[PoC] pg_upgrade: allow to upgrade publisher node

From 3ac0cdd9fef5bbaa3cdd152e3128a8e4747208a1 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Tue, 4 Apr 2023 05:49:34 +0000
Subject: [PATCH v2] pg_upgrade: Add --include-replication-slots option

This commit introduces a new pg_upgrade option called "--include-replication-slots".
This allows nodes with logical replication slots to be upgraded. The commit can
be divided into two parts: one for pg_dump and another for pg_upgrade.

For pg_dump this commit includes a new option called "--replication-slots-only".
This option can be used to dump replication slots. When this option is specified,
the slot_name, plugin, and two_phase parameters are extracted from pg_replication_slots.
An SQL file is then generated which executes pg_create_logical_replication_slot()
with the extracted parameters.

For pg_upgrade, when '--include-replication-slots' is specified, it executes pg_dump
with the new "--replication-slots-only" option and restores from the dump. Apart
from restoring schema, pg_resetwal must not be called after restoring replication
slots. This is because the command discards WAL files and starts from a new segment,
even if they are required by replication slots. This leads to an ERROR: "requested
WAL segment XXX has already been removed". To avoid this, replication slots are
restored at a different time than other objects, after running pg_resetwal.

The significant advantage of this commit is that it makes it easy to continue
logical replication even after upgrading the publisher node. Previously, pg_upgrade
allowed copying publications to a new node. With this new commit, adjusting the
connection string to the new publisher will cause the apply worker on the subscriber
to connect to the new publisher automatically. This enables seamless continuation
of logical replication, even after an upgrade.

Author: Hayato Kuroda
Reviewed-by: Peter Smith, Julien Rouhaud
---
 doc/src/sgml/ref/pg_dump.sgml                 |  10 ++
 doc/src/sgml/ref/pgupgrade.sgml               |  11 ++
 src/bin/pg_dump/pg_backup.h                   |   1 +
 src/bin/pg_dump/pg_dump.c                     | 141 ++++++++++++++++++
 src/bin/pg_dump/pg_dump.h                     |  15 +-
 src/bin/pg_dump/pg_dump_sort.c                |   4 +
 src/bin/pg_upgrade/dump.c                     |  23 +++
 src/bin/pg_upgrade/meson.build                |   1 +
 src/bin/pg_upgrade/option.c                   |   6 +
 src/bin/pg_upgrade/pg_upgrade.c               |  61 ++++++++
 src/bin/pg_upgrade/pg_upgrade.h               |   2 +
 .../pg_upgrade/t/003_logical_replication.pl   |  89 +++++++++++
 12 files changed, 363 insertions(+), 1 deletion(-)
 create mode 100644 src/bin/pg_upgrade/t/003_logical_replication.pl

diff --git a/doc/src/sgml/ref/pg_dump.sgml b/doc/src/sgml/ref/pg_dump.sgml
index e81e35c13b..2cd4fd10b0 100644
--- a/doc/src/sgml/ref/pg_dump.sgml
+++ b/doc/src/sgml/ref/pg_dump.sgml
@@ -1206,6 +1206,16 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--replication-slots-only</option></term>
+      <listitem>
+       <para>
+        Dump only replication slots; not the schema (data definitions), nor
+        data. This is mainly used when upgrading nodes.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
        <term><option>-?</option></term>
        <term><option>--help</option></term>
diff --git a/doc/src/sgml/ref/pgupgrade.sgml b/doc/src/sgml/ref/pgupgrade.sgml
index 7816b4c685..6505b0fd34 100644
--- a/doc/src/sgml/ref/pgupgrade.sgml
+++ b/doc/src/sgml/ref/pgupgrade.sgml
@@ -240,6 +240,17 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--include-replication-slots</option></term>
+      <listitem>
+       <para>
+        Upgrade replication slots. Only logical replication slots are currently
+        supported, and temporary slots are ignored. Note that pg_upgrade does
+        not check the installation of plugins.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
       <term><option>-?</option></term>
       <term><option>--help</option></term>
diff --git a/src/bin/pg_dump/pg_backup.h b/src/bin/pg_dump/pg_backup.h
index aba780ef4b..8a6f25cf2c 100644
--- a/src/bin/pg_dump/pg_backup.h
+++ b/src/bin/pg_dump/pg_backup.h
@@ -187,6 +187,7 @@ typedef struct _dumpOptions
 	int			use_setsessauth;
 	int			enable_row_security;
 	int			load_via_partition_root;
+	int			slot_only;
 
 	/* default, if no "inclusion" switches appear, is to dump everything */
 	bool		include_everything;
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index 7a504dfe25..78c7102d3e 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -328,6 +328,9 @@ static void setupDumpWorker(Archive *AH);
 static TableInfo *getRootTableInfo(const TableInfo *tbinfo);
 static bool forcePartitionRootLoad(const TableInfo *tbinfo);
 
+static void getReplicationSlots(Archive *fout);
+static void dumpReplicationSlot(Archive *fout,
+								const ReplicationSlotInfo * slotinfo);
 
 int
 main(int argc, char **argv)
@@ -431,6 +434,7 @@ main(int argc, char **argv)
 		{"table-and-children", required_argument, NULL, 12},
 		{"exclude-table-and-children", required_argument, NULL, 13},
 		{"exclude-table-data-and-children", required_argument, NULL, 14},
+		{"replication-slots-only", no_argument, NULL, 15},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -657,6 +661,11 @@ main(int argc, char **argv)
 										  optarg);
 				break;
 
+			case 15:			/* dump only replication slot(s) */
+				dopt.slot_only = true;
+				dopt.include_everything = false;
+				break;
+
 			default:
 				/* getopt_long already emitted a complaint */
 				pg_log_error_hint("Try \"%s --help\" for more information.", progname);
@@ -714,6 +723,11 @@ main(int argc, char **argv)
 	if (dopt.do_nothing && dopt.dump_inserts == 0)
 		pg_fatal("option --on-conflict-do-nothing requires option --inserts, --rows-per-insert, or --column-inserts");
 
+	if (dopt.slot_only && dopt.dataOnly)
+		pg_fatal("options --replication-slots-only and -a/--data-only cannot be used together");
+	if (dopt.slot_only && dopt.schemaOnly)
+		pg_fatal("options --replication-slots-only and -s/--schema-only cannot be used together");
+
 	/* Identify archive format to emit */
 	archiveFormat = parseArchiveFormat(format, &archiveMode);
 
@@ -876,6 +890,16 @@ main(int argc, char **argv)
 			pg_fatal("no matching extensions were found");
 	}
 
+	/*
+	 * If dump replication-slots-only was requested, dump only them and skip
+	 * everything else.
+	 */
+	if (dopt.slot_only)
+	{
+		getReplicationSlots(fout);
+		goto dump;
+	}
+
 	/*
 	 * Dumping LOs is the default for dumps where an inclusion switch is not
 	 * used (an "include everything" dump).  -B can be used to exclude LOs
@@ -936,6 +960,8 @@ main(int argc, char **argv)
 	if (!dopt.no_security_labels)
 		collectSecLabels(fout);
 
+dump:
+
 	/* Lastly, create dummy objects to represent the section boundaries */
 	boundaryObjs = createBoundaryObjects();
 
@@ -1119,6 +1145,7 @@ help(const char *progname)
 	printf(_("  --no-unlogged-table-data     do not dump unlogged table data\n"));
 	printf(_("  --on-conflict-do-nothing     add ON CONFLICT DO NOTHING to INSERT commands\n"));
 	printf(_("  --quote-all-identifiers      quote all identifiers, even if not key words\n"));
+	printf(_("  --replication-slots-only     dump only replication slots, no schema or data\n"));
 	printf(_("  --rows-per-insert=NROWS      number of rows per INSERT; implies --inserts\n"));
 	printf(_("  --section=SECTION            dump named section (pre-data, data, or post-data)\n"));
 	printf(_("  --serializable-deferrable    wait until the dump can run without anomalies\n"));
@@ -10252,6 +10279,9 @@ dumpDumpableObject(Archive *fout, DumpableObject *dobj)
 		case DO_SUBSCRIPTION:
 			dumpSubscription(fout, (const SubscriptionInfo *) dobj);
 			break;
+		case DO_REPLICATION_SLOT:
+			dumpReplicationSlot(fout, (const ReplicationSlotInfo *) dobj);
+			break;
 		case DO_PRE_DATA_BOUNDARY:
 		case DO_POST_DATA_BOUNDARY:
 			/* never dumped, nothing to do */
@@ -18227,6 +18257,7 @@ addBoundaryDependencies(DumpableObject **dobjs, int numObjs,
 			case DO_PUBLICATION_REL:
 			case DO_PUBLICATION_TABLE_IN_SCHEMA:
 			case DO_SUBSCRIPTION:
+			case DO_REPLICATION_SLOT:
 				/* Post-data objects: must come after the post-data boundary */
 				addObjectDependency(dobj, postDataBound->dumpId);
 				break;
@@ -18488,3 +18519,113 @@ appendReloptionsArrayAH(PQExpBuffer buffer, const char *reloptions,
 	if (!res)
 		pg_log_warning("could not parse %s array", "reloptions");
 }
+
+/*
+ * getReplicationSlots
+ *	  get information about replication slots
+ */
+static void
+getReplicationSlots(Archive *fout)
+{
+	PGresult   *res;
+	ReplicationSlotInfo *slotinfo;
+	PQExpBuffer query;
+	DumpOptions *dopt = fout->dopt;
+
+	int			i_slotname;
+	int			i_plugin;
+	int			i_twophase;
+	int			i,
+				ntups;
+
+	/* Check whether we should dump or not */
+	if (fout->remoteVersion < 160000 || !dopt->slot_only)
+		return;
+
+	query = createPQExpBuffer();
+
+	resetPQExpBuffer(query);
+
+	/*
+	 * Get replication slots.
+	 *
+	 * XXX: Which information must be extracted from old node? Currently three
+	 * attributes are extracted because they are used by
+	 * pg_create_logical_replication_slot().
+	 * XXX: Do we have to support physical slots?
+	 */
+	appendPQExpBufferStr(query,
+						 "SELECT r.slot_name, r.plugin, r.two_phase "
+						 "FROM pg_replication_slots r "
+						 "WHERE r.database = current_database() AND temporary = false "
+						 "AND wal_status IN ('reserved', 'extended');");
+
+	res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
+
+	ntups = PQntuples(res);
+
+	i_slotname = PQfnumber(res, "slot_name");
+	i_plugin = PQfnumber(res, "plugin");
+	i_twophase = PQfnumber(res, "two_phase");
+
+	slotinfo = pg_malloc(ntups * sizeof(ReplicationSlotInfo));
+
+	for (i = 0; i < ntups; i++)
+	{
+		slotinfo[i].dobj.objType = DO_REPLICATION_SLOT;
+
+		slotinfo[i].dobj.catId.tableoid = InvalidOid;
+		slotinfo[i].dobj.catId.oid = InvalidOid;
+		AssignDumpId(&slotinfo[i].dobj);
+
+		slotinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_slotname));
+
+		slotinfo[i].plugin = pg_strdup(PQgetvalue(res, i, i_plugin));
+		slotinfo[i].twophase = pg_strdup(PQgetvalue(res, i, i_twophase));
+
+		/* FIXME: force dumping */
+		slotinfo[i].dobj.dump = DUMP_COMPONENT_ALL;
+	}
+	PQclear(res);
+
+	destroyPQExpBuffer(query);
+}
+
+/*
+ * dumpReplicationSlot
+ *	  write down a script for pg_restore command
+ */
+static void
+dumpReplicationSlot(Archive *fout, const ReplicationSlotInfo * slotinfo)
+{
+	DumpOptions *dopt = fout->dopt;
+
+	if (!dopt->slot_only)
+		return;
+
+	if (slotinfo->dobj.dump & DUMP_COMPONENT_DEFINITION)
+	{
+		PQExpBuffer query = createPQExpBuffer();
+		char	   *slotname = pg_strdup(slotinfo->dobj.name);
+
+		/*
+		 * XXX: For simplification, pg_create_logical_replication_slot() is
+		 * used. Is it sufficient?
+		 */
+		appendPQExpBuffer(query, "SELECT pg_create_logical_replication_slot('%s', ",
+						  slotname);
+		appendStringLiteralAH(query, slotinfo->plugin, fout);
+		appendPQExpBuffer(query, ", ");
+		appendStringLiteralAH(query, slotinfo->twophase, fout);
+		appendPQExpBuffer(query, ");");
+
+		ArchiveEntry(fout, slotinfo->dobj.catId, slotinfo->dobj.dumpId,
+					 ARCHIVE_OPTS(.tag = slotname,
+								  .description = "REPLICATION SLOT",
+								  .section = SECTION_POST_DATA,
+								  .createStmt = query->data));
+
+		pfree(slotname);
+		destroyPQExpBuffer(query);
+	}
+}
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index ed6ce41ad7..e59cfdd8fa 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -82,7 +82,8 @@ typedef enum
 	DO_PUBLICATION,
 	DO_PUBLICATION_REL,
 	DO_PUBLICATION_TABLE_IN_SCHEMA,
-	DO_SUBSCRIPTION
+	DO_SUBSCRIPTION,
+	DO_REPLICATION_SLOT
 } DumpableObjectType;
 
 /*
@@ -666,6 +667,18 @@ typedef struct _SubscriptionInfo
 	char	   *subpasswordrequired;
 } SubscriptionInfo;
 
+/*
+ * The ReplicationSlotInfo struct is used to represent replication slots.
+ * XXX: add more attrbutes if needed
+ */
+typedef struct _ReplicationSlotInfo
+{
+	DumpableObject dobj;
+	char	   *plugin;
+	char	   *slottype;
+	char	   *twophase;
+}			ReplicationSlotInfo;
+
 /*
  *	common utility functions
  */
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index 8266c117a3..4280283f0d 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -1497,6 +1497,10 @@ describeDumpableObject(DumpableObject *obj, char *buf, int bufsize)
 			snprintf(buf, bufsize,
 					 "SUBSCRIPTION (ID %d OID %u)",
 					 obj->dumpId, obj->catId.oid);
+		case DO_REPLICATION_SLOT:
+			snprintf(buf, bufsize,
+					 "REPLICATION SLOT (ID %d NAME %s)",
+					 obj->dumpId, obj->name);
 			return;
 		case DO_PRE_DATA_BOUNDARY:
 			snprintf(buf, bufsize,
diff --git a/src/bin/pg_upgrade/dump.c b/src/bin/pg_upgrade/dump.c
index 6c8c82dca8..f8d0c6ddde 100644
--- a/src/bin/pg_upgrade/dump.c
+++ b/src/bin/pg_upgrade/dump.c
@@ -59,6 +59,29 @@ generate_old_dump(void)
 						   log_opts.dumpdir,
 						   sql_file_name, escaped_connstr.data);
 
+		/*
+		 * Dump replication slots if needed.
+		 *
+		 * XXX We cannot dump replication slots at the same time as the schema
+		 * dump because we need to separate the timing of restoring
+		 * replication slots and other objects. Replication slots, in
+		 * particular, should not be restored before executing the pg_resetwal
+		 * command because it will remove WALs that are required by the slots.
+		 */
+		if (user_opts.include_slots)
+		{
+			char		slots_file_name[MAXPGPATH];
+
+			snprintf(slots_file_name, sizeof(slots_file_name), DB_DUMP_SLOTS_FILE_MASK, old_db->db_oid);
+			parallel_exec_prog(log_file_name, NULL,
+							   "\"%s/pg_dump\" %s --replication-slots-only "
+							   "--quote-all-identifiers --binary-upgrade %s "
+							   "--file=\"%s/%s\" %s",
+							   new_cluster.bindir, cluster_conn_opts(&old_cluster),
+							   log_opts.verbose ? "--verbose" : "",
+							   log_opts.dumpdir,
+							   slots_file_name, escaped_connstr.data);
+		}
 		termPQExpBuffer(&escaped_connstr);
 	}
 
diff --git a/src/bin/pg_upgrade/meson.build b/src/bin/pg_upgrade/meson.build
index 12a97f84e2..7f5d48b7e1 100644
--- a/src/bin/pg_upgrade/meson.build
+++ b/src/bin/pg_upgrade/meson.build
@@ -42,6 +42,7 @@ tests += {
     'tests': [
       't/001_basic.pl',
       't/002_pg_upgrade.pl',
+      't/003_logical_replication.pl',
     ],
     'test_kwargs': {'priority': 40}, # pg_upgrade tests are slow
   },
diff --git a/src/bin/pg_upgrade/option.c b/src/bin/pg_upgrade/option.c
index 8869b6b60d..d8d9f69b47 100644
--- a/src/bin/pg_upgrade/option.c
+++ b/src/bin/pg_upgrade/option.c
@@ -57,6 +57,7 @@ parseCommandLine(int argc, char *argv[])
 		{"verbose", no_argument, NULL, 'v'},
 		{"clone", no_argument, NULL, 1},
 		{"copy", no_argument, NULL, 2},
+		{"include-replication-slots", no_argument, NULL, 3},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -199,6 +200,10 @@ parseCommandLine(int argc, char *argv[])
 				user_opts.transfer_mode = TRANSFER_MODE_COPY;
 				break;
 
+			case 3:
+				user_opts.include_slots = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						os_info.progname);
@@ -289,6 +294,7 @@ usage(void)
 	printf(_("  -V, --version                 display version information, then exit\n"));
 	printf(_("  --clone                       clone instead of copying files to new cluster\n"));
 	printf(_("  --copy                        copy files to new cluster (default)\n"));
+	printf(_("  --include-replication-slots   upgrade replication slots\n"));
 	printf(_("  -?, --help                    show this help, then exit\n"));
 	printf(_("\n"
 			 "Before running pg_upgrade you must:\n"
diff --git a/src/bin/pg_upgrade/pg_upgrade.c b/src/bin/pg_upgrade/pg_upgrade.c
index 75bab0a04c..04bf1d867a 100644
--- a/src/bin/pg_upgrade/pg_upgrade.c
+++ b/src/bin/pg_upgrade/pg_upgrade.c
@@ -59,6 +59,7 @@ static void copy_xact_xlog_xid(void);
 static void set_frozenxids(bool minmxid_only);
 static void make_outputdirs(char *pgdata);
 static void setup(char *argv0, bool *live_check);
+static void create_replication_slots(void);
 
 ClusterInfo old_cluster,
 			new_cluster;
@@ -188,6 +189,19 @@ main(int argc, char **argv)
 			  new_cluster.pgdata);
 	check_ok();
 
+	/*
+	 * Create replication slots if requested.
+	 *
+	 * Note: This must be done after doing pg_resetwal command because the
+	 * command will remove required WALs.
+	 */
+	if (user_opts.include_slots)
+	{
+		start_postmaster(&new_cluster, true);
+		create_replication_slots();
+		stop_postmaster(false);
+	}
+
 	if (user_opts.do_sync)
 	{
 		prep_status("Sync data directory to disk");
@@ -860,3 +874,50 @@ set_frozenxids(bool minmxid_only)
 
 	check_ok();
 }
+
+/*
+ * create_replication_slots()
+ *
+ * Similar to create_new_objects() but only restores replication slots.
+ */
+static void
+create_replication_slots(void)
+{
+	int			dbnum;
+
+	prep_status_progress("Restoring replication slots in the new cluster");
+
+	for (dbnum = 0; dbnum < old_cluster.dbarr.ndbs; dbnum++)
+	{
+		char		slots_file_name[MAXPGPATH],
+					log_file_name[MAXPGPATH];
+		DbInfo	   *old_db = &old_cluster.dbarr.dbs[dbnum];
+
+		pg_log(PG_STATUS, "%s", old_db->db_name);
+
+		snprintf(slots_file_name, sizeof(slots_file_name),
+				 DB_DUMP_SLOTS_FILE_MASK, old_db->db_oid);
+		snprintf(log_file_name, sizeof(log_file_name),
+				 DB_DUMP_LOG_FILE_MASK, old_db->db_oid);
+
+		parallel_exec_prog(log_file_name,
+						   NULL,
+						   "\"%s/psql\" %s --echo-queries --set ON_ERROR_STOP=on "
+						   "--no-psqlrc --dbname %s -f \"%s/%s\"",
+						   new_cluster.bindir,
+						   cluster_conn_opts(&new_cluster),
+						   old_db->db_name,
+						   log_opts.dumpdir,
+						   slots_file_name);
+	}
+
+	/* reap all children */
+	while (reap_child(true) == true)
+		;
+
+	end_progress_output();
+	check_ok();
+
+	/* update new_cluster info now that we have objects in the databases */
+	get_db_and_rel_infos(&new_cluster);
+}
diff --git a/src/bin/pg_upgrade/pg_upgrade.h b/src/bin/pg_upgrade/pg_upgrade.h
index 3eea0139c7..1aa41f68bc 100644
--- a/src/bin/pg_upgrade/pg_upgrade.h
+++ b/src/bin/pg_upgrade/pg_upgrade.h
@@ -29,6 +29,7 @@
 /* contains both global db information and CREATE DATABASE commands */
 #define GLOBALS_DUMP_FILE	"pg_upgrade_dump_globals.sql"
 #define DB_DUMP_FILE_MASK	"pg_upgrade_dump_%u.custom"
+#define DB_DUMP_SLOTS_FILE_MASK	"pg_upgrade_dump_%u_slots.sql"
 
 /*
  * Base directories that include all the files generated internally, from the
@@ -304,6 +305,7 @@ typedef struct
 	transferMode transfer_mode; /* copy files or link them? */
 	int			jobs;			/* number of processes/threads to use */
 	char	   *socketdir;		/* directory to use for Unix sockets */
+	bool		include_slots;	/* true -> dump and restore replication slots */
 } UserOpts;
 
 typedef struct
diff --git a/src/bin/pg_upgrade/t/003_logical_replication.pl b/src/bin/pg_upgrade/t/003_logical_replication.pl
new file mode 100644
index 0000000000..13ddde3d5f
--- /dev/null
+++ b/src/bin/pg_upgrade/t/003_logical_replication.pl
@@ -0,0 +1,89 @@
+# Copyright (c) 2021-2023, PostgreSQL Global Development Group
+
+# Tests for logical replication, especially for upgrading publisher
+use strict;
+use warnings;
+use PostgreSQL::Test::Cluster;
+use PostgreSQL::Test::Utils;
+use Test::More;
+
+# Can be changed to test the other modes.
+my $mode = $ENV{PG_TEST_PG_UPGRADE_MODE} || '--copy';
+
+# Initialize publisher node
+my $old_publisher = PostgreSQL::Test::Cluster->new('old_publisher');
+$old_publisher->init(allows_streaming => 'logical');
+$old_publisher->start;
+
+# Initialize subscriber node
+my $subscriber = PostgreSQL::Test::Cluster->new('subscriber');
+$subscriber->init(allows_streaming => 'logical');
+$subscriber->start;
+
+$old_publisher->safe_psql('postgres',
+	"CREATE TABLE tbl AS SELECT generate_series(1,10) AS a");
+$subscriber->safe_psql('postgres', "CREATE TABLE tbl (a int)");
+
+# Setup logical replication
+my $old_connstr = $old_publisher->connstr . ' dbname=postgres';
+$old_publisher->safe_psql('postgres',
+	"CREATE PUBLICATION pub FOR ALL TABLES");
+$subscriber->safe_psql('postgres',
+	"CREATE SUBSCRIPTION sub CONNECTION '$old_connstr' PUBLICATION pub");
+
+# Wait for initial table sync to finish
+$subscriber->wait_for_subscription_sync($old_publisher, 'sub');
+
+my $result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(10), 'check initial rows on subscriber');
+
+# Preparations for upgrading publisher
+$old_publisher->stop;
+$subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub DISABLE");
+
+my $new_publisher = PostgreSQL::Test::Cluster->new('new_publisher');
+$new_publisher->init(allows_streaming => 'logical');
+
+my $bindir = $new_publisher->config_data('--bindir');
+
+# Run pg_upgrade. pg_upgrade_output.d is removed at the end
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
+		$mode,        '--include-replication-slot'
+	],
+	'run of pg_upgrade for new publisher');
+ok( !-d $new_publisher->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+# Check whether the replication slot is copied
+$new_publisher->start;
+$result =
+  $new_publisher->safe_psql('postgres',
+	"SELECT count(*) FROM pg_replication_slots");
+is($result, qq(1), 'check the replication slot is copied to new publisher');
+
+# Change connection string and enable logical replication
+my $new_connstr = $new_publisher->connstr . ' dbname=postgres';
+
+$subscriber->safe_psql('postgres',
+	"ALTER SUBSCRIPTION sub CONNECTION '$new_connstr'");
+$subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub ENABLE");
+
+# Check whether changes on the new publisher get replicated to the subscriber
+$new_publisher->safe_psql('postgres',
+	"INSERT INTO tbl VALUES (generate_series(11, 20))");
+
+$new_publisher->wait_for_catchup('sub');
+
+$result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(20), 'check changes are shipped to subscriber');
+
+done_testing();
-- 
2.27.0

From 3f06eb0aa093ac70806d0c3fd6c8713d4e4c0b44 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Tue, 4 Apr 2023 05:49:34 +0000
Subject: [PATCH v3] pg_upgrade: Add --include-logical-replication-slots option

This commit introduces a new pg_upgrade option called "--include-logical-replication-slots".
This allows nodes with logical replication slots to be upgraded. The commit can
be divided into two parts: one for pg_dump and another for pg_upgrade.

For pg_dump this commit includes a new option called "--logical-replication-slots-only".
This option can be used to dump logical replication slots. When this option is
specified, the slot_name, plugin, and two_phase parameters are extracted from
pg_replication_slots. An SQL file is then generated which executes
pg_create_logical_replication_slot() with the extracted parameters.

For pg_upgrade, when '--include-logical-replication-slots' is specified, it executes
pg_dump with the new "--logical-replication-slots-only" option and restores from the
dump. Apart from restoring schema, pg_resetwal must not be called after restoring
replication slots. This is because the command discards WAL files and starts from a
new segment, even if they are required by replication slots. This leads to an ERROR:
"requested WAL segment XXX has already been removed". To avoid this, replication slots
are restored at a different time than other objects, after running pg_resetwal.

The significant advantage of this commit is that it makes it easy to continue
logical replication even after upgrading the publisher node. Previously, pg_upgrade
allowed copying publications to a new node. With this new commit, adjusting the
connection string to the new publisher will cause the apply worker on the subscriber
to connect to the new publisher automatically. This enables seamless continuation
of logical replication, even after an upgrade.

Author: Hayato Kuroda
Reviewed-by: Peter Smith, Julien Rouhaud
---
 doc/src/sgml/ref/pg_dump.sgml                 |  10 ++
 doc/src/sgml/ref/pgupgrade.sgml               |  11 ++
 src/bin/pg_dump/pg_backup.h                   |   1 +
 src/bin/pg_dump/pg_dump.c                     | 144 ++++++++++++++++++
 src/bin/pg_dump/pg_dump.h                     |  16 +-
 src/bin/pg_dump/pg_dump_sort.c                |   4 +
 src/bin/pg_upgrade/dump.c                     |  24 +++
 src/bin/pg_upgrade/meson.build                |   1 +
 src/bin/pg_upgrade/option.c                   |   7 +
 src/bin/pg_upgrade/pg_upgrade.c               |  61 ++++++++
 src/bin/pg_upgrade/pg_upgrade.h               |   3 +
 .../pg_upgrade/t/003_logical_replication.pl   |  89 +++++++++++
 12 files changed, 370 insertions(+), 1 deletion(-)
 create mode 100644 src/bin/pg_upgrade/t/003_logical_replication.pl

diff --git a/doc/src/sgml/ref/pg_dump.sgml b/doc/src/sgml/ref/pg_dump.sgml
index e81e35c13b..6e07f85281 100644
--- a/doc/src/sgml/ref/pg_dump.sgml
+++ b/doc/src/sgml/ref/pg_dump.sgml
@@ -1206,6 +1206,16 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--logical-replication-slots-only</option></term>
+      <listitem>
+       <para>
+        Dump only logical replication slots; not the schema (data definitions),
+        nor data. This is mainly used when upgrading nodes.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
        <term><option>-?</option></term>
        <term><option>--help</option></term>
diff --git a/doc/src/sgml/ref/pgupgrade.sgml b/doc/src/sgml/ref/pgupgrade.sgml
index 7816b4c685..7bcaa388b1 100644
--- a/doc/src/sgml/ref/pgupgrade.sgml
+++ b/doc/src/sgml/ref/pgupgrade.sgml
@@ -240,6 +240,17 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--include-logical-replication-slots</option></term>
+      <listitem>
+       <para>
+        Upgrade logical replication slots. Only permanent replication slots
+        included. Note that pg_upgrade does not check the installation of
+        plugins.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
       <term><option>-?</option></term>
       <term><option>--help</option></term>
diff --git a/src/bin/pg_dump/pg_backup.h b/src/bin/pg_dump/pg_backup.h
index aba780ef4b..5c7832846f 100644
--- a/src/bin/pg_dump/pg_backup.h
+++ b/src/bin/pg_dump/pg_backup.h
@@ -187,6 +187,7 @@ typedef struct _dumpOptions
 	int			use_setsessauth;
 	int			enable_row_security;
 	int			load_via_partition_root;
+	int			logical_slot_only;
 
 	/* default, if no "inclusion" switches appear, is to dump everything */
 	bool		include_everything;
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index 967ced4eed..77d87009e0 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -328,6 +328,9 @@ static void setupDumpWorker(Archive *AH);
 static TableInfo *getRootTableInfo(const TableInfo *tbinfo);
 static bool forcePartitionRootLoad(const TableInfo *tbinfo);
 
+static void getLogicalReplicationSlots(Archive *fout);
+static void dumpLogicalReplicationSlot(Archive *fout,
+									   const LogicalReplicationSlotInfo * slotinfo);
 
 int
 main(int argc, char **argv)
@@ -431,6 +434,7 @@ main(int argc, char **argv)
 		{"table-and-children", required_argument, NULL, 12},
 		{"exclude-table-and-children", required_argument, NULL, 13},
 		{"exclude-table-data-and-children", required_argument, NULL, 14},
+		{"logical-replication-slots-only", no_argument, NULL, 15},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -657,6 +661,11 @@ main(int argc, char **argv)
 										  optarg);
 				break;
 
+			case 15:			/* dump only replication slot(s) */
+				dopt.logical_slot_only = true;
+				dopt.include_everything = false;
+				break;
+
 			default:
 				/* getopt_long already emitted a complaint */
 				pg_log_error_hint("Try \"%s --help\" for more information.", progname);
@@ -714,6 +723,11 @@ main(int argc, char **argv)
 	if (dopt.do_nothing && dopt.dump_inserts == 0)
 		pg_fatal("option --on-conflict-do-nothing requires option --inserts, --rows-per-insert, or --column-inserts");
 
+	if (dopt.logical_slot_only && dopt.dataOnly)
+		pg_fatal("options --logical-replication-slots-only and -a/--data-only cannot be used together");
+	if (dopt.logical_slot_only && dopt.schemaOnly)
+		pg_fatal("options --logical-replication-slots-only and -s/--schema-only cannot be used together");
+
 	/* Identify archive format to emit */
 	archiveFormat = parseArchiveFormat(format, &archiveMode);
 
@@ -876,6 +890,16 @@ main(int argc, char **argv)
 			pg_fatal("no matching extensions were found");
 	}
 
+	/*
+	 * If dump logical-replication-slots-only was requested, dump only them
+	 * and skip everything else.
+	 */
+	if (dopt.logical_slot_only)
+	{
+		getLogicalReplicationSlots(fout);
+		goto dump;
+	}
+
 	/*
 	 * Dumping LOs is the default for dumps where an inclusion switch is not
 	 * used (an "include everything" dump).  -B can be used to exclude LOs
@@ -936,6 +960,8 @@ main(int argc, char **argv)
 	if (!dopt.no_security_labels)
 		collectSecLabels(fout);
 
+dump:
+
 	/* Lastly, create dummy objects to represent the section boundaries */
 	boundaryObjs = createBoundaryObjects();
 
@@ -1119,6 +1145,8 @@ help(const char *progname)
 	printf(_("  --no-unlogged-table-data     do not dump unlogged table data\n"));
 	printf(_("  --on-conflict-do-nothing     add ON CONFLICT DO NOTHING to INSERT commands\n"));
 	printf(_("  --quote-all-identifiers      quote all identifiers, even if not key words\n"));
+	printf(_("  --logical-replication-slots-only\n"
+			 "                               dump only logical replication slots, no schema or data\n"));
 	printf(_("  --rows-per-insert=NROWS      number of rows per INSERT; implies --inserts\n"));
 	printf(_("  --section=SECTION            dump named section (pre-data, data, or post-data)\n"));
 	printf(_("  --serializable-deferrable    wait until the dump can run without anomalies\n"));
@@ -10381,6 +10409,10 @@ dumpDumpableObject(Archive *fout, DumpableObject *dobj)
 		case DO_SUBSCRIPTION:
 			dumpSubscription(fout, (const SubscriptionInfo *) dobj);
 			break;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			dumpLogicalReplicationSlot(fout,
+									   (const LogicalReplicationSlotInfo *) dobj);
+			break;
 		case DO_PRE_DATA_BOUNDARY:
 		case DO_POST_DATA_BOUNDARY:
 			/* never dumped, nothing to do */
@@ -18382,6 +18414,7 @@ addBoundaryDependencies(DumpableObject **dobjs, int numObjs,
 			case DO_PUBLICATION_REL:
 			case DO_PUBLICATION_TABLE_IN_SCHEMA:
 			case DO_SUBSCRIPTION:
+			case DO_LOGICAL_REPLICATION_SLOT:
 				/* Post-data objects: must come after the post-data boundary */
 				addObjectDependency(dobj, postDataBound->dumpId);
 				break;
@@ -18643,3 +18676,114 @@ appendReloptionsArrayAH(PQExpBuffer buffer, const char *reloptions,
 	if (!res)
 		pg_log_warning("could not parse %s array", "reloptions");
 }
+
+/*
+ * getLogicalReplicationSlots
+ *	  get information about replication slots
+ */
+static void
+getLogicalReplicationSlots(Archive *fout)
+{
+	PGresult   *res;
+	LogicalReplicationSlotInfo *slotinfo;
+	PQExpBuffer query;
+	DumpOptions *dopt = fout->dopt;
+
+	int			i_slotname;
+	int			i_plugin;
+	int			i_twophase;
+	int			i,
+				ntups;
+
+	/* Check whether we should dump or not */
+	if (fout->remoteVersion < 160000 || !dopt->logical_slot_only)
+		return;
+
+	query = createPQExpBuffer();
+
+	resetPQExpBuffer(query);
+
+	/*
+	 * Get replication slots.
+	 *
+	 * XXX: Which information must be extracted from old node? Currently three
+	 * attributes are extracted because they are used by
+	 * pg_create_logical_replication_slot().
+	 * XXX: Do we have to support physical slots?
+	 */
+	appendPQExpBufferStr(query,
+						 "SELECT r.slot_name, r.plugin, r.two_phase "
+						 "FROM pg_replication_slots r "
+						 "WHERE r.database = current_database() AND temporary = false "
+						 "AND wal_status IN ('reserved', 'extended');");
+
+	res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
+
+	ntups = PQntuples(res);
+
+	i_slotname = PQfnumber(res, "slot_name");
+	i_plugin = PQfnumber(res, "plugin");
+	i_twophase = PQfnumber(res, "two_phase");
+
+	slotinfo = pg_malloc(ntups * sizeof(LogicalReplicationSlotInfo));
+
+	for (i = 0; i < ntups; i++)
+	{
+		slotinfo[i].dobj.objType = DO_LOGICAL_REPLICATION_SLOT;
+
+		slotinfo[i].dobj.catId.tableoid = InvalidOid;
+		slotinfo[i].dobj.catId.oid = InvalidOid;
+		AssignDumpId(&slotinfo[i].dobj);
+
+		slotinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_slotname));
+
+		slotinfo[i].plugin = pg_strdup(PQgetvalue(res, i, i_plugin));
+		slotinfo[i].twophase = pg_strdup(PQgetvalue(res, i, i_twophase));
+
+		/* FIXME: force dumping */
+		slotinfo[i].dobj.dump = DUMP_COMPONENT_ALL;
+	}
+	PQclear(res);
+
+	destroyPQExpBuffer(query);
+}
+
+/*
+ * dumpLogicalReplicationSlot
+ *	  dump creation functions for the given logical replication slots
+ */
+static void
+dumpLogicalReplicationSlot(Archive *fout,
+						   const LogicalReplicationSlotInfo * slotinfo)
+{
+	DumpOptions *dopt = fout->dopt;
+
+	if (!dopt->logical_slot_only)
+		return;
+
+	if (slotinfo->dobj.dump & DUMP_COMPONENT_DEFINITION)
+	{
+		PQExpBuffer query = createPQExpBuffer();
+		char	   *slotname = pg_strdup(slotinfo->dobj.name);
+
+		/*
+		 * XXX: For simplification, pg_create_logical_replication_slot() is
+		 * used. Is it sufficient?
+		 */
+		appendPQExpBuffer(query, "SELECT pg_create_logical_replication_slot('%s', ",
+						  slotname);
+		appendStringLiteralAH(query, slotinfo->plugin, fout);
+		appendPQExpBuffer(query, ", ");
+		appendStringLiteralAH(query, slotinfo->twophase, fout);
+		appendPQExpBuffer(query, ");");
+
+		ArchiveEntry(fout, slotinfo->dobj.catId, slotinfo->dobj.dumpId,
+					 ARCHIVE_OPTS(.tag = slotname,
+								  .description = "REPLICATION SLOT",
+								  .section = SECTION_POST_DATA,
+								  .createStmt = query->data));
+
+		pfree(slotname);
+		destroyPQExpBuffer(query);
+	}
+}
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index 765fe6399a..e587d4cca4 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -82,7 +82,8 @@ typedef enum
 	DO_PUBLICATION,
 	DO_PUBLICATION_REL,
 	DO_PUBLICATION_TABLE_IN_SCHEMA,
-	DO_SUBSCRIPTION
+	DO_SUBSCRIPTION,
+	DO_LOGICAL_REPLICATION_SLOT
 } DumpableObjectType;
 
 /*
@@ -666,6 +667,19 @@ typedef struct _SubscriptionInfo
 	char	   *subpasswordrequired;
 } SubscriptionInfo;
 
+/*
+ * The LogicalReplicationSlotInfo struct is used to represent replication
+ * slots.
+ * XXX: add more attrbutes if needed
+ */
+typedef struct _LogicalReplicationSlotInfo
+{
+	DumpableObject dobj;
+	char	   *plugin;
+	char	   *slottype;
+	char	   *twophase;
+}			LogicalReplicationSlotInfo;
+
 /*
  *	common utility functions
  */
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index 8266c117a3..0f3c89b9fb 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -1497,6 +1497,10 @@ describeDumpableObject(DumpableObject *obj, char *buf, int bufsize)
 			snprintf(buf, bufsize,
 					 "SUBSCRIPTION (ID %d OID %u)",
 					 obj->dumpId, obj->catId.oid);
+		case DO_LOGICAL_REPLICATION_SLOT:
+			snprintf(buf, bufsize,
+					 "REPLICATION SLOT (ID %d NAME %s)",
+					 obj->dumpId, obj->name);
 			return;
 		case DO_PRE_DATA_BOUNDARY:
 			snprintf(buf, bufsize,
diff --git a/src/bin/pg_upgrade/dump.c b/src/bin/pg_upgrade/dump.c
index 6c8c82dca8..e6b90864f5 100644
--- a/src/bin/pg_upgrade/dump.c
+++ b/src/bin/pg_upgrade/dump.c
@@ -59,6 +59,30 @@ generate_old_dump(void)
 						   log_opts.dumpdir,
 						   sql_file_name, escaped_connstr.data);
 
+		/*
+		 * Dump logical replication slots if needed.
+		 *
+		 * XXX We cannot dump replication slots at the same time as the schema
+		 * dump because we need to separate the timing of restoring
+		 * replication slots and other objects. Replication slots, in
+		 * particular, should not be restored before executing the pg_resetwal
+		 * command because it will remove WALs that are required by the slots.
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			char		slots_file_name[MAXPGPATH];
+
+			snprintf(slots_file_name, sizeof(slots_file_name),
+					 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+			parallel_exec_prog(log_file_name, NULL,
+							   "\"%s/pg_dump\" %s --logical-replication-slots-only "
+							   "--quote-all-identifiers --binary-upgrade %s "
+							   "--file=\"%s/%s\" %s",
+							   new_cluster.bindir, cluster_conn_opts(&old_cluster),
+							   log_opts.verbose ? "--verbose" : "",
+							   log_opts.dumpdir,
+							   slots_file_name, escaped_connstr.data);
+		}
 		termPQExpBuffer(&escaped_connstr);
 	}
 
diff --git a/src/bin/pg_upgrade/meson.build b/src/bin/pg_upgrade/meson.build
index 12a97f84e2..7f5d48b7e1 100644
--- a/src/bin/pg_upgrade/meson.build
+++ b/src/bin/pg_upgrade/meson.build
@@ -42,6 +42,7 @@ tests += {
     'tests': [
       't/001_basic.pl',
       't/002_pg_upgrade.pl',
+      't/003_logical_replication.pl',
     ],
     'test_kwargs': {'priority': 40}, # pg_upgrade tests are slow
   },
diff --git a/src/bin/pg_upgrade/option.c b/src/bin/pg_upgrade/option.c
index 8869b6b60d..f7b5ec9879 100644
--- a/src/bin/pg_upgrade/option.c
+++ b/src/bin/pg_upgrade/option.c
@@ -57,6 +57,7 @@ parseCommandLine(int argc, char *argv[])
 		{"verbose", no_argument, NULL, 'v'},
 		{"clone", no_argument, NULL, 1},
 		{"copy", no_argument, NULL, 2},
+		{"include-logical-replication-slots", no_argument, NULL, 3},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -199,6 +200,10 @@ parseCommandLine(int argc, char *argv[])
 				user_opts.transfer_mode = TRANSFER_MODE_COPY;
 				break;
 
+			case 3:
+				user_opts.include_logical_slots = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						os_info.progname);
@@ -289,6 +294,8 @@ usage(void)
 	printf(_("  -V, --version                 display version information, then exit\n"));
 	printf(_("  --clone                       clone instead of copying files to new cluster\n"));
 	printf(_("  --copy                        copy files to new cluster (default)\n"));
+	printf(_("  --include-logical-replication-slots\n"
+			 "                                upgrade logical replication slots\n"));
 	printf(_("  -?, --help                    show this help, then exit\n"));
 	printf(_("\n"
 			 "Before running pg_upgrade you must:\n"
diff --git a/src/bin/pg_upgrade/pg_upgrade.c b/src/bin/pg_upgrade/pg_upgrade.c
index 75bab0a04c..1241060f4e 100644
--- a/src/bin/pg_upgrade/pg_upgrade.c
+++ b/src/bin/pg_upgrade/pg_upgrade.c
@@ -59,6 +59,7 @@ static void copy_xact_xlog_xid(void);
 static void set_frozenxids(bool minmxid_only);
 static void make_outputdirs(char *pgdata);
 static void setup(char *argv0, bool *live_check);
+static void create_logical_replication_slots(void);
 
 ClusterInfo old_cluster,
 			new_cluster;
@@ -188,6 +189,19 @@ main(int argc, char **argv)
 			  new_cluster.pgdata);
 	check_ok();
 
+	/*
+	 * Create logical replication slots if requested.
+	 *
+	 * Note: This must be done after doing pg_resetwal command because the
+	 * command will remove required WALs.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		start_postmaster(&new_cluster, true);
+		create_logical_replication_slots();
+		stop_postmaster(false);
+	}
+
 	if (user_opts.do_sync)
 	{
 		prep_status("Sync data directory to disk");
@@ -860,3 +874,50 @@ set_frozenxids(bool minmxid_only)
 
 	check_ok();
 }
+
+/*
+ * create_logical_replication_slots()
+ *
+ * Similar to create_new_objects() but only restores logical replication slots.
+ */
+static void
+create_logical_replication_slots(void)
+{
+	int			dbnum;
+
+	prep_status_progress("Restoring logical replication slots in the new cluster");
+
+	for (dbnum = 0; dbnum < old_cluster.dbarr.ndbs; dbnum++)
+	{
+		char		slots_file_name[MAXPGPATH],
+					log_file_name[MAXPGPATH];
+		DbInfo	   *old_db = &old_cluster.dbarr.dbs[dbnum];
+
+		pg_log(PG_STATUS, "%s", old_db->db_name);
+
+		snprintf(slots_file_name, sizeof(slots_file_name),
+				 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+		snprintf(log_file_name, sizeof(log_file_name),
+				 DB_DUMP_LOG_FILE_MASK, old_db->db_oid);
+
+		parallel_exec_prog(log_file_name,
+						   NULL,
+						   "\"%s/psql\" %s --echo-queries --set ON_ERROR_STOP=on "
+						   "--no-psqlrc --dbname %s -f \"%s/%s\"",
+						   new_cluster.bindir,
+						   cluster_conn_opts(&new_cluster),
+						   old_db->db_name,
+						   log_opts.dumpdir,
+						   slots_file_name);
+	}
+
+	/* reap all children */
+	while (reap_child(true) == true)
+		;
+
+	end_progress_output();
+	check_ok();
+
+	/* update new_cluster info now that we have objects in the databases */
+	get_db_and_rel_infos(&new_cluster);
+}
diff --git a/src/bin/pg_upgrade/pg_upgrade.h b/src/bin/pg_upgrade/pg_upgrade.h
index 3eea0139c7..5f3d7a407e 100644
--- a/src/bin/pg_upgrade/pg_upgrade.h
+++ b/src/bin/pg_upgrade/pg_upgrade.h
@@ -29,6 +29,7 @@
 /* contains both global db information and CREATE DATABASE commands */
 #define GLOBALS_DUMP_FILE	"pg_upgrade_dump_globals.sql"
 #define DB_DUMP_FILE_MASK	"pg_upgrade_dump_%u.custom"
+#define DB_DUMP_LOGICAL_SLOTS_FILE_MASK	"pg_upgrade_dump_%u_logical_slots.sql"
 
 /*
  * Base directories that include all the files generated internally, from the
@@ -304,6 +305,8 @@ typedef struct
 	transferMode transfer_mode; /* copy files or link them? */
 	int			jobs;			/* number of processes/threads to use */
 	char	   *socketdir;		/* directory to use for Unix sockets */
+	bool		include_logical_slots;	/* true -> dump and restore logical
+										 * replication slots */
 } UserOpts;
 
 typedef struct
diff --git a/src/bin/pg_upgrade/t/003_logical_replication.pl b/src/bin/pg_upgrade/t/003_logical_replication.pl
new file mode 100644
index 0000000000..8b3be8b0d4
--- /dev/null
+++ b/src/bin/pg_upgrade/t/003_logical_replication.pl
@@ -0,0 +1,89 @@
+# Copyright (c) 2021-2023, PostgreSQL Global Development Group
+
+# Tests for logical replication, especially for upgrading publisher
+use strict;
+use warnings;
+use PostgreSQL::Test::Cluster;
+use PostgreSQL::Test::Utils;
+use Test::More;
+
+# Can be changed to test the other modes.
+my $mode = $ENV{PG_TEST_PG_UPGRADE_MODE} || '--copy';
+
+# Initialize publisher node
+my $old_publisher = PostgreSQL::Test::Cluster->new('old_publisher');
+$old_publisher->init(allows_streaming => 'logical');
+$old_publisher->start;
+
+# Initialize subscriber node
+my $subscriber = PostgreSQL::Test::Cluster->new('subscriber');
+$subscriber->init(allows_streaming => 'logical');
+$subscriber->start;
+
+$old_publisher->safe_psql('postgres',
+	"CREATE TABLE tbl AS SELECT generate_series(1,10) AS a");
+$subscriber->safe_psql('postgres', "CREATE TABLE tbl (a int)");
+
+# Setup logical replication
+my $old_connstr = $old_publisher->connstr . ' dbname=postgres';
+$old_publisher->safe_psql('postgres',
+	"CREATE PUBLICATION pub FOR ALL TABLES");
+$subscriber->safe_psql('postgres',
+	"CREATE SUBSCRIPTION sub CONNECTION '$old_connstr' PUBLICATION pub");
+
+# Wait for initial table sync to finish
+$subscriber->wait_for_subscription_sync($old_publisher, 'sub');
+
+my $result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(10), 'check initial rows on subscriber');
+
+# Preparations for upgrading publisher
+$old_publisher->stop;
+$subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub DISABLE");
+
+my $new_publisher = PostgreSQL::Test::Cluster->new('new_publisher');
+$new_publisher->init(allows_streaming => 'logical');
+
+my $bindir = $new_publisher->config_data('--bindir');
+
+# Run pg_upgrade. pg_upgrade_output.d is removed at the end
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
+		$mode,        '--include-logical-replication-slot'
+	],
+	'run of pg_upgrade for new publisher');
+ok( !-d $new_publisher->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+# Check whether the replication slot is copied
+$new_publisher->start;
+$result =
+  $new_publisher->safe_psql('postgres',
+	"SELECT count(*) FROM pg_replication_slots");
+is($result, qq(1), 'check the replication slot is copied to new publisher');
+
+# Change connection string and enable logical replication
+my $new_connstr = $new_publisher->connstr . ' dbname=postgres';
+
+$subscriber->safe_psql('postgres',
+	"ALTER SUBSCRIPTION sub CONNECTION '$new_connstr'");
+$subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub ENABLE");
+
+# Check whether changes on the new publisher get replicated to the subscriber
+$new_publisher->safe_psql('postgres',
+	"INSERT INTO tbl VALUES (generate_series(11, 20))");
+
+$new_publisher->wait_for_catchup('sub');
+
+$result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(20), 'check changes are shipped to subscriber');
+
+done_testing();
-- 
2.27.0

From 2cc29a84cb4dae1a7db8742f8b165dca42e39ac7 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Tue, 4 Apr 2023 05:49:34 +0000
Subject: [PATCH v4] pg_upgrade: Add --include-logical-replication-slots option

This commit introduces a new pg_upgrade option called "--include-logical-replication-slots".
This allows nodes with logical replication slots to be upgraded. The commit can
be divided into two parts: one for pg_dump and another for pg_upgrade.

For pg_dump this commit includes a new option called "--logical-replication-slots-only".
This option can be used to dump logical replication slots. When this option is
specified, the slot_name, plugin, and two_phase parameters are extracted from
pg_replication_slots. An SQL file is then generated which executes
pg_create_logical_replication_slot() with the extracted parameters.

For pg_upgrade, when '--include-logical-replication-slots' is specified, it executes
pg_dump with the new "--logical-replication-slots-only" option and restores from the
dump. Apart from restoring schema, pg_resetwal must not be called after restoring
replication slots. This is because the command discards WAL files and starts from a
new segment, even if they are required by replication slots. This leads to an ERROR:
"requested WAL segment XXX has already been removed". To avoid this, replication slots
are restored at a different time than other objects, after running pg_resetwal.

The significant advantage of this commit is that it makes it easy to continue
logical replication even after upgrading the publisher node. Previously, pg_upgrade
allowed copying publications to a new node. With this new commit, adjusting the
connection string to the new publisher will cause the apply worker on the subscriber
to connect to the new publisher automatically. This enables seamless continuation
of logical replication, even after an upgrade.

Author: Hayato Kuroda
Reviewed-by: Peter Smith, Julien Rouhaud
---
 doc/src/sgml/ref/pg_dump.sgml                 |  10 ++
 doc/src/sgml/ref/pgupgrade.sgml               |  11 ++
 src/bin/pg_dump/pg_backup.h                   |   1 +
 src/bin/pg_dump/pg_dump.c                     | 144 ++++++++++++++++++
 src/bin/pg_dump/pg_dump.h                     |  16 +-
 src/bin/pg_dump/pg_dump_sort.c                |  11 +-
 src/bin/pg_upgrade/dump.c                     |  24 +++
 src/bin/pg_upgrade/meson.build                |   1 +
 src/bin/pg_upgrade/option.c                   |   7 +
 src/bin/pg_upgrade/pg_upgrade.c               |  61 ++++++++
 src/bin/pg_upgrade/pg_upgrade.h               |   3 +
 .../pg_upgrade/t/003_logical_replication.pl   |  89 +++++++++++
 src/tools/pgindent/typedefs.list              |   1 +
 13 files changed, 376 insertions(+), 3 deletions(-)
 create mode 100644 src/bin/pg_upgrade/t/003_logical_replication.pl

diff --git a/doc/src/sgml/ref/pg_dump.sgml b/doc/src/sgml/ref/pg_dump.sgml
index e81e35c13b..6e07f85281 100644
--- a/doc/src/sgml/ref/pg_dump.sgml
+++ b/doc/src/sgml/ref/pg_dump.sgml
@@ -1206,6 +1206,16 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--logical-replication-slots-only</option></term>
+      <listitem>
+       <para>
+        Dump only logical replication slots; not the schema (data definitions),
+        nor data. This is mainly used when upgrading nodes.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
        <term><option>-?</option></term>
        <term><option>--help</option></term>
diff --git a/doc/src/sgml/ref/pgupgrade.sgml b/doc/src/sgml/ref/pgupgrade.sgml
index 7816b4c685..94e90ff506 100644
--- a/doc/src/sgml/ref/pgupgrade.sgml
+++ b/doc/src/sgml/ref/pgupgrade.sgml
@@ -240,6 +240,17 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--include-logical-replication-slots</option></term>
+      <listitem>
+       <para>
+        Upgrade logical replication slots. Only permanent replication slots
+        are included. Note that pg_upgrade does not check the installation of
+        plugins.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
       <term><option>-?</option></term>
       <term><option>--help</option></term>
diff --git a/src/bin/pg_dump/pg_backup.h b/src/bin/pg_dump/pg_backup.h
index aba780ef4b..5c7832846f 100644
--- a/src/bin/pg_dump/pg_backup.h
+++ b/src/bin/pg_dump/pg_backup.h
@@ -187,6 +187,7 @@ typedef struct _dumpOptions
 	int			use_setsessauth;
 	int			enable_row_security;
 	int			load_via_partition_root;
+	int			logical_slot_only;
 
 	/* default, if no "inclusion" switches appear, is to dump everything */
 	bool		include_everything;
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index 967ced4eed..05ca3f8677 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -328,6 +328,9 @@ static void setupDumpWorker(Archive *AH);
 static TableInfo *getRootTableInfo(const TableInfo *tbinfo);
 static bool forcePartitionRootLoad(const TableInfo *tbinfo);
 
+static void getLogicalReplicationSlots(Archive *fout);
+static void dumpLogicalReplicationSlot(Archive *fout,
+									   const LogicalReplicationSlotInfo *slotinfo);
 
 int
 main(int argc, char **argv)
@@ -431,6 +434,7 @@ main(int argc, char **argv)
 		{"table-and-children", required_argument, NULL, 12},
 		{"exclude-table-and-children", required_argument, NULL, 13},
 		{"exclude-table-data-and-children", required_argument, NULL, 14},
+		{"logical-replication-slots-only", no_argument, NULL, 15},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -657,6 +661,11 @@ main(int argc, char **argv)
 										  optarg);
 				break;
 
+			case 15:			/* dump only replication slot(s) */
+				dopt.logical_slot_only = true;
+				dopt.include_everything = false;
+				break;
+
 			default:
 				/* getopt_long already emitted a complaint */
 				pg_log_error_hint("Try \"%s --help\" for more information.", progname);
@@ -714,6 +723,11 @@ main(int argc, char **argv)
 	if (dopt.do_nothing && dopt.dump_inserts == 0)
 		pg_fatal("option --on-conflict-do-nothing requires option --inserts, --rows-per-insert, or --column-inserts");
 
+	if (dopt.logical_slot_only && dopt.dataOnly)
+		pg_fatal("options --logical-replication-slots-only and -a/--data-only cannot be used together");
+	if (dopt.logical_slot_only && dopt.schemaOnly)
+		pg_fatal("options --logical-replication-slots-only and -s/--schema-only cannot be used together");
+
 	/* Identify archive format to emit */
 	archiveFormat = parseArchiveFormat(format, &archiveMode);
 
@@ -876,6 +890,16 @@ main(int argc, char **argv)
 			pg_fatal("no matching extensions were found");
 	}
 
+	/*
+	 * If dump logical-replication-slots-only was requested, dump only them
+	 * and skip everything else.
+	 */
+	if (dopt.logical_slot_only)
+	{
+		getLogicalReplicationSlots(fout);
+		goto dump;
+	}
+
 	/*
 	 * Dumping LOs is the default for dumps where an inclusion switch is not
 	 * used (an "include everything" dump).  -B can be used to exclude LOs
@@ -936,6 +960,8 @@ main(int argc, char **argv)
 	if (!dopt.no_security_labels)
 		collectSecLabels(fout);
 
+dump:
+
 	/* Lastly, create dummy objects to represent the section boundaries */
 	boundaryObjs = createBoundaryObjects();
 
@@ -1109,6 +1135,8 @@ help(const char *progname)
 			 "                               servers matching PATTERN\n"));
 	printf(_("  --inserts                    dump data as INSERT commands, rather than COPY\n"));
 	printf(_("  --load-via-partition-root    load partitions via the root table\n"));
+	printf(_("  --logical-replication-slots-only\n"
+			 "                               dump only logical replication slots, no schema or data\n"));
 	printf(_("  --no-comments                do not dump comments\n"));
 	printf(_("  --no-publications            do not dump publications\n"));
 	printf(_("  --no-security-labels         do not dump security label assignments\n"));
@@ -10381,6 +10409,10 @@ dumpDumpableObject(Archive *fout, DumpableObject *dobj)
 		case DO_SUBSCRIPTION:
 			dumpSubscription(fout, (const SubscriptionInfo *) dobj);
 			break;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			dumpLogicalReplicationSlot(fout,
+									   (const LogicalReplicationSlotInfo *) dobj);
+			break;
 		case DO_PRE_DATA_BOUNDARY:
 		case DO_POST_DATA_BOUNDARY:
 			/* never dumped, nothing to do */
@@ -18382,6 +18414,7 @@ addBoundaryDependencies(DumpableObject **dobjs, int numObjs,
 			case DO_PUBLICATION_REL:
 			case DO_PUBLICATION_TABLE_IN_SCHEMA:
 			case DO_SUBSCRIPTION:
+			case DO_LOGICAL_REPLICATION_SLOT:
 				/* Post-data objects: must come after the post-data boundary */
 				addObjectDependency(dobj, postDataBound->dumpId);
 				break;
@@ -18643,3 +18676,114 @@ appendReloptionsArrayAH(PQExpBuffer buffer, const char *reloptions,
 	if (!res)
 		pg_log_warning("could not parse %s array", "reloptions");
 }
+
+/*
+ * getLogicalReplicationSlots
+ *	  get information about replication slots
+ */
+static void
+getLogicalReplicationSlots(Archive *fout)
+{
+	PGresult   *res;
+	LogicalReplicationSlotInfo *slotinfo;
+	PQExpBuffer query;
+	DumpOptions *dopt = fout->dopt;
+
+	int			i_slotname;
+	int			i_plugin;
+	int			i_twophase;
+	int			i,
+				ntups;
+
+	/* Check whether we should dump or not */
+	if (fout->remoteVersion < 160000 || !dopt->logical_slot_only)
+		return;
+
+	query = createPQExpBuffer();
+
+	resetPQExpBuffer(query);
+
+	/*
+	 * Get replication slots.
+	 *
+	 * XXX: Which information must be extracted from old node? Currently three
+	 * attributes are extracted because they are used by
+	 * pg_create_logical_replication_slot().
+	 * XXX: Do we have to support physical slots?
+	 */
+	appendPQExpBufferStr(query,
+						 "SELECT r.slot_name, r.plugin, r.two_phase "
+						 "FROM pg_replication_slots r "
+						 "WHERE r.database = current_database() AND temporary = false "
+						 "AND wal_status IN ('reserved', 'extended');");
+
+	res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
+
+	ntups = PQntuples(res);
+
+	i_slotname = PQfnumber(res, "slot_name");
+	i_plugin = PQfnumber(res, "plugin");
+	i_twophase = PQfnumber(res, "two_phase");
+
+	slotinfo = pg_malloc(ntups * sizeof(LogicalReplicationSlotInfo));
+
+	for (i = 0; i < ntups; i++)
+	{
+		slotinfo[i].dobj.objType = DO_LOGICAL_REPLICATION_SLOT;
+
+		slotinfo[i].dobj.catId.tableoid = InvalidOid;
+		slotinfo[i].dobj.catId.oid = InvalidOid;
+		AssignDumpId(&slotinfo[i].dobj);
+
+		slotinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_slotname));
+
+		slotinfo[i].plugin = pg_strdup(PQgetvalue(res, i, i_plugin));
+		slotinfo[i].twophase = pg_strdup(PQgetvalue(res, i, i_twophase));
+
+		/* FIXME: force dumping */
+		slotinfo[i].dobj.dump = DUMP_COMPONENT_ALL;
+	}
+	PQclear(res);
+
+	destroyPQExpBuffer(query);
+}
+
+/*
+ * dumpLogicalReplicationSlot
+ *	  dump creation functions for the given logical replication slots
+ */
+static void
+dumpLogicalReplicationSlot(Archive *fout,
+						   const LogicalReplicationSlotInfo *slotinfo)
+{
+	DumpOptions *dopt = fout->dopt;
+
+	if (!dopt->logical_slot_only)
+		return;
+
+	if (slotinfo->dobj.dump & DUMP_COMPONENT_DEFINITION)
+	{
+		PQExpBuffer query = createPQExpBuffer();
+		char	   *slotname = pg_strdup(slotinfo->dobj.name);
+
+		/*
+		 * XXX: For simplification, pg_create_logical_replication_slot() is
+		 * used. Is it sufficient?
+		 */
+		appendPQExpBuffer(query, "SELECT pg_create_logical_replication_slot('%s', ",
+						  slotname);
+		appendStringLiteralAH(query, slotinfo->plugin, fout);
+		appendPQExpBuffer(query, ", ");
+		appendStringLiteralAH(query, slotinfo->twophase, fout);
+		appendPQExpBuffer(query, ");");
+
+		ArchiveEntry(fout, slotinfo->dobj.catId, slotinfo->dobj.dumpId,
+					 ARCHIVE_OPTS(.tag = slotname,
+								  .description = "REPLICATION SLOT",
+								  .section = SECTION_POST_DATA,
+								  .createStmt = query->data));
+
+		pfree(slotname);
+		destroyPQExpBuffer(query);
+	}
+}
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index 765fe6399a..f5535816d2 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -82,7 +82,8 @@ typedef enum
 	DO_PUBLICATION,
 	DO_PUBLICATION_REL,
 	DO_PUBLICATION_TABLE_IN_SCHEMA,
-	DO_SUBSCRIPTION
+	DO_SUBSCRIPTION,
+	DO_LOGICAL_REPLICATION_SLOT
 } DumpableObjectType;
 
 /*
@@ -666,6 +667,19 @@ typedef struct _SubscriptionInfo
 	char	   *subpasswordrequired;
 } SubscriptionInfo;
 
+/*
+ * The LogicalReplicationSlotInfo struct is used to represent replication
+ * slots.
+ * XXX: add more attrbutes if needed
+ */
+typedef struct _LogicalReplicationSlotInfo
+{
+	DumpableObject dobj;
+	char	   *plugin;
+	char	   *slottype;
+	char	   *twophase;
+} LogicalReplicationSlotInfo;
+
 /*
  *	common utility functions
  */
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index 8266c117a3..b36ced8c8e 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -93,6 +93,7 @@ enum dbObjectTypePriorities
 	PRIO_PUBLICATION_REL,
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,
 	PRIO_SUBSCRIPTION,
+	PRIO_LOGICAL_REPLICATION_SLOT,
 	PRIO_DEFAULT_ACL,			/* done in ACL pass */
 	PRIO_EVENT_TRIGGER,			/* must be next to last! */
 	PRIO_REFRESH_MATVIEW		/* must be last! */
@@ -146,10 +147,11 @@ static const int dbObjectTypePriority[] =
 	PRIO_PUBLICATION,			/* DO_PUBLICATION */
 	PRIO_PUBLICATION_REL,		/* DO_PUBLICATION_REL */
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,	/* DO_PUBLICATION_TABLE_IN_SCHEMA */
-	PRIO_SUBSCRIPTION			/* DO_SUBSCRIPTION */
+	PRIO_SUBSCRIPTION,			/* DO_SUBSCRIPTION */
+	PRIO_LOGICAL_REPLICATION_SLOT	/* DO_LOGICAL_REPLICATION_SLOT */
 };
 
-StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_SUBSCRIPTION + 1),
+StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_LOGICAL_REPLICATION_SLOT + 1),
 				 "array length mismatch");
 
 static DumpId preDataBoundId;
@@ -1498,6 +1500,11 @@ describeDumpableObject(DumpableObject *obj, char *buf, int bufsize)
 					 "SUBSCRIPTION (ID %d OID %u)",
 					 obj->dumpId, obj->catId.oid);
 			return;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			snprintf(buf, bufsize,
+					 "REPLICATION SLOT (ID %d NAME %s)",
+					 obj->dumpId, obj->name);
+			return;
 		case DO_PRE_DATA_BOUNDARY:
 			snprintf(buf, bufsize,
 					 "PRE-DATA BOUNDARY  (ID %d)",
diff --git a/src/bin/pg_upgrade/dump.c b/src/bin/pg_upgrade/dump.c
index 6c8c82dca8..e6b90864f5 100644
--- a/src/bin/pg_upgrade/dump.c
+++ b/src/bin/pg_upgrade/dump.c
@@ -59,6 +59,30 @@ generate_old_dump(void)
 						   log_opts.dumpdir,
 						   sql_file_name, escaped_connstr.data);
 
+		/*
+		 * Dump logical replication slots if needed.
+		 *
+		 * XXX We cannot dump replication slots at the same time as the schema
+		 * dump because we need to separate the timing of restoring
+		 * replication slots and other objects. Replication slots, in
+		 * particular, should not be restored before executing the pg_resetwal
+		 * command because it will remove WALs that are required by the slots.
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			char		slots_file_name[MAXPGPATH];
+
+			snprintf(slots_file_name, sizeof(slots_file_name),
+					 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+			parallel_exec_prog(log_file_name, NULL,
+							   "\"%s/pg_dump\" %s --logical-replication-slots-only "
+							   "--quote-all-identifiers --binary-upgrade %s "
+							   "--file=\"%s/%s\" %s",
+							   new_cluster.bindir, cluster_conn_opts(&old_cluster),
+							   log_opts.verbose ? "--verbose" : "",
+							   log_opts.dumpdir,
+							   slots_file_name, escaped_connstr.data);
+		}
 		termPQExpBuffer(&escaped_connstr);
 	}
 
diff --git a/src/bin/pg_upgrade/meson.build b/src/bin/pg_upgrade/meson.build
index 12a97f84e2..7f5d48b7e1 100644
--- a/src/bin/pg_upgrade/meson.build
+++ b/src/bin/pg_upgrade/meson.build
@@ -42,6 +42,7 @@ tests += {
     'tests': [
       't/001_basic.pl',
       't/002_pg_upgrade.pl',
+      't/003_logical_replication.pl',
     ],
     'test_kwargs': {'priority': 40}, # pg_upgrade tests are slow
   },
diff --git a/src/bin/pg_upgrade/option.c b/src/bin/pg_upgrade/option.c
index 8869b6b60d..f7b5ec9879 100644
--- a/src/bin/pg_upgrade/option.c
+++ b/src/bin/pg_upgrade/option.c
@@ -57,6 +57,7 @@ parseCommandLine(int argc, char *argv[])
 		{"verbose", no_argument, NULL, 'v'},
 		{"clone", no_argument, NULL, 1},
 		{"copy", no_argument, NULL, 2},
+		{"include-logical-replication-slots", no_argument, NULL, 3},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -199,6 +200,10 @@ parseCommandLine(int argc, char *argv[])
 				user_opts.transfer_mode = TRANSFER_MODE_COPY;
 				break;
 
+			case 3:
+				user_opts.include_logical_slots = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						os_info.progname);
@@ -289,6 +294,8 @@ usage(void)
 	printf(_("  -V, --version                 display version information, then exit\n"));
 	printf(_("  --clone                       clone instead of copying files to new cluster\n"));
 	printf(_("  --copy                        copy files to new cluster (default)\n"));
+	printf(_("  --include-logical-replication-slots\n"
+			 "                                upgrade logical replication slots\n"));
 	printf(_("  -?, --help                    show this help, then exit\n"));
 	printf(_("\n"
 			 "Before running pg_upgrade you must:\n"
diff --git a/src/bin/pg_upgrade/pg_upgrade.c b/src/bin/pg_upgrade/pg_upgrade.c
index 75bab0a04c..1241060f4e 100644
--- a/src/bin/pg_upgrade/pg_upgrade.c
+++ b/src/bin/pg_upgrade/pg_upgrade.c
@@ -59,6 +59,7 @@ static void copy_xact_xlog_xid(void);
 static void set_frozenxids(bool minmxid_only);
 static void make_outputdirs(char *pgdata);
 static void setup(char *argv0, bool *live_check);
+static void create_logical_replication_slots(void);
 
 ClusterInfo old_cluster,
 			new_cluster;
@@ -188,6 +189,19 @@ main(int argc, char **argv)
 			  new_cluster.pgdata);
 	check_ok();
 
+	/*
+	 * Create logical replication slots if requested.
+	 *
+	 * Note: This must be done after doing pg_resetwal command because the
+	 * command will remove required WALs.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		start_postmaster(&new_cluster, true);
+		create_logical_replication_slots();
+		stop_postmaster(false);
+	}
+
 	if (user_opts.do_sync)
 	{
 		prep_status("Sync data directory to disk");
@@ -860,3 +874,50 @@ set_frozenxids(bool minmxid_only)
 
 	check_ok();
 }
+
+/*
+ * create_logical_replication_slots()
+ *
+ * Similar to create_new_objects() but only restores logical replication slots.
+ */
+static void
+create_logical_replication_slots(void)
+{
+	int			dbnum;
+
+	prep_status_progress("Restoring logical replication slots in the new cluster");
+
+	for (dbnum = 0; dbnum < old_cluster.dbarr.ndbs; dbnum++)
+	{
+		char		slots_file_name[MAXPGPATH],
+					log_file_name[MAXPGPATH];
+		DbInfo	   *old_db = &old_cluster.dbarr.dbs[dbnum];
+
+		pg_log(PG_STATUS, "%s", old_db->db_name);
+
+		snprintf(slots_file_name, sizeof(slots_file_name),
+				 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+		snprintf(log_file_name, sizeof(log_file_name),
+				 DB_DUMP_LOG_FILE_MASK, old_db->db_oid);
+
+		parallel_exec_prog(log_file_name,
+						   NULL,
+						   "\"%s/psql\" %s --echo-queries --set ON_ERROR_STOP=on "
+						   "--no-psqlrc --dbname %s -f \"%s/%s\"",
+						   new_cluster.bindir,
+						   cluster_conn_opts(&new_cluster),
+						   old_db->db_name,
+						   log_opts.dumpdir,
+						   slots_file_name);
+	}
+
+	/* reap all children */
+	while (reap_child(true) == true)
+		;
+
+	end_progress_output();
+	check_ok();
+
+	/* update new_cluster info now that we have objects in the databases */
+	get_db_and_rel_infos(&new_cluster);
+}
diff --git a/src/bin/pg_upgrade/pg_upgrade.h b/src/bin/pg_upgrade/pg_upgrade.h
index 3eea0139c7..5f3d7a407e 100644
--- a/src/bin/pg_upgrade/pg_upgrade.h
+++ b/src/bin/pg_upgrade/pg_upgrade.h
@@ -29,6 +29,7 @@
 /* contains both global db information and CREATE DATABASE commands */
 #define GLOBALS_DUMP_FILE	"pg_upgrade_dump_globals.sql"
 #define DB_DUMP_FILE_MASK	"pg_upgrade_dump_%u.custom"
+#define DB_DUMP_LOGICAL_SLOTS_FILE_MASK	"pg_upgrade_dump_%u_logical_slots.sql"
 
 /*
  * Base directories that include all the files generated internally, from the
@@ -304,6 +305,8 @@ typedef struct
 	transferMode transfer_mode; /* copy files or link them? */
 	int			jobs;			/* number of processes/threads to use */
 	char	   *socketdir;		/* directory to use for Unix sockets */
+	bool		include_logical_slots;	/* true -> dump and restore logical
+										 * replication slots */
 } UserOpts;
 
 typedef struct
diff --git a/src/bin/pg_upgrade/t/003_logical_replication.pl b/src/bin/pg_upgrade/t/003_logical_replication.pl
new file mode 100644
index 0000000000..4067535fa4
--- /dev/null
+++ b/src/bin/pg_upgrade/t/003_logical_replication.pl
@@ -0,0 +1,89 @@
+# Copyright (c) 2021-2023, PostgreSQL Global Development Group
+
+# Tests for logical replication, especially for upgrading publisher
+use strict;
+use warnings;
+use PostgreSQL::Test::Cluster;
+use PostgreSQL::Test::Utils;
+use Test::More;
+
+# Can be changed to test the other modes.
+my $mode = $ENV{PG_TEST_PG_UPGRADE_MODE} || '--copy';
+
+# Initialize publisher node
+my $old_publisher = PostgreSQL::Test::Cluster->new('old_publisher');
+$old_publisher->init(allows_streaming => 'logical');
+$old_publisher->start;
+
+# Initialize subscriber node
+my $subscriber = PostgreSQL::Test::Cluster->new('subscriber');
+$subscriber->init(allows_streaming => 'logical');
+$subscriber->start;
+
+$old_publisher->safe_psql('postgres',
+	"CREATE TABLE tbl AS SELECT generate_series(1,10) AS a");
+$subscriber->safe_psql('postgres', "CREATE TABLE tbl (a int)");
+
+# Setup logical replication
+my $old_connstr = $old_publisher->connstr . ' dbname=postgres';
+$old_publisher->safe_psql('postgres',
+	"CREATE PUBLICATION pub FOR ALL TABLES");
+$subscriber->safe_psql('postgres',
+	"CREATE SUBSCRIPTION sub CONNECTION '$old_connstr' PUBLICATION pub");
+
+# Wait for initial table sync to finish
+$subscriber->wait_for_subscription_sync($old_publisher, 'sub');
+
+my $result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(10), 'check initial rows on subscriber');
+
+# Preparations for upgrading publisher
+$old_publisher->stop;
+$subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub DISABLE");
+
+my $new_publisher = PostgreSQL::Test::Cluster->new('new_publisher');
+$new_publisher->init(allows_streaming => 'logical');
+
+my $bindir = $new_publisher->config_data('--bindir');
+
+# Run pg_upgrade. pg_upgrade_output.d is removed at the end
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
+		$mode,        '--include-logical-replication-slots'
+	],
+	'run of pg_upgrade of old publisher');
+ok( !-d $new_publisher->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+# Check whether the replication slot is copied
+$new_publisher->start;
+$result =
+  $new_publisher->safe_psql('postgres',
+	"SELECT count(*) FROM pg_replication_slots");
+is($result, qq(1), 'check the replication slot is copied to new publisher');
+
+# Change connection string and enable logical replication
+my $new_connstr = $new_publisher->connstr . ' dbname=postgres';
+
+$subscriber->safe_psql('postgres',
+	"ALTER SUBSCRIPTION sub CONNECTION '$new_connstr'");
+$subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub ENABLE");
+
+# Check whether changes on the new publisher get replicated to the subscriber
+$new_publisher->safe_psql('postgres',
+	"INSERT INTO tbl VALUES (generate_series(11, 20))");
+
+$new_publisher->wait_for_catchup('sub');
+
+$result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(20), 'check changes are shipped to subscriber');
+
+done_testing();
diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list
index b4058b88c3..7e999726c2 100644
--- a/src/tools/pgindent/typedefs.list
+++ b/src/tools/pgindent/typedefs.list
@@ -1479,6 +1479,7 @@ LogicalRepBeginData
 LogicalRepCommitData
 LogicalRepCommitPreparedTxnData
 LogicalRepCtxStruct
+LogicalReplicationSlotInfo
 LogicalRepMode
 LogicalRepMsgType
 LogicalRepPartMapEntry
-- 
2.27.0

From 7dc3d0a297e111ade378e47a43135652d7949715 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Tue, 4 Apr 2023 05:49:34 +0000
Subject: [PATCH v5] pg_upgrade: Add --include-logical-replication-slots option

This commit introduces a new pg_upgrade option called "--include-logical-replication-slots".
This allows nodes with logical replication slots to be upgraded. The commit can
be divided into two parts: one for pg_dump and another for pg_upgrade.

For pg_dump this commit includes a new option called "--logical-replication-slots-only".
This option can be used to dump logical replication slots. When this option is
specified, the slot_name, plugin, and two_phase parameters are extracted from
pg_replication_slots. An SQL file is then generated which executes
pg_create_logical_replication_slot() with the extracted parameters.

For pg_upgrade, when '--include-logical-replication-slots' is specified, it executes
pg_dump with the new "--logical-replication-slots-only" option and restores from the
dump. Apart from restoring schema, pg_resetwal must not be called after restoring
replication slots. This is because the command discards WAL files and starts from a
new segment, even if they are required by replication slots. This leads to an ERROR:
"requested WAL segment XXX has already been removed". To avoid this, replication slots
are restored at a different time than other objects, after running pg_resetwal.

The significant advantage of this commit is that it makes it easy to continue
logical replication even after upgrading the publisher node. Previously, pg_upgrade
allowed copying publications to a new node. With this new commit, adjusting the
connection string to the new publisher will cause the apply worker on the subscriber
to connect to the new publisher automatically. This enables seamless continuation
of logical replication, even after an upgrade.

Author: Hayato Kuroda
Reviewed-by: Peter Smith, Julien Rouhaud
---
 doc/src/sgml/ref/pg_dump.sgml                 |  10 ++
 doc/src/sgml/ref/pgupgrade.sgml               |  11 ++
 src/bin/pg_dump/pg_backup.h                   |   1 +
 src/bin/pg_dump/pg_dump.c                     | 146 ++++++++++++++++++
 src/bin/pg_dump/pg_dump.h                     |  16 +-
 src/bin/pg_dump/pg_dump_sort.c                |  11 +-
 src/bin/pg_upgrade/dump.c                     |  24 +++
 src/bin/pg_upgrade/meson.build                |   1 +
 src/bin/pg_upgrade/option.c                   |   7 +
 src/bin/pg_upgrade/pg_upgrade.c               |  61 ++++++++
 src/bin/pg_upgrade/pg_upgrade.h               |   3 +
 .../pg_upgrade/t/003_logical_replication.pl   |  89 +++++++++++
 src/tools/pgindent/typedefs.list              |   1 +
 13 files changed, 378 insertions(+), 3 deletions(-)
 create mode 100644 src/bin/pg_upgrade/t/003_logical_replication.pl

diff --git a/doc/src/sgml/ref/pg_dump.sgml b/doc/src/sgml/ref/pg_dump.sgml
index e81e35c13b..6e07f85281 100644
--- a/doc/src/sgml/ref/pg_dump.sgml
+++ b/doc/src/sgml/ref/pg_dump.sgml
@@ -1206,6 +1206,16 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--logical-replication-slots-only</option></term>
+      <listitem>
+       <para>
+        Dump only logical replication slots; not the schema (data definitions),
+        nor data. This is mainly used when upgrading nodes.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
        <term><option>-?</option></term>
        <term><option>--help</option></term>
diff --git a/doc/src/sgml/ref/pgupgrade.sgml b/doc/src/sgml/ref/pgupgrade.sgml
index 7816b4c685..94e90ff506 100644
--- a/doc/src/sgml/ref/pgupgrade.sgml
+++ b/doc/src/sgml/ref/pgupgrade.sgml
@@ -240,6 +240,17 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--include-logical-replication-slots</option></term>
+      <listitem>
+       <para>
+        Upgrade logical replication slots. Only permanent replication slots
+        are included. Note that pg_upgrade does not check the installation of
+        plugins.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
       <term><option>-?</option></term>
       <term><option>--help</option></term>
diff --git a/src/bin/pg_dump/pg_backup.h b/src/bin/pg_dump/pg_backup.h
index aba780ef4b..0a4e931f9b 100644
--- a/src/bin/pg_dump/pg_backup.h
+++ b/src/bin/pg_dump/pg_backup.h
@@ -187,6 +187,7 @@ typedef struct _dumpOptions
 	int			use_setsessauth;
 	int			enable_row_security;
 	int			load_via_partition_root;
+	int			logical_slots_only;
 
 	/* default, if no "inclusion" switches appear, is to dump everything */
 	bool		include_everything;
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index 967ced4eed..49526b0486 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -328,6 +328,9 @@ static void setupDumpWorker(Archive *AH);
 static TableInfo *getRootTableInfo(const TableInfo *tbinfo);
 static bool forcePartitionRootLoad(const TableInfo *tbinfo);
 
+static void getLogicalReplicationSlots(Archive *fout);
+static void dumpLogicalReplicationSlot(Archive *fout,
+									   const LogicalReplicationSlotInfo *slotinfo);
 
 int
 main(int argc, char **argv)
@@ -431,6 +434,7 @@ main(int argc, char **argv)
 		{"table-and-children", required_argument, NULL, 12},
 		{"exclude-table-and-children", required_argument, NULL, 13},
 		{"exclude-table-data-and-children", required_argument, NULL, 14},
+		{"logical-replication-slots-only", no_argument, NULL, 15},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -657,6 +661,10 @@ main(int argc, char **argv)
 										  optarg);
 				break;
 
+			case 15:			/* dump only replication slot(s) */
+				dopt.logical_slots_only = true;
+				break;
+
 			default:
 				/* getopt_long already emitted a complaint */
 				pg_log_error_hint("Try \"%s --help\" for more information.", progname);
@@ -714,6 +722,11 @@ main(int argc, char **argv)
 	if (dopt.do_nothing && dopt.dump_inserts == 0)
 		pg_fatal("option --on-conflict-do-nothing requires option --inserts, --rows-per-insert, or --column-inserts");
 
+	if (dopt.logical_slots_only && dopt.dataOnly)
+		pg_fatal("options --logical-replication-slots-only and -a/--data-only cannot be used together");
+	if (dopt.logical_slots_only && dopt.schemaOnly)
+		pg_fatal("options --logical-replication-slots-only and -s/--schema-only cannot be used together");
+
 	/* Identify archive format to emit */
 	archiveFormat = parseArchiveFormat(format, &archiveMode);
 
@@ -876,6 +889,16 @@ main(int argc, char **argv)
 			pg_fatal("no matching extensions were found");
 	}
 
+	/*
+	 * If dump logical-replication-slots-only was requested, dump only them
+	 * and skip everything else.
+	 */
+	if (dopt.logical_slots_only)
+	{
+		getLogicalReplicationSlots(fout);
+		goto dump;
+	}
+
 	/*
 	 * Dumping LOs is the default for dumps where an inclusion switch is not
 	 * used (an "include everything" dump).  -B can be used to exclude LOs
@@ -936,6 +959,8 @@ main(int argc, char **argv)
 	if (!dopt.no_security_labels)
 		collectSecLabels(fout);
 
+dump:
+
 	/* Lastly, create dummy objects to represent the section boundaries */
 	boundaryObjs = createBoundaryObjects();
 
@@ -1109,6 +1134,8 @@ help(const char *progname)
 			 "                               servers matching PATTERN\n"));
 	printf(_("  --inserts                    dump data as INSERT commands, rather than COPY\n"));
 	printf(_("  --load-via-partition-root    load partitions via the root table\n"));
+	printf(_("  --logical-replication-slots-only\n"
+			 "                               dump only logical replication slots, no schema or data\n"));
 	printf(_("  --no-comments                do not dump comments\n"));
 	printf(_("  --no-publications            do not dump publications\n"));
 	printf(_("  --no-security-labels         do not dump security label assignments\n"));
@@ -10381,6 +10408,10 @@ dumpDumpableObject(Archive *fout, DumpableObject *dobj)
 		case DO_SUBSCRIPTION:
 			dumpSubscription(fout, (const SubscriptionInfo *) dobj);
 			break;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			dumpLogicalReplicationSlot(fout,
+									   (const LogicalReplicationSlotInfo *) dobj);
+			break;
 		case DO_PRE_DATA_BOUNDARY:
 		case DO_POST_DATA_BOUNDARY:
 			/* never dumped, nothing to do */
@@ -18382,6 +18413,7 @@ addBoundaryDependencies(DumpableObject **dobjs, int numObjs,
 			case DO_PUBLICATION_REL:
 			case DO_PUBLICATION_TABLE_IN_SCHEMA:
 			case DO_SUBSCRIPTION:
+			case DO_LOGICAL_REPLICATION_SLOT:
 				/* Post-data objects: must come after the post-data boundary */
 				addObjectDependency(dobj, postDataBound->dumpId);
 				break;
@@ -18643,3 +18675,117 @@ appendReloptionsArrayAH(PQExpBuffer buffer, const char *reloptions,
 	if (!res)
 		pg_log_warning("could not parse %s array", "reloptions");
 }
+
+/*
+ * getLogicalReplicationSlots
+ *	  get information about replication slots
+ */
+static void
+getLogicalReplicationSlots(Archive *fout)
+{
+	PGresult   *res;
+	LogicalReplicationSlotInfo *slotinfo;
+	PQExpBuffer query;
+	DumpOptions *dopt = fout->dopt;
+
+	int			i_slotname;
+	int			i_plugin;
+	int			i_twophase;
+	int			i,
+				ntups;
+
+	/* Check whether we should dump or not */
+	if (fout->remoteVersion < 160000 || !dopt->logical_slots_only)
+		return;
+
+	query = createPQExpBuffer();
+
+	resetPQExpBuffer(query);
+
+	/*
+	 * Get replication slots.
+	 *
+	 * XXX: Which information must be extracted from old node? Currently three
+	 * attributes are extracted because they are used by
+	 * pg_create_logical_replication_slot().
+	 * XXX: Do we have to support physical slots?
+	 */
+	appendPQExpBufferStr(query,
+						 "SELECT slot_name, plugin, two_phase "
+						 "FROM pg_catalog.pg_replication_slots "
+						 "WHERE database = current_database() AND temporary = false "
+						 "AND wal_status IN ('reserved', 'extended');");
+
+	res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
+
+	ntups = PQntuples(res);
+
+	i_slotname = PQfnumber(res, "slot_name");
+	i_plugin = PQfnumber(res, "plugin");
+	i_twophase = PQfnumber(res, "two_phase");
+
+	slotinfo = pg_malloc(ntups * sizeof(LogicalReplicationSlotInfo));
+
+	for (i = 0; i < ntups; i++)
+	{
+		slotinfo[i].dobj.objType = DO_LOGICAL_REPLICATION_SLOT;
+
+		slotinfo[i].dobj.catId.tableoid = InvalidOid;
+		slotinfo[i].dobj.catId.oid = InvalidOid;
+		AssignDumpId(&slotinfo[i].dobj);
+
+		slotinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_slotname));
+
+		slotinfo[i].plugin = pg_strdup(PQgetvalue(res, i, i_plugin));
+		slotinfo[i].twophase = pg_strdup(PQgetvalue(res, i, i_twophase));
+
+		/*
+		 * Note: Currently we do not have any options to include/exclude slots
+		 * in dumping, so all the slots must be selected.
+		 */
+		slotinfo[i].dobj.dump = DUMP_COMPONENT_ALL;
+	}
+	PQclear(res);
+
+	destroyPQExpBuffer(query);
+}
+
+/*
+ * dumpLogicalReplicationSlot
+ *	  dump creation functions for the given logical replication slots
+ */
+static void
+dumpLogicalReplicationSlot(Archive *fout,
+						   const LogicalReplicationSlotInfo *slotinfo)
+{
+	DumpOptions *dopt = fout->dopt;
+
+	if (!dopt->logical_slots_only)
+		return;
+
+	if (slotinfo->dobj.dump & DUMP_COMPONENT_DEFINITION)
+	{
+		PQExpBuffer query = createPQExpBuffer();
+		char	   *slotname = pg_strdup(slotinfo->dobj.name);
+
+		/*
+		 * XXX: For simplification, pg_create_logical_replication_slot() is
+		 * used. Is it sufficient?
+		 */
+		appendPQExpBuffer(query, "SELECT pg_catalog.pg_create_logical_replication_slot('%s', ",
+						  slotname);
+		appendStringLiteralAH(query, slotinfo->plugin, fout);
+		appendPQExpBuffer(query, ", ");
+		appendStringLiteralAH(query, slotinfo->twophase, fout);
+		appendPQExpBuffer(query, ");");
+
+		ArchiveEntry(fout, slotinfo->dobj.catId, slotinfo->dobj.dumpId,
+					 ARCHIVE_OPTS(.tag = slotname,
+								  .description = "REPLICATION SLOT",
+								  .section = SECTION_POST_DATA,
+								  .createStmt = query->data));
+
+		pfree(slotname);
+		destroyPQExpBuffer(query);
+	}
+}
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index 765fe6399a..f5535816d2 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -82,7 +82,8 @@ typedef enum
 	DO_PUBLICATION,
 	DO_PUBLICATION_REL,
 	DO_PUBLICATION_TABLE_IN_SCHEMA,
-	DO_SUBSCRIPTION
+	DO_SUBSCRIPTION,
+	DO_LOGICAL_REPLICATION_SLOT
 } DumpableObjectType;
 
 /*
@@ -666,6 +667,19 @@ typedef struct _SubscriptionInfo
 	char	   *subpasswordrequired;
 } SubscriptionInfo;
 
+/*
+ * The LogicalReplicationSlotInfo struct is used to represent replication
+ * slots.
+ * XXX: add more attrbutes if needed
+ */
+typedef struct _LogicalReplicationSlotInfo
+{
+	DumpableObject dobj;
+	char	   *plugin;
+	char	   *slottype;
+	char	   *twophase;
+} LogicalReplicationSlotInfo;
+
 /*
  *	common utility functions
  */
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index 8266c117a3..b36ced8c8e 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -93,6 +93,7 @@ enum dbObjectTypePriorities
 	PRIO_PUBLICATION_REL,
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,
 	PRIO_SUBSCRIPTION,
+	PRIO_LOGICAL_REPLICATION_SLOT,
 	PRIO_DEFAULT_ACL,			/* done in ACL pass */
 	PRIO_EVENT_TRIGGER,			/* must be next to last! */
 	PRIO_REFRESH_MATVIEW		/* must be last! */
@@ -146,10 +147,11 @@ static const int dbObjectTypePriority[] =
 	PRIO_PUBLICATION,			/* DO_PUBLICATION */
 	PRIO_PUBLICATION_REL,		/* DO_PUBLICATION_REL */
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,	/* DO_PUBLICATION_TABLE_IN_SCHEMA */
-	PRIO_SUBSCRIPTION			/* DO_SUBSCRIPTION */
+	PRIO_SUBSCRIPTION,			/* DO_SUBSCRIPTION */
+	PRIO_LOGICAL_REPLICATION_SLOT	/* DO_LOGICAL_REPLICATION_SLOT */
 };
 
-StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_SUBSCRIPTION + 1),
+StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_LOGICAL_REPLICATION_SLOT + 1),
 				 "array length mismatch");
 
 static DumpId preDataBoundId;
@@ -1498,6 +1500,11 @@ describeDumpableObject(DumpableObject *obj, char *buf, int bufsize)
 					 "SUBSCRIPTION (ID %d OID %u)",
 					 obj->dumpId, obj->catId.oid);
 			return;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			snprintf(buf, bufsize,
+					 "REPLICATION SLOT (ID %d NAME %s)",
+					 obj->dumpId, obj->name);
+			return;
 		case DO_PRE_DATA_BOUNDARY:
 			snprintf(buf, bufsize,
 					 "PRE-DATA BOUNDARY  (ID %d)",
diff --git a/src/bin/pg_upgrade/dump.c b/src/bin/pg_upgrade/dump.c
index 6c8c82dca8..e6b90864f5 100644
--- a/src/bin/pg_upgrade/dump.c
+++ b/src/bin/pg_upgrade/dump.c
@@ -59,6 +59,30 @@ generate_old_dump(void)
 						   log_opts.dumpdir,
 						   sql_file_name, escaped_connstr.data);
 
+		/*
+		 * Dump logical replication slots if needed.
+		 *
+		 * XXX We cannot dump replication slots at the same time as the schema
+		 * dump because we need to separate the timing of restoring
+		 * replication slots and other objects. Replication slots, in
+		 * particular, should not be restored before executing the pg_resetwal
+		 * command because it will remove WALs that are required by the slots.
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			char		slots_file_name[MAXPGPATH];
+
+			snprintf(slots_file_name, sizeof(slots_file_name),
+					 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+			parallel_exec_prog(log_file_name, NULL,
+							   "\"%s/pg_dump\" %s --logical-replication-slots-only "
+							   "--quote-all-identifiers --binary-upgrade %s "
+							   "--file=\"%s/%s\" %s",
+							   new_cluster.bindir, cluster_conn_opts(&old_cluster),
+							   log_opts.verbose ? "--verbose" : "",
+							   log_opts.dumpdir,
+							   slots_file_name, escaped_connstr.data);
+		}
 		termPQExpBuffer(&escaped_connstr);
 	}
 
diff --git a/src/bin/pg_upgrade/meson.build b/src/bin/pg_upgrade/meson.build
index 12a97f84e2..7f5d48b7e1 100644
--- a/src/bin/pg_upgrade/meson.build
+++ b/src/bin/pg_upgrade/meson.build
@@ -42,6 +42,7 @@ tests += {
     'tests': [
       't/001_basic.pl',
       't/002_pg_upgrade.pl',
+      't/003_logical_replication.pl',
     ],
     'test_kwargs': {'priority': 40}, # pg_upgrade tests are slow
   },
diff --git a/src/bin/pg_upgrade/option.c b/src/bin/pg_upgrade/option.c
index 8869b6b60d..f7b5ec9879 100644
--- a/src/bin/pg_upgrade/option.c
+++ b/src/bin/pg_upgrade/option.c
@@ -57,6 +57,7 @@ parseCommandLine(int argc, char *argv[])
 		{"verbose", no_argument, NULL, 'v'},
 		{"clone", no_argument, NULL, 1},
 		{"copy", no_argument, NULL, 2},
+		{"include-logical-replication-slots", no_argument, NULL, 3},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -199,6 +200,10 @@ parseCommandLine(int argc, char *argv[])
 				user_opts.transfer_mode = TRANSFER_MODE_COPY;
 				break;
 
+			case 3:
+				user_opts.include_logical_slots = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						os_info.progname);
@@ -289,6 +294,8 @@ usage(void)
 	printf(_("  -V, --version                 display version information, then exit\n"));
 	printf(_("  --clone                       clone instead of copying files to new cluster\n"));
 	printf(_("  --copy                        copy files to new cluster (default)\n"));
+	printf(_("  --include-logical-replication-slots\n"
+			 "                                upgrade logical replication slots\n"));
 	printf(_("  -?, --help                    show this help, then exit\n"));
 	printf(_("\n"
 			 "Before running pg_upgrade you must:\n"
diff --git a/src/bin/pg_upgrade/pg_upgrade.c b/src/bin/pg_upgrade/pg_upgrade.c
index 75bab0a04c..1241060f4e 100644
--- a/src/bin/pg_upgrade/pg_upgrade.c
+++ b/src/bin/pg_upgrade/pg_upgrade.c
@@ -59,6 +59,7 @@ static void copy_xact_xlog_xid(void);
 static void set_frozenxids(bool minmxid_only);
 static void make_outputdirs(char *pgdata);
 static void setup(char *argv0, bool *live_check);
+static void create_logical_replication_slots(void);
 
 ClusterInfo old_cluster,
 			new_cluster;
@@ -188,6 +189,19 @@ main(int argc, char **argv)
 			  new_cluster.pgdata);
 	check_ok();
 
+	/*
+	 * Create logical replication slots if requested.
+	 *
+	 * Note: This must be done after doing pg_resetwal command because the
+	 * command will remove required WALs.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		start_postmaster(&new_cluster, true);
+		create_logical_replication_slots();
+		stop_postmaster(false);
+	}
+
 	if (user_opts.do_sync)
 	{
 		prep_status("Sync data directory to disk");
@@ -860,3 +874,50 @@ set_frozenxids(bool minmxid_only)
 
 	check_ok();
 }
+
+/*
+ * create_logical_replication_slots()
+ *
+ * Similar to create_new_objects() but only restores logical replication slots.
+ */
+static void
+create_logical_replication_slots(void)
+{
+	int			dbnum;
+
+	prep_status_progress("Restoring logical replication slots in the new cluster");
+
+	for (dbnum = 0; dbnum < old_cluster.dbarr.ndbs; dbnum++)
+	{
+		char		slots_file_name[MAXPGPATH],
+					log_file_name[MAXPGPATH];
+		DbInfo	   *old_db = &old_cluster.dbarr.dbs[dbnum];
+
+		pg_log(PG_STATUS, "%s", old_db->db_name);
+
+		snprintf(slots_file_name, sizeof(slots_file_name),
+				 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+		snprintf(log_file_name, sizeof(log_file_name),
+				 DB_DUMP_LOG_FILE_MASK, old_db->db_oid);
+
+		parallel_exec_prog(log_file_name,
+						   NULL,
+						   "\"%s/psql\" %s --echo-queries --set ON_ERROR_STOP=on "
+						   "--no-psqlrc --dbname %s -f \"%s/%s\"",
+						   new_cluster.bindir,
+						   cluster_conn_opts(&new_cluster),
+						   old_db->db_name,
+						   log_opts.dumpdir,
+						   slots_file_name);
+	}
+
+	/* reap all children */
+	while (reap_child(true) == true)
+		;
+
+	end_progress_output();
+	check_ok();
+
+	/* update new_cluster info now that we have objects in the databases */
+	get_db_and_rel_infos(&new_cluster);
+}
diff --git a/src/bin/pg_upgrade/pg_upgrade.h b/src/bin/pg_upgrade/pg_upgrade.h
index 3eea0139c7..5f3d7a407e 100644
--- a/src/bin/pg_upgrade/pg_upgrade.h
+++ b/src/bin/pg_upgrade/pg_upgrade.h
@@ -29,6 +29,7 @@
 /* contains both global db information and CREATE DATABASE commands */
 #define GLOBALS_DUMP_FILE	"pg_upgrade_dump_globals.sql"
 #define DB_DUMP_FILE_MASK	"pg_upgrade_dump_%u.custom"
+#define DB_DUMP_LOGICAL_SLOTS_FILE_MASK	"pg_upgrade_dump_%u_logical_slots.sql"
 
 /*
  * Base directories that include all the files generated internally, from the
@@ -304,6 +305,8 @@ typedef struct
 	transferMode transfer_mode; /* copy files or link them? */
 	int			jobs;			/* number of processes/threads to use */
 	char	   *socketdir;		/* directory to use for Unix sockets */
+	bool		include_logical_slots;	/* true -> dump and restore logical
+										 * replication slots */
 } UserOpts;
 
 typedef struct
diff --git a/src/bin/pg_upgrade/t/003_logical_replication.pl b/src/bin/pg_upgrade/t/003_logical_replication.pl
new file mode 100644
index 0000000000..4067535fa4
--- /dev/null
+++ b/src/bin/pg_upgrade/t/003_logical_replication.pl
@@ -0,0 +1,89 @@
+# Copyright (c) 2021-2023, PostgreSQL Global Development Group
+
+# Tests for logical replication, especially for upgrading publisher
+use strict;
+use warnings;
+use PostgreSQL::Test::Cluster;
+use PostgreSQL::Test::Utils;
+use Test::More;
+
+# Can be changed to test the other modes.
+my $mode = $ENV{PG_TEST_PG_UPGRADE_MODE} || '--copy';
+
+# Initialize publisher node
+my $old_publisher = PostgreSQL::Test::Cluster->new('old_publisher');
+$old_publisher->init(allows_streaming => 'logical');
+$old_publisher->start;
+
+# Initialize subscriber node
+my $subscriber = PostgreSQL::Test::Cluster->new('subscriber');
+$subscriber->init(allows_streaming => 'logical');
+$subscriber->start;
+
+$old_publisher->safe_psql('postgres',
+	"CREATE TABLE tbl AS SELECT generate_series(1,10) AS a");
+$subscriber->safe_psql('postgres', "CREATE TABLE tbl (a int)");
+
+# Setup logical replication
+my $old_connstr = $old_publisher->connstr . ' dbname=postgres';
+$old_publisher->safe_psql('postgres',
+	"CREATE PUBLICATION pub FOR ALL TABLES");
+$subscriber->safe_psql('postgres',
+	"CREATE SUBSCRIPTION sub CONNECTION '$old_connstr' PUBLICATION pub");
+
+# Wait for initial table sync to finish
+$subscriber->wait_for_subscription_sync($old_publisher, 'sub');
+
+my $result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(10), 'check initial rows on subscriber');
+
+# Preparations for upgrading publisher
+$old_publisher->stop;
+$subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub DISABLE");
+
+my $new_publisher = PostgreSQL::Test::Cluster->new('new_publisher');
+$new_publisher->init(allows_streaming => 'logical');
+
+my $bindir = $new_publisher->config_data('--bindir');
+
+# Run pg_upgrade. pg_upgrade_output.d is removed at the end
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
+		$mode,        '--include-logical-replication-slots'
+	],
+	'run of pg_upgrade of old publisher');
+ok( !-d $new_publisher->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+# Check whether the replication slot is copied
+$new_publisher->start;
+$result =
+  $new_publisher->safe_psql('postgres',
+	"SELECT count(*) FROM pg_replication_slots");
+is($result, qq(1), 'check the replication slot is copied to new publisher');
+
+# Change connection string and enable logical replication
+my $new_connstr = $new_publisher->connstr . ' dbname=postgres';
+
+$subscriber->safe_psql('postgres',
+	"ALTER SUBSCRIPTION sub CONNECTION '$new_connstr'");
+$subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub ENABLE");
+
+# Check whether changes on the new publisher get replicated to the subscriber
+$new_publisher->safe_psql('postgres',
+	"INSERT INTO tbl VALUES (generate_series(11, 20))");
+
+$new_publisher->wait_for_catchup('sub');
+
+$result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(20), 'check changes are shipped to subscriber');
+
+done_testing();
diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list
index b4058b88c3..7e999726c2 100644
--- a/src/tools/pgindent/typedefs.list
+++ b/src/tools/pgindent/typedefs.list
@@ -1479,6 +1479,7 @@ LogicalRepBeginData
 LogicalRepCommitData
 LogicalRepCommitPreparedTxnData
 LogicalRepCtxStruct
+LogicalReplicationSlotInfo
 LogicalRepMode
 LogicalRepMsgType
 LogicalRepPartMapEntry
-- 
2.27.0

From 498c48d2573118617fda33fcd7433d874f8cad48 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Tue, 4 Apr 2023 05:49:34 +0000
Subject: [PATCH v6] pg_upgrade: Add --include-logical-replication-slots option

This commit introduces a new pg_upgrade option called "--include-logical-replication-slots".
This allows nodes with logical replication slots to be upgraded. The commit can
be divided into two parts: one for pg_dump and another for pg_upgrade.

For pg_dump this commit includes a new option called "--logical-replication-slots-only".
This option can be used to dump logical replication slots. When this option is
specified, the slot_name, plugin, and two_phase parameters are extracted from
pg_replication_slots. An SQL file is then generated which executes
pg_create_logical_replication_slot() with the extracted parameters.

For pg_upgrade, when '--include-logical-replication-slots' is specified, it executes
pg_dump with the new "--logical-replication-slots-only" option and restores from the
dump. Apart from restoring schema, pg_resetwal must not be called after restoring
replication slots. This is because the command discards WAL files and starts from a
new segment, even if they are required by replication slots. This leads to an ERROR:
"requested WAL segment XXX has already been removed". To avoid this, replication slots
are restored at a different time than other objects, after running pg_resetwal.

The significant advantage of this commit is that it makes it easy to continue
logical replication even after upgrading the publisher node. Previously, pg_upgrade
allowed copying publications to a new node. With this new commit, adjusting the
connection string to the new publisher will cause the apply worker on the subscriber
to connect to the new publisher automatically. This enables seamless continuation
of logical replication, even after an upgrade.

Author: Hayato Kuroda
Reviewed-by: Peter Smith, Julien Rouhaud
---
 doc/src/sgml/ref/pg_dump.sgml                 |  10 ++
 doc/src/sgml/ref/pgupgrade.sgml               |  11 ++
 src/bin/pg_dump/pg_backup.h                   |   1 +
 src/bin/pg_dump/pg_dump.c                     | 147 ++++++++++++++++++
 src/bin/pg_dump/pg_dump.h                     |  17 +-
 src/bin/pg_dump/pg_dump_sort.c                |  11 +-
 src/bin/pg_upgrade/dump.c                     |  24 +++
 src/bin/pg_upgrade/meson.build                |   1 +
 src/bin/pg_upgrade/option.c                   |   7 +
 src/bin/pg_upgrade/pg_upgrade.c               |  61 ++++++++
 src/bin/pg_upgrade/pg_upgrade.h               |   3 +
 .../t/003_logical_replication_slots.pl        |  90 +++++++++++
 src/tools/pgindent/typedefs.list              |   1 +
 13 files changed, 381 insertions(+), 3 deletions(-)
 create mode 100644 src/bin/pg_upgrade/t/003_logical_replication_slots.pl

diff --git a/doc/src/sgml/ref/pg_dump.sgml b/doc/src/sgml/ref/pg_dump.sgml
index e81e35c13b..6e07f85281 100644
--- a/doc/src/sgml/ref/pg_dump.sgml
+++ b/doc/src/sgml/ref/pg_dump.sgml
@@ -1206,6 +1206,16 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--logical-replication-slots-only</option></term>
+      <listitem>
+       <para>
+        Dump only logical replication slots; not the schema (data definitions),
+        nor data. This is mainly used when upgrading nodes.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
        <term><option>-?</option></term>
        <term><option>--help</option></term>
diff --git a/doc/src/sgml/ref/pgupgrade.sgml b/doc/src/sgml/ref/pgupgrade.sgml
index 7816b4c685..94e90ff506 100644
--- a/doc/src/sgml/ref/pgupgrade.sgml
+++ b/doc/src/sgml/ref/pgupgrade.sgml
@@ -240,6 +240,17 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--include-logical-replication-slots</option></term>
+      <listitem>
+       <para>
+        Upgrade logical replication slots. Only permanent replication slots
+        are included. Note that pg_upgrade does not check the installation of
+        plugins.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
       <term><option>-?</option></term>
       <term><option>--help</option></term>
diff --git a/src/bin/pg_dump/pg_backup.h b/src/bin/pg_dump/pg_backup.h
index aba780ef4b..0a4e931f9b 100644
--- a/src/bin/pg_dump/pg_backup.h
+++ b/src/bin/pg_dump/pg_backup.h
@@ -187,6 +187,7 @@ typedef struct _dumpOptions
 	int			use_setsessauth;
 	int			enable_row_security;
 	int			load_via_partition_root;
+	int			logical_slots_only;
 
 	/* default, if no "inclusion" switches appear, is to dump everything */
 	bool		include_everything;
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index 73a6c964ba..6cff4009b3 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -328,6 +328,9 @@ static void setupDumpWorker(Archive *AH);
 static TableInfo *getRootTableInfo(const TableInfo *tbinfo);
 static bool forcePartitionRootLoad(const TableInfo *tbinfo);
 
+static void getLogicalReplicationSlots(Archive *fout);
+static void dumpLogicalReplicationSlot(Archive *fout,
+									   const LogicalReplicationSlotInfo *slotinfo);
 
 int
 main(int argc, char **argv)
@@ -431,6 +434,7 @@ main(int argc, char **argv)
 		{"table-and-children", required_argument, NULL, 12},
 		{"exclude-table-and-children", required_argument, NULL, 13},
 		{"exclude-table-data-and-children", required_argument, NULL, 14},
+		{"logical-replication-slots-only", no_argument, NULL, 15},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -657,6 +661,10 @@ main(int argc, char **argv)
 										  optarg);
 				break;
 
+			case 15:			/* dump only replication slot(s) */
+				dopt.logical_slots_only = true;
+				break;
+
 			default:
 				/* getopt_long already emitted a complaint */
 				pg_log_error_hint("Try \"%s --help\" for more information.", progname);
@@ -714,6 +722,11 @@ main(int argc, char **argv)
 	if (dopt.do_nothing && dopt.dump_inserts == 0)
 		pg_fatal("option --on-conflict-do-nothing requires option --inserts, --rows-per-insert, or --column-inserts");
 
+	if (dopt.logical_slots_only && dopt.dataOnly)
+		pg_fatal("options --logical-replication-slots-only and -a/--data-only cannot be used together");
+	if (dopt.logical_slots_only && dopt.schemaOnly)
+		pg_fatal("options --logical-replication-slots-only and -s/--schema-only cannot be used together");
+
 	/* Identify archive format to emit */
 	archiveFormat = parseArchiveFormat(format, &archiveMode);
 
@@ -876,6 +889,16 @@ main(int argc, char **argv)
 			pg_fatal("no matching extensions were found");
 	}
 
+	/*
+	 * If dump logical-replication-slots-only was requested, dump only them
+	 * and skip everything else.
+	 */
+	if (dopt.logical_slots_only)
+	{
+		getLogicalReplicationSlots(fout);
+		goto dump;
+	}
+
 	/*
 	 * Dumping LOs is the default for dumps where an inclusion switch is not
 	 * used (an "include everything" dump).  -B can be used to exclude LOs
@@ -936,6 +959,8 @@ main(int argc, char **argv)
 	if (!dopt.no_security_labels)
 		collectSecLabels(fout);
 
+dump:
+
 	/* Lastly, create dummy objects to represent the section boundaries */
 	boundaryObjs = createBoundaryObjects();
 
@@ -1109,6 +1134,8 @@ help(const char *progname)
 			 "                               servers matching PATTERN\n"));
 	printf(_("  --inserts                    dump data as INSERT commands, rather than COPY\n"));
 	printf(_("  --load-via-partition-root    load partitions via the root table\n"));
+	printf(_("  --logical-replication-slots-only\n"
+			 "                               dump only logical replication slots, no schema or data\n"));
 	printf(_("  --no-comments                do not dump comments\n"));
 	printf(_("  --no-publications            do not dump publications\n"));
 	printf(_("  --no-security-labels         do not dump security label assignments\n"));
@@ -10252,6 +10279,10 @@ dumpDumpableObject(Archive *fout, DumpableObject *dobj)
 		case DO_SUBSCRIPTION:
 			dumpSubscription(fout, (const SubscriptionInfo *) dobj);
 			break;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			dumpLogicalReplicationSlot(fout,
+									   (const LogicalReplicationSlotInfo *) dobj);
+			break;
 		case DO_PRE_DATA_BOUNDARY:
 		case DO_POST_DATA_BOUNDARY:
 			/* never dumped, nothing to do */
@@ -18227,6 +18258,7 @@ addBoundaryDependencies(DumpableObject **dobjs, int numObjs,
 			case DO_PUBLICATION_REL:
 			case DO_PUBLICATION_TABLE_IN_SCHEMA:
 			case DO_SUBSCRIPTION:
+			case DO_LOGICAL_REPLICATION_SLOT:
 				/* Post-data objects: must come after the post-data boundary */
 				addObjectDependency(dobj, postDataBound->dumpId);
 				break;
@@ -18488,3 +18520,118 @@ appendReloptionsArrayAH(PQExpBuffer buffer, const char *reloptions,
 	if (!res)
 		pg_log_warning("could not parse %s array", "reloptions");
 }
+
+/*
+ * getLogicalReplicationSlots
+ *	  get information about replication slots
+ */
+static void
+getLogicalReplicationSlots(Archive *fout)
+{
+	PGresult   *res;
+	LogicalReplicationSlotInfo *slotinfo;
+	PQExpBuffer query;
+	DumpOptions *dopt = fout->dopt;
+
+	int			i_slotname;
+	int			i_plugin;
+	int			i_twophase;
+	int			i,
+				ntups;
+
+	/* Check whether we should dump or not */
+	if (fout->remoteVersion < 160000 || !dopt->logical_slots_only)
+		return;
+
+	query = createPQExpBuffer();
+
+	resetPQExpBuffer(query);
+
+	/*
+	 * Get replication slots.
+	 *
+	 * XXX: Which information must be extracted from old node? Currently three
+	 * attributes are extracted because they are used by
+	 * pg_create_logical_replication_slot().
+	 *
+	 * XXX: Do we have to support physical slots?
+	 */
+	appendPQExpBufferStr(query,
+						 "SELECT slot_name, plugin, two_phase "
+						 "FROM pg_catalog.pg_replication_slots "
+						 "WHERE database = current_database() AND temporary = false "
+						 "AND wal_status IN ('reserved', 'extended');");
+
+	res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
+
+	ntups = PQntuples(res);
+
+	i_slotname = PQfnumber(res, "slot_name");
+	i_plugin = PQfnumber(res, "plugin");
+	i_twophase = PQfnumber(res, "two_phase");
+
+	slotinfo = pg_malloc(ntups * sizeof(LogicalReplicationSlotInfo));
+
+	for (i = 0; i < ntups; i++)
+	{
+		slotinfo[i].dobj.objType = DO_LOGICAL_REPLICATION_SLOT;
+
+		slotinfo[i].dobj.catId.tableoid = InvalidOid;
+		slotinfo[i].dobj.catId.oid = InvalidOid;
+		AssignDumpId(&slotinfo[i].dobj);
+
+		slotinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_slotname));
+
+		slotinfo[i].plugin = pg_strdup(PQgetvalue(res, i, i_plugin));
+		slotinfo[i].twophase = pg_strdup(PQgetvalue(res, i, i_twophase));
+
+		/*
+		 * Note: Currently we do not have any options to include/exclude slots
+		 * in dumping, so all the slots must be selected.
+		 */
+		slotinfo[i].dobj.dump = DUMP_COMPONENT_ALL;
+	}
+	PQclear(res);
+
+	destroyPQExpBuffer(query);
+}
+
+/*
+ * dumpLogicalReplicationSlot
+ *	  dump creation functions for the given logical replication slots
+ */
+static void
+dumpLogicalReplicationSlot(Archive *fout,
+						   const LogicalReplicationSlotInfo *slotinfo)
+{
+	DumpOptions *dopt = fout->dopt;
+
+	if (!dopt->logical_slots_only)
+		return;
+
+	if (slotinfo->dobj.dump & DUMP_COMPONENT_DEFINITION)
+	{
+		PQExpBuffer query = createPQExpBuffer();
+		char	   *slotname = pg_strdup(slotinfo->dobj.name);
+
+		/*
+		 * XXX: For simplification, pg_create_logical_replication_slot() is
+		 * used. Is it sufficient?
+		 */
+		appendPQExpBuffer(query, "SELECT pg_catalog.pg_create_logical_replication_slot('%s', ",
+						  slotname);
+		appendStringLiteralAH(query, slotinfo->plugin, fout);
+		appendPQExpBuffer(query, ", ");
+		appendStringLiteralAH(query, slotinfo->twophase, fout);
+		appendPQExpBuffer(query, ");");
+
+		ArchiveEntry(fout, slotinfo->dobj.catId, slotinfo->dobj.dumpId,
+					 ARCHIVE_OPTS(.tag = slotname,
+								  .description = "REPLICATION SLOT",
+								  .section = SECTION_POST_DATA,
+								  .createStmt = query->data));
+
+		pfree(slotname);
+		destroyPQExpBuffer(query);
+	}
+}
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index ed6ce41ad7..8028ccf6ff 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -82,7 +82,8 @@ typedef enum
 	DO_PUBLICATION,
 	DO_PUBLICATION_REL,
 	DO_PUBLICATION_TABLE_IN_SCHEMA,
-	DO_SUBSCRIPTION
+	DO_SUBSCRIPTION,
+	DO_LOGICAL_REPLICATION_SLOT
 } DumpableObjectType;
 
 /*
@@ -666,6 +667,20 @@ typedef struct _SubscriptionInfo
 	char	   *subpasswordrequired;
 } SubscriptionInfo;
 
+/*
+ * The LogicalReplicationSlotInfo struct is used to represent replication
+ * slots.
+ *
+ * XXX: add more attributes if needed
+ */
+typedef struct _LogicalReplicationSlotInfo
+{
+	DumpableObject dobj;
+	char	   *plugin;
+	char	   *slottype;
+	char	   *twophase;
+} LogicalReplicationSlotInfo;
+
 /*
  *	common utility functions
  */
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index 8266c117a3..7c37873743 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -93,6 +93,7 @@ enum dbObjectTypePriorities
 	PRIO_PUBLICATION_REL,
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,
 	PRIO_SUBSCRIPTION,
+	PRIO_LOGICAL_REPLICATION_SLOT,
 	PRIO_DEFAULT_ACL,			/* done in ACL pass */
 	PRIO_EVENT_TRIGGER,			/* must be next to last! */
 	PRIO_REFRESH_MATVIEW		/* must be last! */
@@ -146,10 +147,11 @@ static const int dbObjectTypePriority[] =
 	PRIO_PUBLICATION,			/* DO_PUBLICATION */
 	PRIO_PUBLICATION_REL,		/* DO_PUBLICATION_REL */
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,	/* DO_PUBLICATION_TABLE_IN_SCHEMA */
-	PRIO_SUBSCRIPTION			/* DO_SUBSCRIPTION */
+	PRIO_SUBSCRIPTION,			/* DO_SUBSCRIPTION */
+	PRIO_LOGICAL_REPLICATION_SLOT	/* DO_LOGICAL_REPLICATION_SLOT */
 };
 
-StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_SUBSCRIPTION + 1),
+StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_LOGICAL_REPLICATION_SLOT + 1),
 				 "array length mismatch");
 
 static DumpId preDataBoundId;
@@ -1498,6 +1500,11 @@ describeDumpableObject(DumpableObject *obj, char *buf, int bufsize)
 					 "SUBSCRIPTION (ID %d OID %u)",
 					 obj->dumpId, obj->catId.oid);
 			return;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			snprintf(buf, bufsize,
+					 "LOGICAL REPLICATION SLOT (ID %d NAME %s)",
+					 obj->dumpId, obj->name);
+			return;
 		case DO_PRE_DATA_BOUNDARY:
 			snprintf(buf, bufsize,
 					 "PRE-DATA BOUNDARY  (ID %d)",
diff --git a/src/bin/pg_upgrade/dump.c b/src/bin/pg_upgrade/dump.c
index 6c8c82dca8..e6b90864f5 100644
--- a/src/bin/pg_upgrade/dump.c
+++ b/src/bin/pg_upgrade/dump.c
@@ -59,6 +59,30 @@ generate_old_dump(void)
 						   log_opts.dumpdir,
 						   sql_file_name, escaped_connstr.data);
 
+		/*
+		 * Dump logical replication slots if needed.
+		 *
+		 * XXX We cannot dump replication slots at the same time as the schema
+		 * dump because we need to separate the timing of restoring
+		 * replication slots and other objects. Replication slots, in
+		 * particular, should not be restored before executing the pg_resetwal
+		 * command because it will remove WALs that are required by the slots.
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			char		slots_file_name[MAXPGPATH];
+
+			snprintf(slots_file_name, sizeof(slots_file_name),
+					 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+			parallel_exec_prog(log_file_name, NULL,
+							   "\"%s/pg_dump\" %s --logical-replication-slots-only "
+							   "--quote-all-identifiers --binary-upgrade %s "
+							   "--file=\"%s/%s\" %s",
+							   new_cluster.bindir, cluster_conn_opts(&old_cluster),
+							   log_opts.verbose ? "--verbose" : "",
+							   log_opts.dumpdir,
+							   slots_file_name, escaped_connstr.data);
+		}
 		termPQExpBuffer(&escaped_connstr);
 	}
 
diff --git a/src/bin/pg_upgrade/meson.build b/src/bin/pg_upgrade/meson.build
index 12a97f84e2..228f29b688 100644
--- a/src/bin/pg_upgrade/meson.build
+++ b/src/bin/pg_upgrade/meson.build
@@ -42,6 +42,7 @@ tests += {
     'tests': [
       't/001_basic.pl',
       't/002_pg_upgrade.pl',
+      't/003_logical_replication_slots.pl',
     ],
     'test_kwargs': {'priority': 40}, # pg_upgrade tests are slow
   },
diff --git a/src/bin/pg_upgrade/option.c b/src/bin/pg_upgrade/option.c
index 8869b6b60d..f7b5ec9879 100644
--- a/src/bin/pg_upgrade/option.c
+++ b/src/bin/pg_upgrade/option.c
@@ -57,6 +57,7 @@ parseCommandLine(int argc, char *argv[])
 		{"verbose", no_argument, NULL, 'v'},
 		{"clone", no_argument, NULL, 1},
 		{"copy", no_argument, NULL, 2},
+		{"include-logical-replication-slots", no_argument, NULL, 3},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -199,6 +200,10 @@ parseCommandLine(int argc, char *argv[])
 				user_opts.transfer_mode = TRANSFER_MODE_COPY;
 				break;
 
+			case 3:
+				user_opts.include_logical_slots = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						os_info.progname);
@@ -289,6 +294,8 @@ usage(void)
 	printf(_("  -V, --version                 display version information, then exit\n"));
 	printf(_("  --clone                       clone instead of copying files to new cluster\n"));
 	printf(_("  --copy                        copy files to new cluster (default)\n"));
+	printf(_("  --include-logical-replication-slots\n"
+			 "                                upgrade logical replication slots\n"));
 	printf(_("  -?, --help                    show this help, then exit\n"));
 	printf(_("\n"
 			 "Before running pg_upgrade you must:\n"
diff --git a/src/bin/pg_upgrade/pg_upgrade.c b/src/bin/pg_upgrade/pg_upgrade.c
index 75bab0a04c..1241060f4e 100644
--- a/src/bin/pg_upgrade/pg_upgrade.c
+++ b/src/bin/pg_upgrade/pg_upgrade.c
@@ -59,6 +59,7 @@ static void copy_xact_xlog_xid(void);
 static void set_frozenxids(bool minmxid_only);
 static void make_outputdirs(char *pgdata);
 static void setup(char *argv0, bool *live_check);
+static void create_logical_replication_slots(void);
 
 ClusterInfo old_cluster,
 			new_cluster;
@@ -188,6 +189,19 @@ main(int argc, char **argv)
 			  new_cluster.pgdata);
 	check_ok();
 
+	/*
+	 * Create logical replication slots if requested.
+	 *
+	 * Note: This must be done after doing pg_resetwal command because the
+	 * command will remove required WALs.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		start_postmaster(&new_cluster, true);
+		create_logical_replication_slots();
+		stop_postmaster(false);
+	}
+
 	if (user_opts.do_sync)
 	{
 		prep_status("Sync data directory to disk");
@@ -860,3 +874,50 @@ set_frozenxids(bool minmxid_only)
 
 	check_ok();
 }
+
+/*
+ * create_logical_replication_slots()
+ *
+ * Similar to create_new_objects() but only restores logical replication slots.
+ */
+static void
+create_logical_replication_slots(void)
+{
+	int			dbnum;
+
+	prep_status_progress("Restoring logical replication slots in the new cluster");
+
+	for (dbnum = 0; dbnum < old_cluster.dbarr.ndbs; dbnum++)
+	{
+		char		slots_file_name[MAXPGPATH],
+					log_file_name[MAXPGPATH];
+		DbInfo	   *old_db = &old_cluster.dbarr.dbs[dbnum];
+
+		pg_log(PG_STATUS, "%s", old_db->db_name);
+
+		snprintf(slots_file_name, sizeof(slots_file_name),
+				 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+		snprintf(log_file_name, sizeof(log_file_name),
+				 DB_DUMP_LOG_FILE_MASK, old_db->db_oid);
+
+		parallel_exec_prog(log_file_name,
+						   NULL,
+						   "\"%s/psql\" %s --echo-queries --set ON_ERROR_STOP=on "
+						   "--no-psqlrc --dbname %s -f \"%s/%s\"",
+						   new_cluster.bindir,
+						   cluster_conn_opts(&new_cluster),
+						   old_db->db_name,
+						   log_opts.dumpdir,
+						   slots_file_name);
+	}
+
+	/* reap all children */
+	while (reap_child(true) == true)
+		;
+
+	end_progress_output();
+	check_ok();
+
+	/* update new_cluster info now that we have objects in the databases */
+	get_db_and_rel_infos(&new_cluster);
+}
diff --git a/src/bin/pg_upgrade/pg_upgrade.h b/src/bin/pg_upgrade/pg_upgrade.h
index 3eea0139c7..5f3d7a407e 100644
--- a/src/bin/pg_upgrade/pg_upgrade.h
+++ b/src/bin/pg_upgrade/pg_upgrade.h
@@ -29,6 +29,7 @@
 /* contains both global db information and CREATE DATABASE commands */
 #define GLOBALS_DUMP_FILE	"pg_upgrade_dump_globals.sql"
 #define DB_DUMP_FILE_MASK	"pg_upgrade_dump_%u.custom"
+#define DB_DUMP_LOGICAL_SLOTS_FILE_MASK	"pg_upgrade_dump_%u_logical_slots.sql"
 
 /*
  * Base directories that include all the files generated internally, from the
@@ -304,6 +305,8 @@ typedef struct
 	transferMode transfer_mode; /* copy files or link them? */
 	int			jobs;			/* number of processes/threads to use */
 	char	   *socketdir;		/* directory to use for Unix sockets */
+	bool		include_logical_slots;	/* true -> dump and restore logical
+										 * replication slots */
 } UserOpts;
 
 typedef struct
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
new file mode 100644
index 0000000000..81917f3b14
--- /dev/null
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -0,0 +1,90 @@
+# Copyright (c) 2021-2023, PostgreSQL Global Development Group
+
+# Tests for logical replication, especially for upgrading replication slots
+
+use strict;
+use warnings;
+use PostgreSQL::Test::Cluster;
+use PostgreSQL::Test::Utils;
+use Test::More;
+
+# Can be changed to test the other modes.
+my $mode = $ENV{PG_TEST_PG_UPGRADE_MODE} || '--copy';
+
+# Initialize publisher node
+my $old_publisher = PostgreSQL::Test::Cluster->new('old_publisher');
+$old_publisher->init(allows_streaming => 'logical');
+$old_publisher->start;
+
+# Initialize subscriber node
+my $subscriber = PostgreSQL::Test::Cluster->new('subscriber');
+$subscriber->init(allows_streaming => 'logical');
+$subscriber->start;
+
+$old_publisher->safe_psql('postgres',
+	"CREATE TABLE tbl AS SELECT generate_series(1,10) AS a");
+$subscriber->safe_psql('postgres', "CREATE TABLE tbl (a int)");
+
+# Setup logical replication
+my $old_connstr = $old_publisher->connstr . ' dbname=postgres';
+$old_publisher->safe_psql('postgres',
+	"CREATE PUBLICATION pub FOR ALL TABLES");
+$subscriber->safe_psql('postgres',
+	"CREATE SUBSCRIPTION sub CONNECTION '$old_connstr' PUBLICATION pub");
+
+# Wait for initial table sync to finish
+$subscriber->wait_for_subscription_sync($old_publisher, 'sub');
+
+my $result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(10), 'check initial rows on subscriber');
+
+# Preparations for upgrading publisher
+$old_publisher->stop;
+$subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub DISABLE");
+
+my $new_publisher = PostgreSQL::Test::Cluster->new('new_publisher');
+$new_publisher->init(allows_streaming => 'logical');
+
+my $bindir = $new_publisher->config_data('--bindir');
+
+# Run pg_upgrade. pg_upgrade_output.d is removed at the end
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
+		$mode,        '--include-logical-replication-slots'
+	],
+	'run of pg_upgrade of old publisher');
+ok( !-d $new_publisher->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+# Check whether the replication slot is copied
+$new_publisher->start;
+$result =
+  $new_publisher->safe_psql('postgres',
+	"SELECT count(*) FROM pg_replication_slots");
+is($result, qq(1), 'check the replication slot is copied to new publisher');
+
+# Change connection string and enable logical replication
+my $new_connstr = $new_publisher->connstr . ' dbname=postgres';
+
+$subscriber->safe_psql('postgres',
+	"ALTER SUBSCRIPTION sub CONNECTION '$new_connstr'");
+$subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub ENABLE");
+
+# Check whether changes on the new publisher get replicated to the subscriber
+$new_publisher->safe_psql('postgres',
+	"INSERT INTO tbl VALUES (generate_series(11, 20))");
+
+$new_publisher->wait_for_catchup('sub');
+
+$result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(20), 'check changes are shipped to subscriber');
+
+done_testing();
diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list
index b4058b88c3..7e999726c2 100644
--- a/src/tools/pgindent/typedefs.list
+++ b/src/tools/pgindent/typedefs.list
@@ -1479,6 +1479,7 @@ LogicalRepBeginData
 LogicalRepCommitData
 LogicalRepCommitPreparedTxnData
 LogicalRepCtxStruct
+LogicalReplicationSlotInfo
 LogicalRepMode
 LogicalRepMsgType
 LogicalRepPartMapEntry
-- 
2.27.0

From 77c3d2d361893de857627e036d0eaaf01cfe91c1 Mon Sep 17 00:00:00 2001
From: Julien Rouhaud <julien.rouhaud@free.fr>
Date: Fri, 14 Apr 2023 13:49:09 +0800
Subject: [PATCH v1] Always persist to disk logical slots during a shutdown
 checkpoint.

It's entirely possible for a logical slot to have a confirmed_flush_lsn higher
than the last value saved on disk while not being marked as dirty.  It's
currently not a problem to lose that value during a clean shutdown / restart
cycle, but a later patch adding support for pg_upgrade of publications and
logical slots will rely on that value being properly persisted to disk.

Author: Julien Rouhaud
Reviewed-by: FIXME
Discussion: FIXME
---
 src/backend/access/transam/xlog.c |  2 +-
 src/backend/replication/slot.c    | 26 ++++++++++++++++----------
 src/include/replication/slot.h    |  2 +-
 3 files changed, 18 insertions(+), 12 deletions(-)

diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index b540ee293b..8100ca656e 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7011,7 +7011,7 @@ static void
 CheckPointGuts(XLogRecPtr checkPointRedo, int flags)
 {
 	CheckPointRelationMap();
-	CheckPointReplicationSlots();
+	CheckPointReplicationSlots(flags & CHECKPOINT_IS_SHUTDOWN);
 	CheckPointSnapBuild();
 	CheckPointLogicalRewriteHeap();
 	CheckPointReplicationOrigin();
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index 8021aaa0a8..2bbf2af770 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -109,7 +109,8 @@ static void ReplicationSlotDropPtr(ReplicationSlot *slot);
 /* internal persistency functions */
 static void RestoreSlotFromDisk(const char *name);
 static void CreateSlotOnDisk(ReplicationSlot *slot);
-static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel);
+static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+							bool is_shutdown);
 
 /*
  * Report shared-memory space needed by ReplicationSlotsShmemInit.
@@ -783,7 +784,7 @@ ReplicationSlotSave(void)
 	Assert(MyReplicationSlot != NULL);
 
 	sprintf(path, "pg_replslot/%s", NameStr(MyReplicationSlot->data.name));
-	SaveSlotToPath(MyReplicationSlot, path, ERROR);
+	SaveSlotToPath(MyReplicationSlot, path, ERROR, false);
 }
 
 /*
@@ -1565,11 +1566,10 @@ restart:
 /*
  * Flush all replication slots to disk.
  *
- * This needn't actually be part of a checkpoint, but it's a convenient
- * location.
+ * is_shutdown is true in case of a shutdown checkpoint.
  */
 void
-CheckPointReplicationSlots(void)
+CheckPointReplicationSlots(bool is_shutdown)
 {
 	int			i;
 
@@ -1594,7 +1594,7 @@ CheckPointReplicationSlots(void)
 
 		/* save the slot to disk, locking is handled in SaveSlotToPath() */
 		sprintf(path, "pg_replslot/%s", NameStr(s->data.name));
-		SaveSlotToPath(s, path, LOG);
+		SaveSlotToPath(s, path, LOG, is_shutdown);
 	}
 	LWLockRelease(ReplicationSlotAllocationLock);
 }
@@ -1700,7 +1700,7 @@ CreateSlotOnDisk(ReplicationSlot *slot)
 
 	/* Write the actual state file. */
 	slot->dirty = true;			/* signal that we really need to write */
-	SaveSlotToPath(slot, tmppath, ERROR);
+	SaveSlotToPath(slot, tmppath, ERROR, false);
 
 	/* Rename the directory into place. */
 	if (rename(tmppath, path) != 0)
@@ -1726,7 +1726,8 @@ CreateSlotOnDisk(ReplicationSlot *slot)
  * Shared functionality between saving and creating a replication slot.
  */
 static void
-SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
+SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+			   bool is_shutdown)
 {
 	char		tmppath[MAXPGPATH];
 	char		path[MAXPGPATH];
@@ -1740,8 +1741,13 @@ SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
 	slot->just_dirtied = false;
 	SpinLockRelease(&slot->mutex);
 
-	/* and don't do anything if there's nothing to write */
-	if (!was_dirty)
+	/*
+	 * and don't do anything if there's nothing to write, unless it's this is
+	 * called for a logical slot during a shutdown checkpoint, as we want to
+	 * persist the confirmed_flush_lsn in that case, even if that's the only
+	 * modification.
+	 */
+	if (!was_dirty && !is_shutdown && !SlotIsLogical(slot))
 		return;
 
 	LWLockAcquire(&slot->io_in_progress_lock, LW_EXCLUSIVE);
diff --git a/src/include/replication/slot.h b/src/include/replication/slot.h
index a8a89dc784..7ca37c9f70 100644
--- a/src/include/replication/slot.h
+++ b/src/include/replication/slot.h
@@ -241,7 +241,7 @@ extern void ReplicationSlotNameForTablesync(Oid suboid, Oid relid, char *syncslo
 extern void ReplicationSlotDropAtPubNode(WalReceiverConn *wrconn, char *slotname, bool missing_ok);
 
 extern void StartupReplicationSlots(void);
-extern void CheckPointReplicationSlots(void);
+extern void CheckPointReplicationSlots(bool is_shutdown);
 
 extern void CheckSlotRequirements(void);
 extern void CheckSlotPermissions(void);
-- 
2.37.0

From 512b817c55d8d2dec7f70684250e5c67f1562711 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Tue, 4 Apr 2023 05:49:34 +0000
Subject: [PATCH v7 1/3] pg_upgrade: Add --include-logical-replication-slots
 option

This commit introduces a new pg_upgrade option called "--include-logical-replication-slots".
This allows nodes with logical replication slots to be upgraded. The commit can
be divided into two parts: one for pg_dump and another for pg_upgrade.

For pg_dump this commit includes a new option called "--logical-replication-slots-only".
This option can be used to dump logical replication slots. When this option is
specified, the slot_name, plugin, and two_phase parameters are extracted from
pg_replication_slots. An SQL file is then generated which executes
pg_create_logical_replication_slot() with the extracted parameters.

For pg_upgrade, when '--include-logical-replication-slots' is specified, it executes
pg_dump with the new "--logical-replication-slots-only" option and restores from the
dump. Apart from restoring schema, pg_resetwal must not be called after restoring
replication slots. This is because the command discards WAL files and starts from a
new segment, even if they are required by replication slots. This leads to an ERROR:
"requested WAL segment XXX has already been removed". To avoid this, replication slots
are restored at a different time than other objects, after running pg_resetwal.

The significant advantage of this commit is that it makes it easy to continue
logical replication even after upgrading the publisher node. Previously, pg_upgrade
allowed copying publications to a new node. With this new commit, adjusting the
connection string to the new publisher will cause the apply worker on the subscriber
to connect to the new publisher automatically. This enables seamless continuation
of logical replication, even after an upgrade.

Author: Hayato Kuroda
Reviewed-by: Peter Smith, Julien Rouhaud
---
 doc/src/sgml/ref/pg_dump.sgml                 |  10 ++
 doc/src/sgml/ref/pgupgrade.sgml               |  11 ++
 src/bin/pg_dump/pg_backup.h                   |   1 +
 src/bin/pg_dump/pg_dump.c                     | 147 ++++++++++++++++++
 src/bin/pg_dump/pg_dump.h                     |  17 +-
 src/bin/pg_dump/pg_dump_sort.c                |  11 +-
 src/bin/pg_upgrade/dump.c                     |  24 +++
 src/bin/pg_upgrade/meson.build                |   1 +
 src/bin/pg_upgrade/option.c                   |   7 +
 src/bin/pg_upgrade/pg_upgrade.c               |  61 ++++++++
 src/bin/pg_upgrade/pg_upgrade.h               |   3 +
 .../t/003_logical_replication_slots.pl        |  90 +++++++++++
 src/tools/pgindent/typedefs.list              |   1 +
 13 files changed, 381 insertions(+), 3 deletions(-)
 create mode 100644 src/bin/pg_upgrade/t/003_logical_replication_slots.pl

diff --git a/doc/src/sgml/ref/pg_dump.sgml b/doc/src/sgml/ref/pg_dump.sgml
index e81e35c13b..6e07f85281 100644
--- a/doc/src/sgml/ref/pg_dump.sgml
+++ b/doc/src/sgml/ref/pg_dump.sgml
@@ -1206,6 +1206,16 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--logical-replication-slots-only</option></term>
+      <listitem>
+       <para>
+        Dump only logical replication slots; not the schema (data definitions),
+        nor data. This is mainly used when upgrading nodes.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
        <term><option>-?</option></term>
        <term><option>--help</option></term>
diff --git a/doc/src/sgml/ref/pgupgrade.sgml b/doc/src/sgml/ref/pgupgrade.sgml
index 7816b4c685..94e90ff506 100644
--- a/doc/src/sgml/ref/pgupgrade.sgml
+++ b/doc/src/sgml/ref/pgupgrade.sgml
@@ -240,6 +240,17 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--include-logical-replication-slots</option></term>
+      <listitem>
+       <para>
+        Upgrade logical replication slots. Only permanent replication slots
+        are included. Note that pg_upgrade does not check the installation of
+        plugins.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
       <term><option>-?</option></term>
       <term><option>--help</option></term>
diff --git a/src/bin/pg_dump/pg_backup.h b/src/bin/pg_dump/pg_backup.h
index aba780ef4b..0a4e931f9b 100644
--- a/src/bin/pg_dump/pg_backup.h
+++ b/src/bin/pg_dump/pg_backup.h
@@ -187,6 +187,7 @@ typedef struct _dumpOptions
 	int			use_setsessauth;
 	int			enable_row_security;
 	int			load_via_partition_root;
+	int			logical_slots_only;
 
 	/* default, if no "inclusion" switches appear, is to dump everything */
 	bool		include_everything;
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index 73a6c964ba..6cff4009b3 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -328,6 +328,9 @@ static void setupDumpWorker(Archive *AH);
 static TableInfo *getRootTableInfo(const TableInfo *tbinfo);
 static bool forcePartitionRootLoad(const TableInfo *tbinfo);
 
+static void getLogicalReplicationSlots(Archive *fout);
+static void dumpLogicalReplicationSlot(Archive *fout,
+									   const LogicalReplicationSlotInfo *slotinfo);
 
 int
 main(int argc, char **argv)
@@ -431,6 +434,7 @@ main(int argc, char **argv)
 		{"table-and-children", required_argument, NULL, 12},
 		{"exclude-table-and-children", required_argument, NULL, 13},
 		{"exclude-table-data-and-children", required_argument, NULL, 14},
+		{"logical-replication-slots-only", no_argument, NULL, 15},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -657,6 +661,10 @@ main(int argc, char **argv)
 										  optarg);
 				break;
 
+			case 15:			/* dump only replication slot(s) */
+				dopt.logical_slots_only = true;
+				break;
+
 			default:
 				/* getopt_long already emitted a complaint */
 				pg_log_error_hint("Try \"%s --help\" for more information.", progname);
@@ -714,6 +722,11 @@ main(int argc, char **argv)
 	if (dopt.do_nothing && dopt.dump_inserts == 0)
 		pg_fatal("option --on-conflict-do-nothing requires option --inserts, --rows-per-insert, or --column-inserts");
 
+	if (dopt.logical_slots_only && dopt.dataOnly)
+		pg_fatal("options --logical-replication-slots-only and -a/--data-only cannot be used together");
+	if (dopt.logical_slots_only && dopt.schemaOnly)
+		pg_fatal("options --logical-replication-slots-only and -s/--schema-only cannot be used together");
+
 	/* Identify archive format to emit */
 	archiveFormat = parseArchiveFormat(format, &archiveMode);
 
@@ -876,6 +889,16 @@ main(int argc, char **argv)
 			pg_fatal("no matching extensions were found");
 	}
 
+	/*
+	 * If dump logical-replication-slots-only was requested, dump only them
+	 * and skip everything else.
+	 */
+	if (dopt.logical_slots_only)
+	{
+		getLogicalReplicationSlots(fout);
+		goto dump;
+	}
+
 	/*
 	 * Dumping LOs is the default for dumps where an inclusion switch is not
 	 * used (an "include everything" dump).  -B can be used to exclude LOs
@@ -936,6 +959,8 @@ main(int argc, char **argv)
 	if (!dopt.no_security_labels)
 		collectSecLabels(fout);
 
+dump:
+
 	/* Lastly, create dummy objects to represent the section boundaries */
 	boundaryObjs = createBoundaryObjects();
 
@@ -1109,6 +1134,8 @@ help(const char *progname)
 			 "                               servers matching PATTERN\n"));
 	printf(_("  --inserts                    dump data as INSERT commands, rather than COPY\n"));
 	printf(_("  --load-via-partition-root    load partitions via the root table\n"));
+	printf(_("  --logical-replication-slots-only\n"
+			 "                               dump only logical replication slots, no schema or data\n"));
 	printf(_("  --no-comments                do not dump comments\n"));
 	printf(_("  --no-publications            do not dump publications\n"));
 	printf(_("  --no-security-labels         do not dump security label assignments\n"));
@@ -10252,6 +10279,10 @@ dumpDumpableObject(Archive *fout, DumpableObject *dobj)
 		case DO_SUBSCRIPTION:
 			dumpSubscription(fout, (const SubscriptionInfo *) dobj);
 			break;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			dumpLogicalReplicationSlot(fout,
+									   (const LogicalReplicationSlotInfo *) dobj);
+			break;
 		case DO_PRE_DATA_BOUNDARY:
 		case DO_POST_DATA_BOUNDARY:
 			/* never dumped, nothing to do */
@@ -18227,6 +18258,7 @@ addBoundaryDependencies(DumpableObject **dobjs, int numObjs,
 			case DO_PUBLICATION_REL:
 			case DO_PUBLICATION_TABLE_IN_SCHEMA:
 			case DO_SUBSCRIPTION:
+			case DO_LOGICAL_REPLICATION_SLOT:
 				/* Post-data objects: must come after the post-data boundary */
 				addObjectDependency(dobj, postDataBound->dumpId);
 				break;
@@ -18488,3 +18520,118 @@ appendReloptionsArrayAH(PQExpBuffer buffer, const char *reloptions,
 	if (!res)
 		pg_log_warning("could not parse %s array", "reloptions");
 }
+
+/*
+ * getLogicalReplicationSlots
+ *	  get information about replication slots
+ */
+static void
+getLogicalReplicationSlots(Archive *fout)
+{
+	PGresult   *res;
+	LogicalReplicationSlotInfo *slotinfo;
+	PQExpBuffer query;
+	DumpOptions *dopt = fout->dopt;
+
+	int			i_slotname;
+	int			i_plugin;
+	int			i_twophase;
+	int			i,
+				ntups;
+
+	/* Check whether we should dump or not */
+	if (fout->remoteVersion < 160000 || !dopt->logical_slots_only)
+		return;
+
+	query = createPQExpBuffer();
+
+	resetPQExpBuffer(query);
+
+	/*
+	 * Get replication slots.
+	 *
+	 * XXX: Which information must be extracted from old node? Currently three
+	 * attributes are extracted because they are used by
+	 * pg_create_logical_replication_slot().
+	 *
+	 * XXX: Do we have to support physical slots?
+	 */
+	appendPQExpBufferStr(query,
+						 "SELECT slot_name, plugin, two_phase "
+						 "FROM pg_catalog.pg_replication_slots "
+						 "WHERE database = current_database() AND temporary = false "
+						 "AND wal_status IN ('reserved', 'extended');");
+
+	res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
+
+	ntups = PQntuples(res);
+
+	i_slotname = PQfnumber(res, "slot_name");
+	i_plugin = PQfnumber(res, "plugin");
+	i_twophase = PQfnumber(res, "two_phase");
+
+	slotinfo = pg_malloc(ntups * sizeof(LogicalReplicationSlotInfo));
+
+	for (i = 0; i < ntups; i++)
+	{
+		slotinfo[i].dobj.objType = DO_LOGICAL_REPLICATION_SLOT;
+
+		slotinfo[i].dobj.catId.tableoid = InvalidOid;
+		slotinfo[i].dobj.catId.oid = InvalidOid;
+		AssignDumpId(&slotinfo[i].dobj);
+
+		slotinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_slotname));
+
+		slotinfo[i].plugin = pg_strdup(PQgetvalue(res, i, i_plugin));
+		slotinfo[i].twophase = pg_strdup(PQgetvalue(res, i, i_twophase));
+
+		/*
+		 * Note: Currently we do not have any options to include/exclude slots
+		 * in dumping, so all the slots must be selected.
+		 */
+		slotinfo[i].dobj.dump = DUMP_COMPONENT_ALL;
+	}
+	PQclear(res);
+
+	destroyPQExpBuffer(query);
+}
+
+/*
+ * dumpLogicalReplicationSlot
+ *	  dump creation functions for the given logical replication slots
+ */
+static void
+dumpLogicalReplicationSlot(Archive *fout,
+						   const LogicalReplicationSlotInfo *slotinfo)
+{
+	DumpOptions *dopt = fout->dopt;
+
+	if (!dopt->logical_slots_only)
+		return;
+
+	if (slotinfo->dobj.dump & DUMP_COMPONENT_DEFINITION)
+	{
+		PQExpBuffer query = createPQExpBuffer();
+		char	   *slotname = pg_strdup(slotinfo->dobj.name);
+
+		/*
+		 * XXX: For simplification, pg_create_logical_replication_slot() is
+		 * used. Is it sufficient?
+		 */
+		appendPQExpBuffer(query, "SELECT pg_catalog.pg_create_logical_replication_slot('%s', ",
+						  slotname);
+		appendStringLiteralAH(query, slotinfo->plugin, fout);
+		appendPQExpBuffer(query, ", ");
+		appendStringLiteralAH(query, slotinfo->twophase, fout);
+		appendPQExpBuffer(query, ");");
+
+		ArchiveEntry(fout, slotinfo->dobj.catId, slotinfo->dobj.dumpId,
+					 ARCHIVE_OPTS(.tag = slotname,
+								  .description = "REPLICATION SLOT",
+								  .section = SECTION_POST_DATA,
+								  .createStmt = query->data));
+
+		pfree(slotname);
+		destroyPQExpBuffer(query);
+	}
+}
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index ed6ce41ad7..8028ccf6ff 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -82,7 +82,8 @@ typedef enum
 	DO_PUBLICATION,
 	DO_PUBLICATION_REL,
 	DO_PUBLICATION_TABLE_IN_SCHEMA,
-	DO_SUBSCRIPTION
+	DO_SUBSCRIPTION,
+	DO_LOGICAL_REPLICATION_SLOT
 } DumpableObjectType;
 
 /*
@@ -666,6 +667,20 @@ typedef struct _SubscriptionInfo
 	char	   *subpasswordrequired;
 } SubscriptionInfo;
 
+/*
+ * The LogicalReplicationSlotInfo struct is used to represent replication
+ * slots.
+ *
+ * XXX: add more attributes if needed
+ */
+typedef struct _LogicalReplicationSlotInfo
+{
+	DumpableObject dobj;
+	char	   *plugin;
+	char	   *slottype;
+	char	   *twophase;
+} LogicalReplicationSlotInfo;
+
 /*
  *	common utility functions
  */
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index 8266c117a3..7c37873743 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -93,6 +93,7 @@ enum dbObjectTypePriorities
 	PRIO_PUBLICATION_REL,
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,
 	PRIO_SUBSCRIPTION,
+	PRIO_LOGICAL_REPLICATION_SLOT,
 	PRIO_DEFAULT_ACL,			/* done in ACL pass */
 	PRIO_EVENT_TRIGGER,			/* must be next to last! */
 	PRIO_REFRESH_MATVIEW		/* must be last! */
@@ -146,10 +147,11 @@ static const int dbObjectTypePriority[] =
 	PRIO_PUBLICATION,			/* DO_PUBLICATION */
 	PRIO_PUBLICATION_REL,		/* DO_PUBLICATION_REL */
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,	/* DO_PUBLICATION_TABLE_IN_SCHEMA */
-	PRIO_SUBSCRIPTION			/* DO_SUBSCRIPTION */
+	PRIO_SUBSCRIPTION,			/* DO_SUBSCRIPTION */
+	PRIO_LOGICAL_REPLICATION_SLOT	/* DO_LOGICAL_REPLICATION_SLOT */
 };
 
-StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_SUBSCRIPTION + 1),
+StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_LOGICAL_REPLICATION_SLOT + 1),
 				 "array length mismatch");
 
 static DumpId preDataBoundId;
@@ -1498,6 +1500,11 @@ describeDumpableObject(DumpableObject *obj, char *buf, int bufsize)
 					 "SUBSCRIPTION (ID %d OID %u)",
 					 obj->dumpId, obj->catId.oid);
 			return;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			snprintf(buf, bufsize,
+					 "LOGICAL REPLICATION SLOT (ID %d NAME %s)",
+					 obj->dumpId, obj->name);
+			return;
 		case DO_PRE_DATA_BOUNDARY:
 			snprintf(buf, bufsize,
 					 "PRE-DATA BOUNDARY  (ID %d)",
diff --git a/src/bin/pg_upgrade/dump.c b/src/bin/pg_upgrade/dump.c
index 6c8c82dca8..e6b90864f5 100644
--- a/src/bin/pg_upgrade/dump.c
+++ b/src/bin/pg_upgrade/dump.c
@@ -59,6 +59,30 @@ generate_old_dump(void)
 						   log_opts.dumpdir,
 						   sql_file_name, escaped_connstr.data);
 
+		/*
+		 * Dump logical replication slots if needed.
+		 *
+		 * XXX We cannot dump replication slots at the same time as the schema
+		 * dump because we need to separate the timing of restoring
+		 * replication slots and other objects. Replication slots, in
+		 * particular, should not be restored before executing the pg_resetwal
+		 * command because it will remove WALs that are required by the slots.
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			char		slots_file_name[MAXPGPATH];
+
+			snprintf(slots_file_name, sizeof(slots_file_name),
+					 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+			parallel_exec_prog(log_file_name, NULL,
+							   "\"%s/pg_dump\" %s --logical-replication-slots-only "
+							   "--quote-all-identifiers --binary-upgrade %s "
+							   "--file=\"%s/%s\" %s",
+							   new_cluster.bindir, cluster_conn_opts(&old_cluster),
+							   log_opts.verbose ? "--verbose" : "",
+							   log_opts.dumpdir,
+							   slots_file_name, escaped_connstr.data);
+		}
 		termPQExpBuffer(&escaped_connstr);
 	}
 
diff --git a/src/bin/pg_upgrade/meson.build b/src/bin/pg_upgrade/meson.build
index 12a97f84e2..228f29b688 100644
--- a/src/bin/pg_upgrade/meson.build
+++ b/src/bin/pg_upgrade/meson.build
@@ -42,6 +42,7 @@ tests += {
     'tests': [
       't/001_basic.pl',
       't/002_pg_upgrade.pl',
+      't/003_logical_replication_slots.pl',
     ],
     'test_kwargs': {'priority': 40}, # pg_upgrade tests are slow
   },
diff --git a/src/bin/pg_upgrade/option.c b/src/bin/pg_upgrade/option.c
index 8869b6b60d..f7b5ec9879 100644
--- a/src/bin/pg_upgrade/option.c
+++ b/src/bin/pg_upgrade/option.c
@@ -57,6 +57,7 @@ parseCommandLine(int argc, char *argv[])
 		{"verbose", no_argument, NULL, 'v'},
 		{"clone", no_argument, NULL, 1},
 		{"copy", no_argument, NULL, 2},
+		{"include-logical-replication-slots", no_argument, NULL, 3},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -199,6 +200,10 @@ parseCommandLine(int argc, char *argv[])
 				user_opts.transfer_mode = TRANSFER_MODE_COPY;
 				break;
 
+			case 3:
+				user_opts.include_logical_slots = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						os_info.progname);
@@ -289,6 +294,8 @@ usage(void)
 	printf(_("  -V, --version                 display version information, then exit\n"));
 	printf(_("  --clone                       clone instead of copying files to new cluster\n"));
 	printf(_("  --copy                        copy files to new cluster (default)\n"));
+	printf(_("  --include-logical-replication-slots\n"
+			 "                                upgrade logical replication slots\n"));
 	printf(_("  -?, --help                    show this help, then exit\n"));
 	printf(_("\n"
 			 "Before running pg_upgrade you must:\n"
diff --git a/src/bin/pg_upgrade/pg_upgrade.c b/src/bin/pg_upgrade/pg_upgrade.c
index 75bab0a04c..1241060f4e 100644
--- a/src/bin/pg_upgrade/pg_upgrade.c
+++ b/src/bin/pg_upgrade/pg_upgrade.c
@@ -59,6 +59,7 @@ static void copy_xact_xlog_xid(void);
 static void set_frozenxids(bool minmxid_only);
 static void make_outputdirs(char *pgdata);
 static void setup(char *argv0, bool *live_check);
+static void create_logical_replication_slots(void);
 
 ClusterInfo old_cluster,
 			new_cluster;
@@ -188,6 +189,19 @@ main(int argc, char **argv)
 			  new_cluster.pgdata);
 	check_ok();
 
+	/*
+	 * Create logical replication slots if requested.
+	 *
+	 * Note: This must be done after doing pg_resetwal command because the
+	 * command will remove required WALs.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		start_postmaster(&new_cluster, true);
+		create_logical_replication_slots();
+		stop_postmaster(false);
+	}
+
 	if (user_opts.do_sync)
 	{
 		prep_status("Sync data directory to disk");
@@ -860,3 +874,50 @@ set_frozenxids(bool minmxid_only)
 
 	check_ok();
 }
+
+/*
+ * create_logical_replication_slots()
+ *
+ * Similar to create_new_objects() but only restores logical replication slots.
+ */
+static void
+create_logical_replication_slots(void)
+{
+	int			dbnum;
+
+	prep_status_progress("Restoring logical replication slots in the new cluster");
+
+	for (dbnum = 0; dbnum < old_cluster.dbarr.ndbs; dbnum++)
+	{
+		char		slots_file_name[MAXPGPATH],
+					log_file_name[MAXPGPATH];
+		DbInfo	   *old_db = &old_cluster.dbarr.dbs[dbnum];
+
+		pg_log(PG_STATUS, "%s", old_db->db_name);
+
+		snprintf(slots_file_name, sizeof(slots_file_name),
+				 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+		snprintf(log_file_name, sizeof(log_file_name),
+				 DB_DUMP_LOG_FILE_MASK, old_db->db_oid);
+
+		parallel_exec_prog(log_file_name,
+						   NULL,
+						   "\"%s/psql\" %s --echo-queries --set ON_ERROR_STOP=on "
+						   "--no-psqlrc --dbname %s -f \"%s/%s\"",
+						   new_cluster.bindir,
+						   cluster_conn_opts(&new_cluster),
+						   old_db->db_name,
+						   log_opts.dumpdir,
+						   slots_file_name);
+	}
+
+	/* reap all children */
+	while (reap_child(true) == true)
+		;
+
+	end_progress_output();
+	check_ok();
+
+	/* update new_cluster info now that we have objects in the databases */
+	get_db_and_rel_infos(&new_cluster);
+}
diff --git a/src/bin/pg_upgrade/pg_upgrade.h b/src/bin/pg_upgrade/pg_upgrade.h
index 3eea0139c7..5f3d7a407e 100644
--- a/src/bin/pg_upgrade/pg_upgrade.h
+++ b/src/bin/pg_upgrade/pg_upgrade.h
@@ -29,6 +29,7 @@
 /* contains both global db information and CREATE DATABASE commands */
 #define GLOBALS_DUMP_FILE	"pg_upgrade_dump_globals.sql"
 #define DB_DUMP_FILE_MASK	"pg_upgrade_dump_%u.custom"
+#define DB_DUMP_LOGICAL_SLOTS_FILE_MASK	"pg_upgrade_dump_%u_logical_slots.sql"
 
 /*
  * Base directories that include all the files generated internally, from the
@@ -304,6 +305,8 @@ typedef struct
 	transferMode transfer_mode; /* copy files or link them? */
 	int			jobs;			/* number of processes/threads to use */
 	char	   *socketdir;		/* directory to use for Unix sockets */
+	bool		include_logical_slots;	/* true -> dump and restore logical
+										 * replication slots */
 } UserOpts;
 
 typedef struct
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
new file mode 100644
index 0000000000..81917f3b14
--- /dev/null
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -0,0 +1,90 @@
+# Copyright (c) 2021-2023, PostgreSQL Global Development Group
+
+# Tests for logical replication, especially for upgrading replication slots
+
+use strict;
+use warnings;
+use PostgreSQL::Test::Cluster;
+use PostgreSQL::Test::Utils;
+use Test::More;
+
+# Can be changed to test the other modes.
+my $mode = $ENV{PG_TEST_PG_UPGRADE_MODE} || '--copy';
+
+# Initialize publisher node
+my $old_publisher = PostgreSQL::Test::Cluster->new('old_publisher');
+$old_publisher->init(allows_streaming => 'logical');
+$old_publisher->start;
+
+# Initialize subscriber node
+my $subscriber = PostgreSQL::Test::Cluster->new('subscriber');
+$subscriber->init(allows_streaming => 'logical');
+$subscriber->start;
+
+$old_publisher->safe_psql('postgres',
+	"CREATE TABLE tbl AS SELECT generate_series(1,10) AS a");
+$subscriber->safe_psql('postgres', "CREATE TABLE tbl (a int)");
+
+# Setup logical replication
+my $old_connstr = $old_publisher->connstr . ' dbname=postgres';
+$old_publisher->safe_psql('postgres',
+	"CREATE PUBLICATION pub FOR ALL TABLES");
+$subscriber->safe_psql('postgres',
+	"CREATE SUBSCRIPTION sub CONNECTION '$old_connstr' PUBLICATION pub");
+
+# Wait for initial table sync to finish
+$subscriber->wait_for_subscription_sync($old_publisher, 'sub');
+
+my $result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(10), 'check initial rows on subscriber');
+
+# Preparations for upgrading publisher
+$old_publisher->stop;
+$subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub DISABLE");
+
+my $new_publisher = PostgreSQL::Test::Cluster->new('new_publisher');
+$new_publisher->init(allows_streaming => 'logical');
+
+my $bindir = $new_publisher->config_data('--bindir');
+
+# Run pg_upgrade. pg_upgrade_output.d is removed at the end
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
+		$mode,        '--include-logical-replication-slots'
+	],
+	'run of pg_upgrade of old publisher');
+ok( !-d $new_publisher->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+# Check whether the replication slot is copied
+$new_publisher->start;
+$result =
+  $new_publisher->safe_psql('postgres',
+	"SELECT count(*) FROM pg_replication_slots");
+is($result, qq(1), 'check the replication slot is copied to new publisher');
+
+# Change connection string and enable logical replication
+my $new_connstr = $new_publisher->connstr . ' dbname=postgres';
+
+$subscriber->safe_psql('postgres',
+	"ALTER SUBSCRIPTION sub CONNECTION '$new_connstr'");
+$subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub ENABLE");
+
+# Check whether changes on the new publisher get replicated to the subscriber
+$new_publisher->safe_psql('postgres',
+	"INSERT INTO tbl VALUES (generate_series(11, 20))");
+
+$new_publisher->wait_for_catchup('sub');
+
+$result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(20), 'check changes are shipped to subscriber');
+
+done_testing();
diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list
index b4058b88c3..7e999726c2 100644
--- a/src/tools/pgindent/typedefs.list
+++ b/src/tools/pgindent/typedefs.list
@@ -1479,6 +1479,7 @@ LogicalRepBeginData
 LogicalRepCommitData
 LogicalRepCommitPreparedTxnData
 LogicalRepCtxStruct
+LogicalReplicationSlotInfo
 LogicalRepMode
 LogicalRepMsgType
 LogicalRepPartMapEntry
-- 
2.27.0

From baf1b71bd2c51b130fba324ce7c8d1dddd7a1a5a Mon Sep 17 00:00:00 2001
From: Julien Rouhaud <julien.rouhaud@free.fr>
Date: Fri, 14 Apr 2023 13:49:09 +0800
Subject: [PATCH v7 2/3] Always persist to disk logical slots during a shutdown
 checkpoint.

It's entirely possible for a logical slot to have a confirmed_flush_lsn higher
than the last value saved on disk while not being marked as dirty.  It's
currently not a problem to lose that value during a clean shutdown / restart
cycle, but a later patch adding support for pg_upgrade of publications and
logical slots will rely on that value being properly persisted to disk.

Author: Julien Rouhaud
Reviewed-by: FIXME
Discussion: FIXME
---
 src/backend/access/transam/xlog.c |  2 +-
 src/backend/replication/slot.c    | 26 ++++++++++++++++----------
 src/include/replication/slot.h    |  2 +-
 3 files changed, 18 insertions(+), 12 deletions(-)

diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index b540ee293b..8100ca656e 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7011,7 +7011,7 @@ static void
 CheckPointGuts(XLogRecPtr checkPointRedo, int flags)
 {
 	CheckPointRelationMap();
-	CheckPointReplicationSlots();
+	CheckPointReplicationSlots(flags & CHECKPOINT_IS_SHUTDOWN);
 	CheckPointSnapBuild();
 	CheckPointLogicalRewriteHeap();
 	CheckPointReplicationOrigin();
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index 8021aaa0a8..aeea6ffd1f 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -109,7 +109,8 @@ static void ReplicationSlotDropPtr(ReplicationSlot *slot);
 /* internal persistency functions */
 static void RestoreSlotFromDisk(const char *name);
 static void CreateSlotOnDisk(ReplicationSlot *slot);
-static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel);
+static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+						   bool is_shutdown);
 
 /*
  * Report shared-memory space needed by ReplicationSlotsShmemInit.
@@ -783,7 +784,7 @@ ReplicationSlotSave(void)
 	Assert(MyReplicationSlot != NULL);
 
 	sprintf(path, "pg_replslot/%s", NameStr(MyReplicationSlot->data.name));
-	SaveSlotToPath(MyReplicationSlot, path, ERROR);
+	SaveSlotToPath(MyReplicationSlot, path, ERROR, false);
 }
 
 /*
@@ -1565,11 +1566,10 @@ restart:
 /*
  * Flush all replication slots to disk.
  *
- * This needn't actually be part of a checkpoint, but it's a convenient
- * location.
+ * is_shutdown is true in case of a shutdown checkpoint.
  */
 void
-CheckPointReplicationSlots(void)
+CheckPointReplicationSlots(bool is_shutdown)
 {
 	int			i;
 
@@ -1594,7 +1594,7 @@ CheckPointReplicationSlots(void)
 
 		/* save the slot to disk, locking is handled in SaveSlotToPath() */
 		sprintf(path, "pg_replslot/%s", NameStr(s->data.name));
-		SaveSlotToPath(s, path, LOG);
+		SaveSlotToPath(s, path, LOG, is_shutdown);
 	}
 	LWLockRelease(ReplicationSlotAllocationLock);
 }
@@ -1700,7 +1700,7 @@ CreateSlotOnDisk(ReplicationSlot *slot)
 
 	/* Write the actual state file. */
 	slot->dirty = true;			/* signal that we really need to write */
-	SaveSlotToPath(slot, tmppath, ERROR);
+	SaveSlotToPath(slot, tmppath, ERROR, false);
 
 	/* Rename the directory into place. */
 	if (rename(tmppath, path) != 0)
@@ -1726,7 +1726,8 @@ CreateSlotOnDisk(ReplicationSlot *slot)
  * Shared functionality between saving and creating a replication slot.
  */
 static void
-SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
+SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+			   bool is_shutdown)
 {
 	char		tmppath[MAXPGPATH];
 	char		path[MAXPGPATH];
@@ -1740,8 +1741,13 @@ SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
 	slot->just_dirtied = false;
 	SpinLockRelease(&slot->mutex);
 
-	/* and don't do anything if there's nothing to write */
-	if (!was_dirty)
+	/*
+	 * and don't do anything if there's nothing to write, unless it's this is
+	 * called for a logical slot during a shutdown checkpoint, as we want to
+	 * persist the confirmed_flush_lsn in that case, even if that's the only
+	 * modification.
+	 */
+	if (!was_dirty && !is_shutdown && !SlotIsLogical(slot))
 		return;
 
 	LWLockAcquire(&slot->io_in_progress_lock, LW_EXCLUSIVE);
diff --git a/src/include/replication/slot.h b/src/include/replication/slot.h
index a8a89dc784..7ca37c9f70 100644
--- a/src/include/replication/slot.h
+++ b/src/include/replication/slot.h
@@ -241,7 +241,7 @@ extern void ReplicationSlotNameForTablesync(Oid suboid, Oid relid, char *syncslo
 extern void ReplicationSlotDropAtPubNode(WalReceiverConn *wrconn, char *slotname, bool missing_ok);
 
 extern void StartupReplicationSlots(void);
-extern void CheckPointReplicationSlots(void);
+extern void CheckPointReplicationSlots(bool is_shutdown);
 
 extern void CheckSlotRequirements(void);
 extern void CheckSlotPermissions(void);
-- 
2.27.0

From ddd3739eff35141bf00e64a9c5e5500f3c97383b Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Fri, 14 Apr 2023 08:59:03 +0000
Subject: [PATCH v7 3/3] pg_upgrade: Add check function for
 --include-logical-replication-slots option

---
 src/bin/pg_upgrade/check.c                    | 76 +++++++++++++++++++
 .../t/003_logical_replication_slots.pl        | 58 +++++++++++---
 2 files changed, 123 insertions(+), 11 deletions(-)

diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index fea159689e..8dda227bf2 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -9,7 +9,10 @@
 
 #include "postgres_fe.h"
 
+#include "access/xlogrecord.h"
+#include "access/xlog_internal.h"
 #include "catalog/pg_authid_d.h"
+#include "catalog/pg_control.h"
 #include "catalog/pg_collation.h"
 #include "fe_utils/string_utils.h"
 #include "mb/pg_wchar.h"
@@ -30,6 +33,7 @@ static void check_for_jsonb_9_4_usage(ClusterInfo *cluster);
 static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
+static void check_for_logical_replication_slots(ClusterInfo *cluster);
 
 
 /*
@@ -103,6 +107,8 @@ check_and_dump_old_cluster(bool live_check)
 	check_for_composite_data_type_usage(&old_cluster);
 	check_for_reg_data_type_usage(&old_cluster);
 	check_for_isn_and_int8_passing_mismatch(&old_cluster);
+	if (user_opts.include_logical_slots)
+		check_for_logical_replication_slots(&old_cluster);
 
 	/*
 	 * PG 16 increased the size of the 'aclitem' type, which breaks the on-disk
@@ -1402,3 +1408,73 @@ check_for_user_defined_encoding_conversions(ClusterInfo *cluster)
 	else
 		check_ok();
 }
+
+/*
+ * Verify that all logical replication slots consumed all WALs, except a
+ * CHECKPOINT_SHUTDOWN record.
+ */
+static void
+check_for_logical_replication_slots(ClusterInfo *cluster)
+{
+	int			i,
+				ntups,
+				i_slotname;
+	bool		is_error = false;
+	PGresult   *res;
+	DbInfo	   *active_db = &cluster->dbarr.dbs[0];
+	PGconn	   *conn = connectToServer(cluster, active_db->db_name);
+
+	Assert(user_opts.include_logical_slots);
+
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(cluster->major_version < 1600))
+		return;
+
+	prep_status("Checking for logical replication slots");
+
+
+	/*
+	 * Check that all logical replication slots have reached the current WAL
+	 * position, except for the CHECKPOINT_SHUTDOWN record. Even if all WALs
+	 * are consumed before shutting down the node, the checkpointer generates
+	 * a CHECKPOINT_SHUTDOWN record at shutdown, which cannot be consumed by
+	 * any slots. Therefore, we must allow for a difference between
+	 * pg_current_wal_insert_lsn() and confirmed_flush_lsn.
+	 */
+#define SHUTDOWN_RECORD_SIZE  (SizeOfXLogRecord + \
+							   SizeOfXLogRecordDataHeaderShort + \
+							   sizeof(CheckPoint))
+
+	res = executeQueryOrDie(conn,
+							"SELECT slot_name FROM pg_catalog.pg_replication_slots "
+							"WHERE (pg_catalog.pg_current_wal_insert_lsn() - confirmed_flush_lsn) > %d "
+							"AND temporary = false AND wal_status IN ('reserved', 'extended');",
+							(int) (SizeOfXLogShortPHD + SHUTDOWN_RECORD_SIZE));
+
+#undef SHUTDOWN_RECORD_SIZE
+
+	ntups = PQntuples(res);
+	i_slotname = PQfnumber(res, "slot_name");
+
+	for (i = 0; i < ntups; i++)
+	{
+		char	   *slotname;
+
+		is_error = true;
+
+		slotname = PQgetvalue(res, i, i_slotname);
+
+		pg_log(PG_WARNING,
+			   "\nWARNING: logical replication slot \"%s\" has not consumed WALs yet.",
+			   slotname);
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	if (is_error)
+		pg_fatal("--include-logical-replication-slot requires that all "
+				 "logical replication slots consumed all the WALs");
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
index 81917f3b14..527d448a00 100644
--- a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -4,6 +4,9 @@
 
 use strict;
 use warnings;
+
+use File::Path qw(rmtree);
+
 use PostgreSQL::Test::Cluster;
 use PostgreSQL::Test::Utils;
 use Test::More;
@@ -11,7 +14,7 @@ use Test::More;
 # Can be changed to test the other modes.
 my $mode = $ENV{PG_TEST_PG_UPGRADE_MODE} || '--copy';
 
-# Initialize publisher node
+# Initialize old publisher node
 my $old_publisher = PostgreSQL::Test::Cluster->new('old_publisher');
 $old_publisher->init(allows_streaming => 'logical');
 $old_publisher->start;
@@ -21,10 +24,48 @@ my $subscriber = PostgreSQL::Test::Cluster->new('subscriber');
 $subscriber->init(allows_streaming => 'logical');
 $subscriber->start;
 
+# Initialize new publisher node
+my $new_publisher = PostgreSQL::Test::Cluster->new('new_publisher');
+$new_publisher->init(allows_streaming => 'logical');
+my $bindir = $new_publisher->config_data('--bindir');
+
+# Schema setup
 $old_publisher->safe_psql('postgres',
 	"CREATE TABLE tbl AS SELECT generate_series(1,10) AS a");
 $subscriber->safe_psql('postgres', "CREATE TABLE tbl (a int)");
 
+# Create a dummy slot on old publisher to fail the test
+$old_publisher->safe_psql('postgres',
+	"SELECT pg_create_logical_replication_slot('dropme_slot', 'pgoutput')");
+$old_publisher->safe_psql('postgres',
+	"INSERT INTO tbl VALUES (generate_series(11, 20))");
+$old_publisher->stop;
+
+# Cause a failure at the start of pg_upgrade because dropme_slot is not
+# advanced till the INSERT statement
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
+		$mode,        '--include-logical-replication-slots',
+		'--check',
+	],
+	'run of pg_upgrade --check for new instance with inactive slot');
+ok( -d $new_publisher->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_publisher->data_dir . "/pg_upgrade_output.d");
+$old_publisher->start;
+$old_publisher->safe_psql('postgres',
+	"SELECT pg_drop_replication_slot('dropme_slot')");
+
 # Setup logical replication
 my $old_connstr = $old_publisher->connstr . ' dbname=postgres';
 $old_publisher->safe_psql('postgres',
@@ -36,18 +77,13 @@ $subscriber->safe_psql('postgres',
 $subscriber->wait_for_subscription_sync($old_publisher, 'sub');
 
 my $result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
-is($result, qq(10), 'check initial rows on subscriber');
+is($result, qq(20), 'check initial rows on subscriber');
 
 # Preparations for upgrading publisher
 $old_publisher->stop;
 $subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub DISABLE");
 
-my $new_publisher = PostgreSQL::Test::Cluster->new('new_publisher');
-$new_publisher->init(allows_streaming => 'logical');
-
-my $bindir = $new_publisher->config_data('--bindir');
-
-# Run pg_upgrade. pg_upgrade_output.d is removed at the end
+# Run pg_upgrade again. pg_upgrade_output.d is removed at the end
 command_ok(
 	[
 		'pg_upgrade', '--no-sync',
@@ -64,7 +100,7 @@ command_ok(
 ok( !-d $new_publisher->data_dir . "/pg_upgrade_output.d",
 	"pg_upgrade_output.d/ removed after pg_upgrade success");
 
-# Check whether the replication slot is copied
+# Check whether the replication slot is copied to new publisher
 $new_publisher->start;
 $result =
   $new_publisher->safe_psql('postgres',
@@ -80,11 +116,11 @@ $subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub ENABLE");
 
 # Check whether changes on the new publisher get replicated to the subscriber
 $new_publisher->safe_psql('postgres',
-	"INSERT INTO tbl VALUES (generate_series(11, 20))");
+	"INSERT INTO tbl VALUES (generate_series(21, 30))");
 
 $new_publisher->wait_for_catchup('sub');
 
 $result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
-is($result, qq(20), 'check changes are shipped to subscriber');
+is($result, qq(30), 'check changes are shipped to subscriber');
 
 done_testing();
-- 
2.27.0

From 94a5cffd6475ecd97f7a4699f993ecc61e0f5ead Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Tue, 4 Apr 2023 05:49:34 +0000
Subject: [PATCH v8 1/3] pg_upgrade: Add --include-logical-replication-slots
 option

This commit introduces a new pg_upgrade option called "--include-logical-replication-slots".
This allows nodes with logical replication slots to be upgraded. The commit can
be divided into two parts: one for pg_dump and another for pg_upgrade.

For pg_dump this commit includes a new option called "--logical-replication-slots-only".
This option can be used to dump logical replication slots. When this option is
specified, the slot_name, plugin, and two_phase parameters are extracted from
pg_replication_slots. An SQL file is then generated which executes
pg_create_logical_replication_slot() with the extracted parameters.

For pg_upgrade, when '--include-logical-replication-slots' is specified, it executes
pg_dump with the new "--logical-replication-slots-only" option and restores from the
dump. Apart from restoring schema, pg_resetwal must not be called after restoring
replication slots. This is because the command discards WAL files and starts from a
new segment, even if they are required by replication slots. This leads to an ERROR:
"requested WAL segment XXX has already been removed". To avoid this, replication slots
are restored at a different time than other objects, after running pg_resetwal.

The significant advantage of this commit is that it makes it easy to continue
logical replication even after upgrading the publisher node. Previously, pg_upgrade
allowed copying publications to a new node. With this new commit, adjusting the
connection string to the new publisher will cause the apply worker on the subscriber
to connect to the new publisher automatically. This enables seamless continuation
of logical replication, even after an upgrade.

Author: Hayato Kuroda
Reviewed-by: Peter Smith, Julien Rouhaud, Vignesh C
---
 doc/src/sgml/ref/pg_dump.sgml                 |  10 ++
 doc/src/sgml/ref/pgupgrade.sgml               |  11 ++
 src/bin/pg_dump/pg_backup.h                   |   1 +
 src/bin/pg_dump/pg_dump.c                     | 147 +++++++++++++++++
 src/bin/pg_dump/pg_dump.h                     |  17 +-
 src/bin/pg_dump/pg_dump_sort.c                |  11 +-
 src/bin/pg_upgrade/check.c                    |  39 +++++
 src/bin/pg_upgrade/dump.c                     |  24 +++
 src/bin/pg_upgrade/meson.build                |   1 +
 src/bin/pg_upgrade/option.c                   |   7 +
 src/bin/pg_upgrade/pg_upgrade.c               |  61 +++++++
 src/bin/pg_upgrade/pg_upgrade.h               |   3 +
 .../t/003_logical_replication_slots.pl        | 150 ++++++++++++++++++
 src/tools/pgindent/typedefs.list              |   1 +
 14 files changed, 480 insertions(+), 3 deletions(-)
 create mode 100644 src/bin/pg_upgrade/t/003_logical_replication_slots.pl

diff --git a/doc/src/sgml/ref/pg_dump.sgml b/doc/src/sgml/ref/pg_dump.sgml
index e81e35c13b..6e07f85281 100644
--- a/doc/src/sgml/ref/pg_dump.sgml
+++ b/doc/src/sgml/ref/pg_dump.sgml
@@ -1206,6 +1206,16 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--logical-replication-slots-only</option></term>
+      <listitem>
+       <para>
+        Dump only logical replication slots; not the schema (data definitions),
+        nor data. This is mainly used when upgrading nodes.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
        <term><option>-?</option></term>
        <term><option>--help</option></term>
diff --git a/doc/src/sgml/ref/pgupgrade.sgml b/doc/src/sgml/ref/pgupgrade.sgml
index 7816b4c685..94e90ff506 100644
--- a/doc/src/sgml/ref/pgupgrade.sgml
+++ b/doc/src/sgml/ref/pgupgrade.sgml
@@ -240,6 +240,17 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--include-logical-replication-slots</option></term>
+      <listitem>
+       <para>
+        Upgrade logical replication slots. Only permanent replication slots
+        are included. Note that pg_upgrade does not check the installation of
+        plugins.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
       <term><option>-?</option></term>
       <term><option>--help</option></term>
diff --git a/src/bin/pg_dump/pg_backup.h b/src/bin/pg_dump/pg_backup.h
index aba780ef4b..0a4e931f9b 100644
--- a/src/bin/pg_dump/pg_backup.h
+++ b/src/bin/pg_dump/pg_backup.h
@@ -187,6 +187,7 @@ typedef struct _dumpOptions
 	int			use_setsessauth;
 	int			enable_row_security;
 	int			load_via_partition_root;
+	int			logical_slots_only;
 
 	/* default, if no "inclusion" switches appear, is to dump everything */
 	bool		include_everything;
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index 058244cd17..d824150d79 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -328,6 +328,9 @@ static void setupDumpWorker(Archive *AH);
 static TableInfo *getRootTableInfo(const TableInfo *tbinfo);
 static bool forcePartitionRootLoad(const TableInfo *tbinfo);
 
+static void getLogicalReplicationSlots(Archive *fout);
+static void dumpLogicalReplicationSlot(Archive *fout,
+									   const LogicalReplicationSlotInfo *slotinfo);
 
 int
 main(int argc, char **argv)
@@ -431,6 +434,7 @@ main(int argc, char **argv)
 		{"table-and-children", required_argument, NULL, 12},
 		{"exclude-table-and-children", required_argument, NULL, 13},
 		{"exclude-table-data-and-children", required_argument, NULL, 14},
+		{"logical-replication-slots-only", no_argument, NULL, 15},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -657,6 +661,10 @@ main(int argc, char **argv)
 										  optarg);
 				break;
 
+			case 15:			/* dump only replication slot(s) */
+				dopt.logical_slots_only = true;
+				break;
+
 			default:
 				/* getopt_long already emitted a complaint */
 				pg_log_error_hint("Try \"%s --help\" for more information.", progname);
@@ -714,6 +722,11 @@ main(int argc, char **argv)
 	if (dopt.do_nothing && dopt.dump_inserts == 0)
 		pg_fatal("option --on-conflict-do-nothing requires option --inserts, --rows-per-insert, or --column-inserts");
 
+	if (dopt.logical_slots_only && dopt.dataOnly)
+		pg_fatal("options --logical-replication-slots-only and -a/--data-only cannot be used together");
+	if (dopt.logical_slots_only && dopt.schemaOnly)
+		pg_fatal("options --logical-replication-slots-only and -s/--schema-only cannot be used together");
+
 	/* Identify archive format to emit */
 	archiveFormat = parseArchiveFormat(format, &archiveMode);
 
@@ -876,6 +889,16 @@ main(int argc, char **argv)
 			pg_fatal("no matching extensions were found");
 	}
 
+	/*
+	 * If dump logical-replication-slots-only was requested, dump only them
+	 * and skip everything else.
+	 */
+	if (dopt.logical_slots_only)
+	{
+		getLogicalReplicationSlots(fout);
+		goto dump;
+	}
+
 	/*
 	 * Dumping LOs is the default for dumps where an inclusion switch is not
 	 * used (an "include everything" dump).  -B can be used to exclude LOs
@@ -936,6 +959,8 @@ main(int argc, char **argv)
 	if (!dopt.no_security_labels)
 		collectSecLabels(fout);
 
+dump:
+
 	/* Lastly, create dummy objects to represent the section boundaries */
 	boundaryObjs = createBoundaryObjects();
 
@@ -1109,6 +1134,8 @@ help(const char *progname)
 			 "                               servers matching PATTERN\n"));
 	printf(_("  --inserts                    dump data as INSERT commands, rather than COPY\n"));
 	printf(_("  --load-via-partition-root    load partitions via the root table\n"));
+	printf(_("  --logical-replication-slots-only\n"
+			 "                               dump only logical replication slots, no schema or data\n"));
 	printf(_("  --no-comments                do not dump comments\n"));
 	printf(_("  --no-publications            do not dump publications\n"));
 	printf(_("  --no-security-labels         do not dump security label assignments\n"));
@@ -10252,6 +10279,10 @@ dumpDumpableObject(Archive *fout, DumpableObject *dobj)
 		case DO_SUBSCRIPTION:
 			dumpSubscription(fout, (const SubscriptionInfo *) dobj);
 			break;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			dumpLogicalReplicationSlot(fout,
+									   (const LogicalReplicationSlotInfo *) dobj);
+			break;
 		case DO_PRE_DATA_BOUNDARY:
 		case DO_POST_DATA_BOUNDARY:
 			/* never dumped, nothing to do */
@@ -18227,6 +18258,7 @@ addBoundaryDependencies(DumpableObject **dobjs, int numObjs,
 			case DO_PUBLICATION_REL:
 			case DO_PUBLICATION_TABLE_IN_SCHEMA:
 			case DO_SUBSCRIPTION:
+			case DO_LOGICAL_REPLICATION_SLOT:
 				/* Post-data objects: must come after the post-data boundary */
 				addObjectDependency(dobj, postDataBound->dumpId);
 				break;
@@ -18488,3 +18520,118 @@ appendReloptionsArrayAH(PQExpBuffer buffer, const char *reloptions,
 	if (!res)
 		pg_log_warning("could not parse %s array", "reloptions");
 }
+
+/*
+ * getLogicalReplicationSlots
+ *	  get information about replication slots
+ */
+static void
+getLogicalReplicationSlots(Archive *fout)
+{
+	PGresult   *res;
+	LogicalReplicationSlotInfo *slotinfo;
+	PQExpBuffer query;
+	DumpOptions *dopt = fout->dopt;
+
+	int			i_slotname;
+	int			i_plugin;
+	int			i_twophase;
+	int			i,
+				ntups;
+
+	/* Check whether we should dump or not */
+	if (fout->remoteVersion < 160000 || !dopt->logical_slots_only)
+		return;
+
+	query = createPQExpBuffer();
+
+	resetPQExpBuffer(query);
+
+	/*
+	 * Get replication slots.
+	 *
+	 * XXX: Which information must be extracted from old node? Currently three
+	 * attributes are extracted because they are used by
+	 * pg_create_logical_replication_slot().
+	 *
+	 * XXX: Do we have to support physical slots?
+	 */
+	appendPQExpBufferStr(query,
+						 "SELECT slot_name, plugin, two_phase "
+						 "FROM pg_catalog.pg_replication_slots "
+						 "WHERE database = current_database() AND temporary = false "
+						 "AND wal_status IN ('reserved', 'extended');");
+
+	res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
+
+	ntups = PQntuples(res);
+
+	i_slotname = PQfnumber(res, "slot_name");
+	i_plugin = PQfnumber(res, "plugin");
+	i_twophase = PQfnumber(res, "two_phase");
+
+	slotinfo = pg_malloc(ntups * sizeof(LogicalReplicationSlotInfo));
+
+	for (i = 0; i < ntups; i++)
+	{
+		slotinfo[i].dobj.objType = DO_LOGICAL_REPLICATION_SLOT;
+
+		slotinfo[i].dobj.catId.tableoid = InvalidOid;
+		slotinfo[i].dobj.catId.oid = InvalidOid;
+		AssignDumpId(&slotinfo[i].dobj);
+
+		slotinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_slotname));
+
+		slotinfo[i].plugin = pg_strdup(PQgetvalue(res, i, i_plugin));
+		slotinfo[i].twophase = pg_strdup(PQgetvalue(res, i, i_twophase));
+
+		/*
+		 * Note: Currently we do not have any options to include/exclude slots
+		 * in dumping, so all the slots must be selected.
+		 */
+		slotinfo[i].dobj.dump = DUMP_COMPONENT_ALL;
+	}
+	PQclear(res);
+
+	destroyPQExpBuffer(query);
+}
+
+/*
+ * dumpLogicalReplicationSlot
+ *	  dump creation functions for the given logical replication slots
+ */
+static void
+dumpLogicalReplicationSlot(Archive *fout,
+						   const LogicalReplicationSlotInfo *slotinfo)
+{
+	DumpOptions *dopt = fout->dopt;
+
+	if (!dopt->logical_slots_only)
+		return;
+
+	if (slotinfo->dobj.dump & DUMP_COMPONENT_DEFINITION)
+	{
+		PQExpBuffer query = createPQExpBuffer();
+		char	   *slotname = pg_strdup(slotinfo->dobj.name);
+
+		/*
+		 * XXX: For simplification, pg_create_logical_replication_slot() is
+		 * used. Is it sufficient?
+		 */
+		appendPQExpBuffer(query, "SELECT pg_catalog.pg_create_logical_replication_slot('%s', ",
+						  slotname);
+		appendStringLiteralAH(query, slotinfo->plugin, fout);
+		appendPQExpBuffer(query, ", false, ");
+		appendStringLiteralAH(query, slotinfo->twophase, fout);
+		appendPQExpBuffer(query, ");");
+
+		ArchiveEntry(fout, slotinfo->dobj.catId, slotinfo->dobj.dumpId,
+					 ARCHIVE_OPTS(.tag = slotname,
+								  .description = "REPLICATION SLOT",
+								  .section = SECTION_POST_DATA,
+								  .createStmt = query->data));
+
+		pfree(slotname);
+		destroyPQExpBuffer(query);
+	}
+}
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index ed6ce41ad7..8028ccf6ff 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -82,7 +82,8 @@ typedef enum
 	DO_PUBLICATION,
 	DO_PUBLICATION_REL,
 	DO_PUBLICATION_TABLE_IN_SCHEMA,
-	DO_SUBSCRIPTION
+	DO_SUBSCRIPTION,
+	DO_LOGICAL_REPLICATION_SLOT
 } DumpableObjectType;
 
 /*
@@ -666,6 +667,20 @@ typedef struct _SubscriptionInfo
 	char	   *subpasswordrequired;
 } SubscriptionInfo;
 
+/*
+ * The LogicalReplicationSlotInfo struct is used to represent replication
+ * slots.
+ *
+ * XXX: add more attributes if needed
+ */
+typedef struct _LogicalReplicationSlotInfo
+{
+	DumpableObject dobj;
+	char	   *plugin;
+	char	   *slottype;
+	char	   *twophase;
+} LogicalReplicationSlotInfo;
+
 /*
  *	common utility functions
  */
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index 745578d855..4e12e46dc5 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -93,6 +93,7 @@ enum dbObjectTypePriorities
 	PRIO_PUBLICATION_REL,
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,
 	PRIO_SUBSCRIPTION,
+	PRIO_LOGICAL_REPLICATION_SLOT,
 	PRIO_DEFAULT_ACL,			/* done in ACL pass */
 	PRIO_EVENT_TRIGGER,			/* must be next to last! */
 	PRIO_REFRESH_MATVIEW		/* must be last! */
@@ -146,10 +147,11 @@ static const int dbObjectTypePriority[] =
 	PRIO_PUBLICATION,			/* DO_PUBLICATION */
 	PRIO_PUBLICATION_REL,		/* DO_PUBLICATION_REL */
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,	/* DO_PUBLICATION_TABLE_IN_SCHEMA */
-	PRIO_SUBSCRIPTION			/* DO_SUBSCRIPTION */
+	PRIO_SUBSCRIPTION,			/* DO_SUBSCRIPTION */
+	PRIO_LOGICAL_REPLICATION_SLOT	/* DO_LOGICAL_REPLICATION_SLOT */
 };
 
-StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_SUBSCRIPTION + 1),
+StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_LOGICAL_REPLICATION_SLOT + 1),
 				 "array length mismatch");
 
 static DumpId preDataBoundId;
@@ -1498,6 +1500,11 @@ describeDumpableObject(DumpableObject *obj, char *buf, int bufsize)
 					 "SUBSCRIPTION (ID %d OID %u)",
 					 obj->dumpId, obj->catId.oid);
 			return;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			snprintf(buf, bufsize,
+					 "LOGICAL REPLICATION SLOT (ID %d NAME %s)",
+					 obj->dumpId, obj->name);
+			return;
 		case DO_PRE_DATA_BOUNDARY:
 			snprintf(buf, bufsize,
 					 "PRE-DATA BOUNDARY  (ID %d)",
diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index fea159689e..46a7d64448 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -30,6 +30,7 @@ static void check_for_jsonb_9_4_usage(ClusterInfo *cluster);
 static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
+static void check_for_parameter_settings(ClusterInfo *new_cluster);
 
 
 /*
@@ -210,6 +211,9 @@ check_new_cluster(void)
 	check_for_prepared_transactions(&new_cluster);
 
 	check_for_new_tablespace_dir(&new_cluster);
+
+	if (user_opts.include_logical_slots)
+		check_for_parameter_settings(&new_cluster);
 }
 
 
@@ -1402,3 +1406,38 @@ check_for_user_defined_encoding_conversions(ClusterInfo *cluster)
 	else
 		check_ok();
 }
+
+/*
+ * Verify parameter settings for creating logical replication slots
+ */
+static void
+check_for_parameter_settings(ClusterInfo *new_cluster)
+{
+	PGresult   *res;
+	PGconn	   *conn = connectToServer(new_cluster, "template1");
+	int			max_replication_slots;
+	char	   *wal_level;
+
+	prep_status("Checking for logical replication slots");
+
+	res = executeQueryOrDie(conn, "SHOW max_replication_slots;");
+	max_replication_slots = atoi(PQgetvalue(res, 0, 0));
+
+	if (max_replication_slots == 0)
+		pg_fatal("max_replication_slots must be greater than 0");
+
+	PQclear(res);
+
+	res = executeQueryOrDie(conn, "SHOW wal_level;");
+	wal_level = PQgetvalue(res, 0, 0);
+
+	if (strcmp(wal_level, "logical") != 0)
+		pg_fatal("wal_level must be \"logical\", but set to \"%s\"",
+				  wal_level);
+
+	PQclear(res);
+
+	PQfinish(conn);
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/dump.c b/src/bin/pg_upgrade/dump.c
index 6c8c82dca8..e6b90864f5 100644
--- a/src/bin/pg_upgrade/dump.c
+++ b/src/bin/pg_upgrade/dump.c
@@ -59,6 +59,30 @@ generate_old_dump(void)
 						   log_opts.dumpdir,
 						   sql_file_name, escaped_connstr.data);
 
+		/*
+		 * Dump logical replication slots if needed.
+		 *
+		 * XXX We cannot dump replication slots at the same time as the schema
+		 * dump because we need to separate the timing of restoring
+		 * replication slots and other objects. Replication slots, in
+		 * particular, should not be restored before executing the pg_resetwal
+		 * command because it will remove WALs that are required by the slots.
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			char		slots_file_name[MAXPGPATH];
+
+			snprintf(slots_file_name, sizeof(slots_file_name),
+					 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+			parallel_exec_prog(log_file_name, NULL,
+							   "\"%s/pg_dump\" %s --logical-replication-slots-only "
+							   "--quote-all-identifiers --binary-upgrade %s "
+							   "--file=\"%s/%s\" %s",
+							   new_cluster.bindir, cluster_conn_opts(&old_cluster),
+							   log_opts.verbose ? "--verbose" : "",
+							   log_opts.dumpdir,
+							   slots_file_name, escaped_connstr.data);
+		}
 		termPQExpBuffer(&escaped_connstr);
 	}
 
diff --git a/src/bin/pg_upgrade/meson.build b/src/bin/pg_upgrade/meson.build
index 12a97f84e2..228f29b688 100644
--- a/src/bin/pg_upgrade/meson.build
+++ b/src/bin/pg_upgrade/meson.build
@@ -42,6 +42,7 @@ tests += {
     'tests': [
       't/001_basic.pl',
       't/002_pg_upgrade.pl',
+      't/003_logical_replication_slots.pl',
     ],
     'test_kwargs': {'priority': 40}, # pg_upgrade tests are slow
   },
diff --git a/src/bin/pg_upgrade/option.c b/src/bin/pg_upgrade/option.c
index 8869b6b60d..f7b5ec9879 100644
--- a/src/bin/pg_upgrade/option.c
+++ b/src/bin/pg_upgrade/option.c
@@ -57,6 +57,7 @@ parseCommandLine(int argc, char *argv[])
 		{"verbose", no_argument, NULL, 'v'},
 		{"clone", no_argument, NULL, 1},
 		{"copy", no_argument, NULL, 2},
+		{"include-logical-replication-slots", no_argument, NULL, 3},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -199,6 +200,10 @@ parseCommandLine(int argc, char *argv[])
 				user_opts.transfer_mode = TRANSFER_MODE_COPY;
 				break;
 
+			case 3:
+				user_opts.include_logical_slots = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						os_info.progname);
@@ -289,6 +294,8 @@ usage(void)
 	printf(_("  -V, --version                 display version information, then exit\n"));
 	printf(_("  --clone                       clone instead of copying files to new cluster\n"));
 	printf(_("  --copy                        copy files to new cluster (default)\n"));
+	printf(_("  --include-logical-replication-slots\n"
+			 "                                upgrade logical replication slots\n"));
 	printf(_("  -?, --help                    show this help, then exit\n"));
 	printf(_("\n"
 			 "Before running pg_upgrade you must:\n"
diff --git a/src/bin/pg_upgrade/pg_upgrade.c b/src/bin/pg_upgrade/pg_upgrade.c
index 75bab0a04c..1241060f4e 100644
--- a/src/bin/pg_upgrade/pg_upgrade.c
+++ b/src/bin/pg_upgrade/pg_upgrade.c
@@ -59,6 +59,7 @@ static void copy_xact_xlog_xid(void);
 static void set_frozenxids(bool minmxid_only);
 static void make_outputdirs(char *pgdata);
 static void setup(char *argv0, bool *live_check);
+static void create_logical_replication_slots(void);
 
 ClusterInfo old_cluster,
 			new_cluster;
@@ -188,6 +189,19 @@ main(int argc, char **argv)
 			  new_cluster.pgdata);
 	check_ok();
 
+	/*
+	 * Create logical replication slots if requested.
+	 *
+	 * Note: This must be done after doing pg_resetwal command because the
+	 * command will remove required WALs.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		start_postmaster(&new_cluster, true);
+		create_logical_replication_slots();
+		stop_postmaster(false);
+	}
+
 	if (user_opts.do_sync)
 	{
 		prep_status("Sync data directory to disk");
@@ -860,3 +874,50 @@ set_frozenxids(bool minmxid_only)
 
 	check_ok();
 }
+
+/*
+ * create_logical_replication_slots()
+ *
+ * Similar to create_new_objects() but only restores logical replication slots.
+ */
+static void
+create_logical_replication_slots(void)
+{
+	int			dbnum;
+
+	prep_status_progress("Restoring logical replication slots in the new cluster");
+
+	for (dbnum = 0; dbnum < old_cluster.dbarr.ndbs; dbnum++)
+	{
+		char		slots_file_name[MAXPGPATH],
+					log_file_name[MAXPGPATH];
+		DbInfo	   *old_db = &old_cluster.dbarr.dbs[dbnum];
+
+		pg_log(PG_STATUS, "%s", old_db->db_name);
+
+		snprintf(slots_file_name, sizeof(slots_file_name),
+				 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+		snprintf(log_file_name, sizeof(log_file_name),
+				 DB_DUMP_LOG_FILE_MASK, old_db->db_oid);
+
+		parallel_exec_prog(log_file_name,
+						   NULL,
+						   "\"%s/psql\" %s --echo-queries --set ON_ERROR_STOP=on "
+						   "--no-psqlrc --dbname %s -f \"%s/%s\"",
+						   new_cluster.bindir,
+						   cluster_conn_opts(&new_cluster),
+						   old_db->db_name,
+						   log_opts.dumpdir,
+						   slots_file_name);
+	}
+
+	/* reap all children */
+	while (reap_child(true) == true)
+		;
+
+	end_progress_output();
+	check_ok();
+
+	/* update new_cluster info now that we have objects in the databases */
+	get_db_and_rel_infos(&new_cluster);
+}
diff --git a/src/bin/pg_upgrade/pg_upgrade.h b/src/bin/pg_upgrade/pg_upgrade.h
index 3eea0139c7..5f3d7a407e 100644
--- a/src/bin/pg_upgrade/pg_upgrade.h
+++ b/src/bin/pg_upgrade/pg_upgrade.h
@@ -29,6 +29,7 @@
 /* contains both global db information and CREATE DATABASE commands */
 #define GLOBALS_DUMP_FILE	"pg_upgrade_dump_globals.sql"
 #define DB_DUMP_FILE_MASK	"pg_upgrade_dump_%u.custom"
+#define DB_DUMP_LOGICAL_SLOTS_FILE_MASK	"pg_upgrade_dump_%u_logical_slots.sql"
 
 /*
  * Base directories that include all the files generated internally, from the
@@ -304,6 +305,8 @@ typedef struct
 	transferMode transfer_mode; /* copy files or link them? */
 	int			jobs;			/* number of processes/threads to use */
 	char	   *socketdir;		/* directory to use for Unix sockets */
+	bool		include_logical_slots;	/* true -> dump and restore logical
+										 * replication slots */
 } UserOpts;
 
 typedef struct
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
new file mode 100644
index 0000000000..1159487f94
--- /dev/null
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -0,0 +1,150 @@
+# Copyright (c) 2023, PostgreSQL Global Development Group
+
+# Tests for logical replication, especially for upgrading replication slots
+
+use strict;
+use warnings;
+
+use File::Path qw(rmtree);
+
+use PostgreSQL::Test::Cluster;
+use PostgreSQL::Test::Utils;
+use Test::More;
+
+# Can be changed to test the other modes.
+my $mode = $ENV{PG_TEST_PG_UPGRADE_MODE} || '--copy';
+
+# Initialize old publisher node
+my $old_publisher = PostgreSQL::Test::Cluster->new('old_publisher');
+$old_publisher->init(allows_streaming => 'logical');
+$old_publisher->start;
+
+# Initialize subscriber node
+my $subscriber = PostgreSQL::Test::Cluster->new('subscriber');
+$subscriber->init(allows_streaming => 'logical');
+$subscriber->start;
+
+# Schema setup
+$old_publisher->safe_psql('postgres',
+	"CREATE TABLE tbl AS SELECT generate_series(1,10) AS a");
+$subscriber->safe_psql('postgres', "CREATE TABLE tbl (a int)");
+
+# Initialize new publisher node
+my $new_publisher = PostgreSQL::Test::Cluster->new('new_publisher');
+$new_publisher->init(allows_streaming => 1);
+
+my $bindir = $new_publisher->config_data('--bindir');
+
+$old_publisher->stop;
+
+# Cause a failure at the start of pg_upgrade because wal_level is replica
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
+		$mode,        '--include-logical-replication-slots',
+		'--check'
+	],
+	'run of pg_upgrade of old publisher with wrong wal_level');
+ok( -d $new_publisher->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_publisher->data_dir . "/pg_upgrade_output.d");
+
+$new_publisher->append_conf('postgresql.conf', "wal_level = 'logical'");
+$new_publisher->append_conf('postgresql.conf', "max_replication_slots = 0");
+
+# Cause a failure at the start of pg_upgrade because max_replication_slots is 0
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
+		$mode,        '--include-logical-replication-slots',
+		'--check'
+	],
+	'run of pg_upgrade of old publisher with wrong max_replication_slots');
+ok( -d $new_publisher->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_publisher->data_dir . "/pg_upgrade_output.d");
+$new_publisher->append_conf('postgresql.conf', "max_replication_slots = 10");
+
+$old_publisher->start;
+
+# Setup logical replication
+my $old_connstr = $old_publisher->connstr . ' dbname=postgres';
+$old_publisher->safe_psql('postgres',
+	"CREATE PUBLICATION pub FOR ALL TABLES");
+$subscriber->safe_psql('postgres',
+	"CREATE SUBSCRIPTION sub CONNECTION '$old_connstr' PUBLICATION pub");
+
+# Wait for initial table sync to finish
+$subscriber->wait_for_subscription_sync($old_publisher, 'sub');
+
+my $result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(10), 'check initial rows on subscriber');
+
+# Define another replication slot which allows to decode prepared transactions
+$old_publisher->safe_psql('postgres',
+	"SELECT pg_catalog.pg_create_logical_replication_slot('twophase_slot', 'pgoutput', false, true)");
+
+# Preparations for upgrading publisher
+$old_publisher->stop;
+$subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub DISABLE");
+
+# Actual run, pg_upgrade_output.d is removed at the end.
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
+		$mode,        '--include-logical-replication-slots'
+	],
+	'run of pg_upgrade of old publisher');
+ok( !-d $new_publisher->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+# Check whether the replication slot is copied to new publisher
+$new_publisher->start;
+$result =
+  $new_publisher->safe_psql('postgres',
+	"SELECT slot_name, two_phase FROM pg_replication_slots");
+is($result, qq(sub|f
+twophase_slot|t), 'check the replication slot is copied to new publisher');
+
+# Change connection string and enable logical replication
+my $new_connstr = $new_publisher->connstr . ' dbname=postgres';
+
+$subscriber->safe_psql('postgres',
+	"ALTER SUBSCRIPTION sub CONNECTION '$new_connstr'");
+$subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub ENABLE");
+
+# Check whether changes on the new publisher get replicated to the subscriber
+$new_publisher->safe_psql('postgres',
+	"INSERT INTO tbl VALUES (generate_series(11, 20))");
+
+$new_publisher->wait_for_catchup('sub');
+
+$result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(20), 'check changes are shipped to subscriber');
+
+done_testing();
diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list
index b4058b88c3..7e999726c2 100644
--- a/src/tools/pgindent/typedefs.list
+++ b/src/tools/pgindent/typedefs.list
@@ -1479,6 +1479,7 @@ LogicalRepBeginData
 LogicalRepCommitData
 LogicalRepCommitPreparedTxnData
 LogicalRepCtxStruct
+LogicalReplicationSlotInfo
 LogicalRepMode
 LogicalRepMsgType
 LogicalRepPartMapEntry
-- 
2.27.0

From 8af435355351dc655c00e732a568b323466a1f87 Mon Sep 17 00:00:00 2001
From: Julien Rouhaud <julien.rouhaud@free.fr>
Date: Fri, 14 Apr 2023 13:49:09 +0800
Subject: [PATCH v8 2/3] Always persist to disk logical slots during a shutdown
 checkpoint.

It's entirely possible for a logical slot to have a confirmed_flush_lsn higher
than the last value saved on disk while not being marked as dirty.  It's
currently not a problem to lose that value during a clean shutdown / restart
cycle, but a later patch adding support for pg_upgrade of publications and
logical slots will rely on that value being properly persisted to disk.

Author: Julien Rouhaud
Reviewed-by: FIXME
Discussion: FIXME
---
 src/backend/access/transam/xlog.c |  2 +-
 src/backend/replication/slot.c    | 26 ++++++++++++++++----------
 src/include/replication/slot.h    |  2 +-
 3 files changed, 18 insertions(+), 12 deletions(-)

diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 63481d826f..07d775cf33 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7011,7 +7011,7 @@ static void
 CheckPointGuts(XLogRecPtr checkPointRedo, int flags)
 {
 	CheckPointRelationMap();
-	CheckPointReplicationSlots();
+	CheckPointReplicationSlots(flags & CHECKPOINT_IS_SHUTDOWN);
 	CheckPointSnapBuild();
 	CheckPointLogicalRewriteHeap();
 	CheckPointReplicationOrigin();
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index 8021aaa0a8..aeea6ffd1f 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -109,7 +109,8 @@ static void ReplicationSlotDropPtr(ReplicationSlot *slot);
 /* internal persistency functions */
 static void RestoreSlotFromDisk(const char *name);
 static void CreateSlotOnDisk(ReplicationSlot *slot);
-static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel);
+static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+						   bool is_shutdown);
 
 /*
  * Report shared-memory space needed by ReplicationSlotsShmemInit.
@@ -783,7 +784,7 @@ ReplicationSlotSave(void)
 	Assert(MyReplicationSlot != NULL);
 
 	sprintf(path, "pg_replslot/%s", NameStr(MyReplicationSlot->data.name));
-	SaveSlotToPath(MyReplicationSlot, path, ERROR);
+	SaveSlotToPath(MyReplicationSlot, path, ERROR, false);
 }
 
 /*
@@ -1565,11 +1566,10 @@ restart:
 /*
  * Flush all replication slots to disk.
  *
- * This needn't actually be part of a checkpoint, but it's a convenient
- * location.
+ * is_shutdown is true in case of a shutdown checkpoint.
  */
 void
-CheckPointReplicationSlots(void)
+CheckPointReplicationSlots(bool is_shutdown)
 {
 	int			i;
 
@@ -1594,7 +1594,7 @@ CheckPointReplicationSlots(void)
 
 		/* save the slot to disk, locking is handled in SaveSlotToPath() */
 		sprintf(path, "pg_replslot/%s", NameStr(s->data.name));
-		SaveSlotToPath(s, path, LOG);
+		SaveSlotToPath(s, path, LOG, is_shutdown);
 	}
 	LWLockRelease(ReplicationSlotAllocationLock);
 }
@@ -1700,7 +1700,7 @@ CreateSlotOnDisk(ReplicationSlot *slot)
 
 	/* Write the actual state file. */
 	slot->dirty = true;			/* signal that we really need to write */
-	SaveSlotToPath(slot, tmppath, ERROR);
+	SaveSlotToPath(slot, tmppath, ERROR, false);
 
 	/* Rename the directory into place. */
 	if (rename(tmppath, path) != 0)
@@ -1726,7 +1726,8 @@ CreateSlotOnDisk(ReplicationSlot *slot)
  * Shared functionality between saving and creating a replication slot.
  */
 static void
-SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
+SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+			   bool is_shutdown)
 {
 	char		tmppath[MAXPGPATH];
 	char		path[MAXPGPATH];
@@ -1740,8 +1741,13 @@ SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
 	slot->just_dirtied = false;
 	SpinLockRelease(&slot->mutex);
 
-	/* and don't do anything if there's nothing to write */
-	if (!was_dirty)
+	/*
+	 * and don't do anything if there's nothing to write, unless it's this is
+	 * called for a logical slot during a shutdown checkpoint, as we want to
+	 * persist the confirmed_flush_lsn in that case, even if that's the only
+	 * modification.
+	 */
+	if (!was_dirty && !is_shutdown && !SlotIsLogical(slot))
 		return;
 
 	LWLockAcquire(&slot->io_in_progress_lock, LW_EXCLUSIVE);
diff --git a/src/include/replication/slot.h b/src/include/replication/slot.h
index a8a89dc784..7ca37c9f70 100644
--- a/src/include/replication/slot.h
+++ b/src/include/replication/slot.h
@@ -241,7 +241,7 @@ extern void ReplicationSlotNameForTablesync(Oid suboid, Oid relid, char *syncslo
 extern void ReplicationSlotDropAtPubNode(WalReceiverConn *wrconn, char *slotname, bool missing_ok);
 
 extern void StartupReplicationSlots(void);
-extern void CheckPointReplicationSlots(void);
+extern void CheckPointReplicationSlots(bool is_shutdown);
 
 extern void CheckSlotRequirements(void);
 extern void CheckSlotPermissions(void);
-- 
2.27.0

From 18cc4b037143d576c025d3a37a8eccbcea66ce63 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Fri, 14 Apr 2023 08:59:03 +0000
Subject: [PATCH v8 3/3] pg_upgrade: Add check function for
 --include-logical-replication-slots option

---
 src/bin/pg_upgrade/check.c                    | 81 ++++++++++++++++++-
 .../t/003_logical_replication_slots.pl        | 38 ++++++++-
 2 files changed, 115 insertions(+), 4 deletions(-)

diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index 46a7d64448..b1e1f516c9 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -9,7 +9,10 @@
 
 #include "postgres_fe.h"
 
+#include "access/xlogrecord.h"
+#include "access/xlog_internal.h"
 #include "catalog/pg_authid_d.h"
+#include "catalog/pg_control.h"
 #include "catalog/pg_collation.h"
 #include "fe_utils/string_utils.h"
 #include "mb/pg_wchar.h"
@@ -31,7 +34,7 @@ static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
 static void check_for_parameter_settings(ClusterInfo *new_cluster);
-
+static void check_for_confirmed_flush_lsn(ClusterInfo *cluster);
 
 /*
  * fix_path_separator
@@ -104,6 +107,8 @@ check_and_dump_old_cluster(bool live_check)
 	check_for_composite_data_type_usage(&old_cluster);
 	check_for_reg_data_type_usage(&old_cluster);
 	check_for_isn_and_int8_passing_mismatch(&old_cluster);
+	if (user_opts.include_logical_slots)
+		check_for_confirmed_flush_lsn(&old_cluster);
 
 	/*
 	 * PG 16 increased the size of the 'aclitem' type, which breaks the on-disk
@@ -1418,6 +1423,10 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 	int			max_replication_slots;
 	char	   *wal_level;
 
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(new_cluster->major_version < 1600))
+		return;
+
 	prep_status("Checking for logical replication slots");
 
 	res = executeQueryOrDie(conn, "SHOW max_replication_slots;");
@@ -1441,3 +1450,73 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 
 	check_ok();
 }
+
+/*
+ * Verify that all logical replication slots consumed all WALs, except a
+ * CHECKPOINT_SHUTDOWN record.
+ */
+static void
+check_for_confirmed_flush_lsn(ClusterInfo *cluster)
+{
+	int			i,
+				ntups,
+				i_slotname;
+	bool		is_error = false;
+	PGresult   *res;
+	DbInfo	   *active_db = &cluster->dbarr.dbs[0];
+	PGconn	   *conn = connectToServer(cluster, active_db->db_name);
+
+	Assert(user_opts.include_logical_slots);
+
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(cluster->major_version < 1600))
+		return;
+
+	prep_status("Checking for logical replication slots");
+
+
+	/*
+	 * Check that all logical replication slots have reached the current WAL
+	 * position, except for the CHECKPOINT_SHUTDOWN record. Even if all WALs
+	 * are consumed before shutting down the node, the checkpointer generates
+	 * a CHECKPOINT_SHUTDOWN record at shutdown, which cannot be consumed by
+	 * any slots. Therefore, we must allow for a difference between
+	 * pg_current_wal_insert_lsn() and confirmed_flush_lsn.
+	 */
+#define SHUTDOWN_RECORD_SIZE  (SizeOfXLogRecord + \
+							   SizeOfXLogRecordDataHeaderShort + \
+							   sizeof(CheckPoint))
+
+	res = executeQueryOrDie(conn,
+							"SELECT slot_name FROM pg_catalog.pg_replication_slots "
+							"WHERE (pg_catalog.pg_current_wal_insert_lsn() - confirmed_flush_lsn) > %d "
+							"AND temporary = false AND wal_status IN ('reserved', 'extended');",
+							(int) (SizeOfXLogShortPHD + SHUTDOWN_RECORD_SIZE));
+
+#undef SHUTDOWN_RECORD_SIZE
+
+	ntups = PQntuples(res);
+	i_slotname = PQfnumber(res, "slot_name");
+
+	for (i = 0; i < ntups; i++)
+	{
+		char	   *slotname;
+
+		is_error = true;
+
+		slotname = PQgetvalue(res, i, i_slotname);
+
+		pg_log(PG_WARNING,
+			   "\nWARNING: logical replication slot \"%s\" has not consumed WALs yet.",
+			   slotname);
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	if (is_error)
+		pg_fatal("--include-logical-replication-slot requires that all "
+				 "logical replication slots consumed all the WALs");
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
index 1159487f94..cb7acb3302 100644
--- a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -85,6 +85,38 @@ $new_publisher->append_conf('postgresql.conf', "max_replication_slots = 10");
 
 $old_publisher->start;
 
+# Create a dummy slot on old publisher to fail the test
+$old_publisher->safe_psql('postgres',
+	"SELECT pg_create_logical_replication_slot('dropme_slot', 'pgoutput')");
+$old_publisher->safe_psql('postgres',
+	"INSERT INTO tbl VALUES (generate_series(11, 20))");
+$old_publisher->stop;
+
+# Cause a failure at the start of pg_upgrade because dropme_slot is not
+# advanced till the INSERT statement
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
+		$mode,        '--include-logical-replication-slots',
+		'--check'
+	],
+	'run of pg_upgrade of old publisher with idle replication slots');
+ok( -d $new_publisher->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_publisher->data_dir . "/pg_upgrade_output.d");
+$old_publisher->start;
+$old_publisher->safe_psql('postgres',
+	"SELECT pg_drop_replication_slot('dropme_slot')");
+
 # Setup logical replication
 my $old_connstr = $old_publisher->connstr . ' dbname=postgres';
 $old_publisher->safe_psql('postgres',
@@ -96,7 +128,7 @@ $subscriber->safe_psql('postgres',
 $subscriber->wait_for_subscription_sync($old_publisher, 'sub');
 
 my $result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
-is($result, qq(10), 'check initial rows on subscriber');
+is($result, qq(20), 'check initial rows on subscriber');
 
 # Define another replication slot which allows to decode prepared transactions
 $old_publisher->safe_psql('postgres',
@@ -140,11 +172,11 @@ $subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub ENABLE");
 
 # Check whether changes on the new publisher get replicated to the subscriber
 $new_publisher->safe_psql('postgres',
-	"INSERT INTO tbl VALUES (generate_series(11, 20))");
+	"INSERT INTO tbl VALUES (generate_series(21, 30))");
 
 $new_publisher->wait_for_catchup('sub');
 
 $result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
-is($result, qq(20), 'check changes are shipped to subscriber');
+is($result, qq(30), 'check changes are shipped to subscriber');
 
 done_testing();
-- 
2.27.0

From abc975cb9eae0088231b72fc6264024d0f5869fe Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Tue, 4 Apr 2023 05:49:34 +0000
Subject: [PATCH v9 1/4] pg_upgrade: Add --include-logical-replication-slots
 option

This commit introduces a new pg_upgrade option called "--include-logical-replication-slots".
This allows nodes with logical replication slots to be upgraded. The commit can
be divided into two parts: one for pg_dump and another for pg_upgrade.

For pg_dump this commit includes a new option called "--logical-replication-slots-only".
This option can be used to dump logical replication slots. When this option is
specified, the slot_name, plugin, and two_phase parameters are extracted from
pg_replication_slots. An SQL file is then generated which executes
pg_create_logical_replication_slot() with the extracted parameters.

For pg_upgrade, when '--include-logical-replication-slots' is specified, it executes
pg_dump with the new "--logical-replication-slots-only" option and restores from the
dump. Apart from restoring schema, pg_resetwal must not be called after restoring
replication slots. This is because the command discards WAL files and starts from a
new segment, even if they are required by replication slots. This leads to an ERROR:
"requested WAL segment XXX has already been removed". To avoid this, replication slots
are restored at a different time than other objects, after running pg_resetwal.

The significant advantage of this commit is that it makes it easy to continue
logical replication even after upgrading the publisher node. Previously, pg_upgrade
allowed copying publications to a new node. With this new commit, adjusting the
connection string to the new publisher will cause the apply worker on the subscriber
to connect to the new publisher automatically. This enables seamless continuation
of logical replication, even after an upgrade.

Author: Hayato Kuroda
Reviewed-by: Peter Smith, Julien Rouhaud, Vignesh C
---
 doc/src/sgml/ref/pg_dump.sgml                 |  10 ++
 doc/src/sgml/ref/pgupgrade.sgml               |  11 ++
 src/bin/pg_dump/pg_backup.h                   |   1 +
 src/bin/pg_dump/pg_dump.c                     | 150 ++++++++++++++++++
 src/bin/pg_dump/pg_dump.h                     |  17 +-
 src/bin/pg_dump/pg_dump_sort.c                |  11 +-
 src/bin/pg_upgrade/check.c                    |  57 +++++++
 src/bin/pg_upgrade/dump.c                     |  24 +++
 src/bin/pg_upgrade/info.c                     | 116 +++++++++++++-
 src/bin/pg_upgrade/meson.build                |   1 +
 src/bin/pg_upgrade/option.c                   |   7 +
 src/bin/pg_upgrade/pg_upgrade.c               |  61 +++++++
 src/bin/pg_upgrade/pg_upgrade.h               |  18 +++
 .../t/003_logical_replication_slots.pl        | 150 ++++++++++++++++++
 src/tools/pgindent/typedefs.list              |   3 +
 15 files changed, 633 insertions(+), 4 deletions(-)
 create mode 100644 src/bin/pg_upgrade/t/003_logical_replication_slots.pl

diff --git a/doc/src/sgml/ref/pg_dump.sgml b/doc/src/sgml/ref/pg_dump.sgml
index e81e35c13b..6e07f85281 100644
--- a/doc/src/sgml/ref/pg_dump.sgml
+++ b/doc/src/sgml/ref/pg_dump.sgml
@@ -1206,6 +1206,16 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--logical-replication-slots-only</option></term>
+      <listitem>
+       <para>
+        Dump only logical replication slots; not the schema (data definitions),
+        nor data. This is mainly used when upgrading nodes.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
        <term><option>-?</option></term>
        <term><option>--help</option></term>
diff --git a/doc/src/sgml/ref/pgupgrade.sgml b/doc/src/sgml/ref/pgupgrade.sgml
index 7816b4c685..94e90ff506 100644
--- a/doc/src/sgml/ref/pgupgrade.sgml
+++ b/doc/src/sgml/ref/pgupgrade.sgml
@@ -240,6 +240,17 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--include-logical-replication-slots</option></term>
+      <listitem>
+       <para>
+        Upgrade logical replication slots. Only permanent replication slots
+        are included. Note that pg_upgrade does not check the installation of
+        plugins.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
       <term><option>-?</option></term>
       <term><option>--help</option></term>
diff --git a/src/bin/pg_dump/pg_backup.h b/src/bin/pg_dump/pg_backup.h
index aba780ef4b..0a4e931f9b 100644
--- a/src/bin/pg_dump/pg_backup.h
+++ b/src/bin/pg_dump/pg_backup.h
@@ -187,6 +187,7 @@ typedef struct _dumpOptions
 	int			use_setsessauth;
 	int			enable_row_security;
 	int			load_via_partition_root;
+	int			logical_slots_only;
 
 	/* default, if no "inclusion" switches appear, is to dump everything */
 	bool		include_everything;
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index 058244cd17..ed98572cce 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -328,6 +328,9 @@ static void setupDumpWorker(Archive *AH);
 static TableInfo *getRootTableInfo(const TableInfo *tbinfo);
 static bool forcePartitionRootLoad(const TableInfo *tbinfo);
 
+static void getLogicalReplicationSlots(Archive *fout);
+static void dumpLogicalReplicationSlot(Archive *fout,
+									   const LogicalReplicationSlotInfo *slotinfo);
 
 int
 main(int argc, char **argv)
@@ -431,6 +434,7 @@ main(int argc, char **argv)
 		{"table-and-children", required_argument, NULL, 12},
 		{"exclude-table-and-children", required_argument, NULL, 13},
 		{"exclude-table-data-and-children", required_argument, NULL, 14},
+		{"logical-replication-slots-only", no_argument, NULL, 15},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -657,6 +661,10 @@ main(int argc, char **argv)
 										  optarg);
 				break;
 
+			case 15:			/* dump only replication slot(s) */
+				dopt.logical_slots_only = true;
+				break;
+
 			default:
 				/* getopt_long already emitted a complaint */
 				pg_log_error_hint("Try \"%s --help\" for more information.", progname);
@@ -714,6 +722,14 @@ main(int argc, char **argv)
 	if (dopt.do_nothing && dopt.dump_inserts == 0)
 		pg_fatal("option --on-conflict-do-nothing requires option --inserts, --rows-per-insert, or --column-inserts");
 
+	if (dopt.logical_slots_only && !dopt.binary_upgrade)
+		pg_fatal("options --logical-replication-slots-only requires option --binary-upgrade");
+
+	if (dopt.logical_slots_only && dopt.dataOnly)
+		pg_fatal("options --logical-replication-slots-only and -a/--data-only cannot be used together");
+	if (dopt.logical_slots_only && dopt.schemaOnly)
+		pg_fatal("options --logical-replication-slots-only and -s/--schema-only cannot be used together");
+
 	/* Identify archive format to emit */
 	archiveFormat = parseArchiveFormat(format, &archiveMode);
 
@@ -876,6 +892,16 @@ main(int argc, char **argv)
 			pg_fatal("no matching extensions were found");
 	}
 
+	/*
+	 * If dump logical-replication-slots-only was requested, dump only them
+	 * and skip everything else.
+	 */
+	if (dopt.logical_slots_only)
+	{
+		getLogicalReplicationSlots(fout);
+		goto dump;
+	}
+
 	/*
 	 * Dumping LOs is the default for dumps where an inclusion switch is not
 	 * used (an "include everything" dump).  -B can be used to exclude LOs
@@ -936,6 +962,8 @@ main(int argc, char **argv)
 	if (!dopt.no_security_labels)
 		collectSecLabels(fout);
 
+dump:
+
 	/* Lastly, create dummy objects to represent the section boundaries */
 	boundaryObjs = createBoundaryObjects();
 
@@ -1109,6 +1137,8 @@ help(const char *progname)
 			 "                               servers matching PATTERN\n"));
 	printf(_("  --inserts                    dump data as INSERT commands, rather than COPY\n"));
 	printf(_("  --load-via-partition-root    load partitions via the root table\n"));
+	printf(_("  --logical-replication-slots-only\n"
+			 "                               dump only logical replication slots, no schema or data\n"));
 	printf(_("  --no-comments                do not dump comments\n"));
 	printf(_("  --no-publications            do not dump publications\n"));
 	printf(_("  --no-security-labels         do not dump security label assignments\n"));
@@ -10252,6 +10282,10 @@ dumpDumpableObject(Archive *fout, DumpableObject *dobj)
 		case DO_SUBSCRIPTION:
 			dumpSubscription(fout, (const SubscriptionInfo *) dobj);
 			break;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			dumpLogicalReplicationSlot(fout,
+									   (const LogicalReplicationSlotInfo *) dobj);
+			break;
 		case DO_PRE_DATA_BOUNDARY:
 		case DO_POST_DATA_BOUNDARY:
 			/* never dumped, nothing to do */
@@ -18227,6 +18261,7 @@ addBoundaryDependencies(DumpableObject **dobjs, int numObjs,
 			case DO_PUBLICATION_REL:
 			case DO_PUBLICATION_TABLE_IN_SCHEMA:
 			case DO_SUBSCRIPTION:
+			case DO_LOGICAL_REPLICATION_SLOT:
 				/* Post-data objects: must come after the post-data boundary */
 				addObjectDependency(dobj, postDataBound->dumpId);
 				break;
@@ -18488,3 +18523,118 @@ appendReloptionsArrayAH(PQExpBuffer buffer, const char *reloptions,
 	if (!res)
 		pg_log_warning("could not parse %s array", "reloptions");
 }
+
+/*
+ * getLogicalReplicationSlots
+ *	  get information about replication slots
+ */
+static void
+getLogicalReplicationSlots(Archive *fout)
+{
+	PGresult   *res;
+	LogicalReplicationSlotInfo *slotinfo;
+	PQExpBuffer query;
+	DumpOptions *dopt = fout->dopt;
+
+	int			i_slotname;
+	int			i_plugin;
+	int			i_twophase;
+	int			i,
+				ntups;
+
+	/* Check whether we should dump or not */
+	if (fout->remoteVersion < 160000 || !dopt->logical_slots_only)
+		return;
+
+	query = createPQExpBuffer();
+
+	resetPQExpBuffer(query);
+
+	/*
+	 * Get replication slots.
+	 *
+	 * XXX: Which information must be extracted from old node? Currently three
+	 * attributes are extracted because they are used by
+	 * pg_create_logical_replication_slot().
+	 *
+	 * XXX: Do we have to support physical slots?
+	 */
+	appendPQExpBufferStr(query,
+						 "SELECT slot_name, plugin, two_phase "
+						 "FROM pg_catalog.pg_replication_slots "
+						 "WHERE database = current_database() AND temporary = false "
+						 "AND wal_status IN ('reserved', 'extended');");
+
+	res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
+
+	ntups = PQntuples(res);
+
+	i_slotname = PQfnumber(res, "slot_name");
+	i_plugin = PQfnumber(res, "plugin");
+	i_twophase = PQfnumber(res, "two_phase");
+
+	slotinfo = pg_malloc(ntups * sizeof(LogicalReplicationSlotInfo));
+
+	for (i = 0; i < ntups; i++)
+	{
+		slotinfo[i].dobj.objType = DO_LOGICAL_REPLICATION_SLOT;
+
+		slotinfo[i].dobj.catId.tableoid = InvalidOid;
+		slotinfo[i].dobj.catId.oid = InvalidOid;
+		AssignDumpId(&slotinfo[i].dobj);
+
+		slotinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_slotname));
+
+		slotinfo[i].plugin = pg_strdup(PQgetvalue(res, i, i_plugin));
+		slotinfo[i].twophase = (strcmp(PQgetvalue(res, i, i_twophase), "t") == 0);
+
+		/*
+		 * Note: Currently we do not have any options to include/exclude slots
+		 * in dumping, so all the slots must be selected.
+		 */
+		slotinfo[i].dobj.dump = DUMP_COMPONENT_ALL;
+	}
+	PQclear(res);
+
+	destroyPQExpBuffer(query);
+}
+
+/*
+ * dumpLogicalReplicationSlot
+ *	  dump creation functions for the given logical replication slots
+ */
+static void
+dumpLogicalReplicationSlot(Archive *fout,
+						   const LogicalReplicationSlotInfo *slotinfo)
+{
+	DumpOptions *dopt = fout->dopt;
+
+	if (!dopt->logical_slots_only)
+		return;
+
+	if (slotinfo->dobj.dump & DUMP_COMPONENT_DEFINITION)
+	{
+		PQExpBuffer query = createPQExpBuffer();
+		char	   *slotname = pg_strdup(slotinfo->dobj.name);
+
+		/*
+		 * XXX: For simplification, pg_create_logical_replication_slot() is
+		 * used. Is it sufficient?
+		 */
+		appendPQExpBuffer(query, "SELECT pg_catalog.pg_create_logical_replication_slot(");
+		appendStringLiteralAH(query, slotname, fout);
+		appendPQExpBuffer(query, ", ");
+		appendStringLiteralAH(query, slotinfo->plugin, fout);
+		appendPQExpBuffer(query, ", false, %s);",
+						  slotinfo->twophase ? "true" : "false");
+
+		ArchiveEntry(fout, slotinfo->dobj.catId, slotinfo->dobj.dumpId,
+					 ARCHIVE_OPTS(.tag = slotname,
+								  .description = "REPLICATION SLOT",
+								  .section = SECTION_POST_DATA,
+								  .createStmt = query->data));
+
+		free(slotname);
+		destroyPQExpBuffer(query);
+	}
+}
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index ed6ce41ad7..de081c35ae 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -82,7 +82,8 @@ typedef enum
 	DO_PUBLICATION,
 	DO_PUBLICATION_REL,
 	DO_PUBLICATION_TABLE_IN_SCHEMA,
-	DO_SUBSCRIPTION
+	DO_SUBSCRIPTION,
+	DO_LOGICAL_REPLICATION_SLOT
 } DumpableObjectType;
 
 /*
@@ -666,6 +667,20 @@ typedef struct _SubscriptionInfo
 	char	   *subpasswordrequired;
 } SubscriptionInfo;
 
+/*
+ * The LogicalReplicationSlotInfo struct is used to represent replication
+ * slots.
+ *
+ * XXX: add more attributes if needed
+ */
+typedef struct _LogicalReplicationSlotInfo
+{
+	DumpableObject dobj;
+	char	   *plugin;
+	char	   *slottype;
+	bool		twophase;
+} LogicalReplicationSlotInfo;
+
 /*
  *	common utility functions
  */
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index 745578d855..4e12e46dc5 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -93,6 +93,7 @@ enum dbObjectTypePriorities
 	PRIO_PUBLICATION_REL,
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,
 	PRIO_SUBSCRIPTION,
+	PRIO_LOGICAL_REPLICATION_SLOT,
 	PRIO_DEFAULT_ACL,			/* done in ACL pass */
 	PRIO_EVENT_TRIGGER,			/* must be next to last! */
 	PRIO_REFRESH_MATVIEW		/* must be last! */
@@ -146,10 +147,11 @@ static const int dbObjectTypePriority[] =
 	PRIO_PUBLICATION,			/* DO_PUBLICATION */
 	PRIO_PUBLICATION_REL,		/* DO_PUBLICATION_REL */
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,	/* DO_PUBLICATION_TABLE_IN_SCHEMA */
-	PRIO_SUBSCRIPTION			/* DO_SUBSCRIPTION */
+	PRIO_SUBSCRIPTION,			/* DO_SUBSCRIPTION */
+	PRIO_LOGICAL_REPLICATION_SLOT	/* DO_LOGICAL_REPLICATION_SLOT */
 };
 
-StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_SUBSCRIPTION + 1),
+StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_LOGICAL_REPLICATION_SLOT + 1),
 				 "array length mismatch");
 
 static DumpId preDataBoundId;
@@ -1498,6 +1500,11 @@ describeDumpableObject(DumpableObject *obj, char *buf, int bufsize)
 					 "SUBSCRIPTION (ID %d OID %u)",
 					 obj->dumpId, obj->catId.oid);
 			return;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			snprintf(buf, bufsize,
+					 "LOGICAL REPLICATION SLOT (ID %d NAME %s)",
+					 obj->dumpId, obj->name);
+			return;
 		case DO_PRE_DATA_BOUNDARY:
 			snprintf(buf, bufsize,
 					 "PRE-DATA BOUNDARY  (ID %d)",
diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index fea159689e..9674c1a2ce 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -30,6 +30,7 @@ static void check_for_jsonb_9_4_usage(ClusterInfo *cluster);
 static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
+static void check_for_parameter_settings(ClusterInfo *new_cluster);
 
 
 /*
@@ -88,6 +89,7 @@ check_and_dump_old_cluster(bool live_check)
 
 	/* Extract a list of databases and tables from the old cluster */
 	get_db_and_rel_infos(&old_cluster);
+	get_logical_slot_infos(&old_cluster);
 
 	init_tablespaces();
 
@@ -188,6 +190,7 @@ void
 check_new_cluster(void)
 {
 	get_db_and_rel_infos(&new_cluster);
+	get_logical_slot_infos(&new_cluster);
 
 	check_new_cluster_is_empty();
 
@@ -210,6 +213,9 @@ check_new_cluster(void)
 	check_for_prepared_transactions(&new_cluster);
 
 	check_for_new_tablespace_dir(&new_cluster);
+
+	if (user_opts.include_logical_slots)
+		check_for_parameter_settings(&new_cluster);
 }
 
 
@@ -364,6 +370,22 @@ check_new_cluster_is_empty(void)
 						 rel_arr->rels[relnum].nspname,
 						 rel_arr->rels[relnum].relname);
 		}
+
+		/*
+		 * If --include-logical-replication-slots is required, check the
+		 * existing of slots
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			LogicalSlotInfoArr *slot_arr = &new_cluster.dbarr.dbs[dbnum].slot_arr;
+
+			/* if nslots > 0, report just first entry and exit */
+			if (slot_arr->nslots)
+				pg_fatal("New cluster database \"%s\" is not empty: found logical replication slot \"%s\"",
+						 new_cluster.dbarr.dbs[dbnum].db_name,
+						 slot_arr->slots[0].slotname);
+		}
+
 	}
 }
 
@@ -1402,3 +1424,38 @@ check_for_user_defined_encoding_conversions(ClusterInfo *cluster)
 	else
 		check_ok();
 }
+
+/*
+ * Verify parameter settings for creating logical replication slots
+ */
+static void
+check_for_parameter_settings(ClusterInfo *new_cluster)
+{
+	PGresult   *res;
+	PGconn	   *conn = connectToServer(new_cluster, "template1");
+	int			max_replication_slots;
+	char	   *wal_level;
+
+	prep_status("Checking for logical replication slots");
+
+	res = executeQueryOrDie(conn, "SHOW max_replication_slots;");
+	max_replication_slots = atoi(PQgetvalue(res, 0, 0));
+
+	if (max_replication_slots == 0)
+		pg_fatal("max_replication_slots must be greater than 0");
+
+	PQclear(res);
+
+	res = executeQueryOrDie(conn, "SHOW wal_level;");
+	wal_level = PQgetvalue(res, 0, 0);
+
+	if (strcmp(wal_level, "logical") != 0)
+		pg_fatal("wal_level must be \"logical\", but set to \"%s\"",
+				 wal_level);
+
+	PQclear(res);
+
+	PQfinish(conn);
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/dump.c b/src/bin/pg_upgrade/dump.c
index 6c8c82dca8..e6b90864f5 100644
--- a/src/bin/pg_upgrade/dump.c
+++ b/src/bin/pg_upgrade/dump.c
@@ -59,6 +59,30 @@ generate_old_dump(void)
 						   log_opts.dumpdir,
 						   sql_file_name, escaped_connstr.data);
 
+		/*
+		 * Dump logical replication slots if needed.
+		 *
+		 * XXX We cannot dump replication slots at the same time as the schema
+		 * dump because we need to separate the timing of restoring
+		 * replication slots and other objects. Replication slots, in
+		 * particular, should not be restored before executing the pg_resetwal
+		 * command because it will remove WALs that are required by the slots.
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			char		slots_file_name[MAXPGPATH];
+
+			snprintf(slots_file_name, sizeof(slots_file_name),
+					 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+			parallel_exec_prog(log_file_name, NULL,
+							   "\"%s/pg_dump\" %s --logical-replication-slots-only "
+							   "--quote-all-identifiers --binary-upgrade %s "
+							   "--file=\"%s/%s\" %s",
+							   new_cluster.bindir, cluster_conn_opts(&old_cluster),
+							   log_opts.verbose ? "--verbose" : "",
+							   log_opts.dumpdir,
+							   slots_file_name, escaped_connstr.data);
+		}
 		termPQExpBuffer(&escaped_connstr);
 	}
 
diff --git a/src/bin/pg_upgrade/info.c b/src/bin/pg_upgrade/info.c
index 85ed15ae4a..0996fdb0a8 100644
--- a/src/bin/pg_upgrade/info.c
+++ b/src/bin/pg_upgrade/info.c
@@ -23,10 +23,12 @@ static void free_db_and_rel_infos(DbInfoArr *db_arr);
 static void get_template0_info(ClusterInfo *cluster);
 static void get_db_infos(ClusterInfo *cluster);
 static void get_rel_infos(ClusterInfo *cluster, DbInfo *dbinfo);
+static void get_logical_slot_infos_per_db(ClusterInfo *cluster, DbInfo *dbinfo);
 static void free_rel_infos(RelInfoArr *rel_arr);
+static void free_logical_slot_infos(LogicalSlotInfoArr *slot_arr);
 static void print_db_infos(DbInfoArr *db_arr);
 static void print_rel_infos(RelInfoArr *rel_arr);
-
+static void print_slot_infos(LogicalSlotInfoArr *slot_arr);
 
 /*
  * gen_db_file_maps()
@@ -600,6 +602,93 @@ get_rel_infos(ClusterInfo *cluster, DbInfo *dbinfo)
 	dbinfo->rel_arr.nrels = num_rels;
 }
 
+/*
+ * Higher level routine to generate LogicalSlotInfoArr for all databases.
+ */
+void
+get_logical_slot_infos(ClusterInfo *cluster)
+{
+	int			dbnum;
+
+	for (dbnum = 0; dbnum < cluster->dbarr.ndbs; dbnum++)
+	{
+		if (!cluster->dbarr.dbs[dbnum].slot_arr.slots)
+			free_logical_slot_infos(&cluster->dbarr.dbs[dbnum].slot_arr);
+
+		get_logical_slot_infos_per_db(cluster, &cluster->dbarr.dbs[dbnum]);
+	}
+
+	if (cluster == &old_cluster)
+		pg_log(PG_VERBOSE, "\nsource databases:");
+	else
+		pg_log(PG_VERBOSE, "\ntarget databases:");
+
+	if (log_opts.verbose)
+	{
+		for (dbnum = 0; dbnum < cluster->dbarr.ndbs; dbnum++)
+		{
+			pg_log(PG_VERBOSE, "Database: %s", cluster->dbarr.dbs[dbnum].db_name);
+			print_slot_infos(&cluster->dbarr.dbs[dbnum].slot_arr);
+		}
+	}
+}
+
+/*
+ * get_logical_slot_infos_per_db()
+ *
+ * gets the LogicalSlotInfos for all the logical replication slots of the database
+ * referred to by "dbinfo".
+ */
+static void
+get_logical_slot_infos_per_db(ClusterInfo *cluster, DbInfo *dbinfo)
+{
+	PGconn	   *conn = connectToServer(cluster,
+									   dbinfo->db_name);
+	PGresult   *res;
+	LogicalSlotInfo *slotinfos;
+
+	int			ntups;
+	int			slotnum;
+	int			num_slots = 0;
+
+	int			i_slotname;
+	int			i_plugin;
+	int			i_twophase;
+
+	char		query[QUERY_ALLOC];
+
+	query[0] = '\0';			/* initialize query string to empty */
+
+	snprintf(query + strlen(query), sizeof(query) - strlen(query),
+			 "SELECT slot_name, plugin, two_phase "
+			 "FROM pg_catalog.pg_replication_slots "
+			 "WHERE database = current_database() AND temporary = false "
+			 "AND wal_status IN ('reserved', 'extended');");
+
+	res = executeQueryOrDie(conn, "%s", query);
+
+	ntups = PQntuples(res);
+
+	slotinfos = (LogicalSlotInfo *) pg_malloc(sizeof(LogicalSlotInfo) * ntups);
+
+	i_slotname = PQfnumber(res, "slot_name");
+	i_plugin = PQfnumber(res, "plugin");
+	i_twophase = PQfnumber(res, "two_phase");
+
+	for (slotnum = 0; slotnum < ntups; slotnum++)
+	{
+		LogicalSlotInfo *curr = &slotinfos[num_slots++];
+
+		curr->slotname = pg_strdup(PQgetvalue(res, slotnum, i_slotname));
+		curr->plugin = pg_strdup(PQgetvalue(res, slotnum, i_plugin));
+		curr->two_phase = (strcmp(PQgetvalue(res, slotnum, i_twophase), "t") == 0);
+	}
+
+	PQfinish(conn);
+
+	dbinfo->slot_arr.slots = slotinfos;
+	dbinfo->slot_arr.nslots = num_slots;
+}
 
 static void
 free_db_and_rel_infos(DbInfoArr *db_arr)
@@ -634,6 +723,19 @@ free_rel_infos(RelInfoArr *rel_arr)
 	rel_arr->nrels = 0;
 }
 
+static void
+free_logical_slot_infos(LogicalSlotInfoArr *slot_arr)
+{
+	int			slotnum;
+
+	for (slotnum = 0; slotnum < slot_arr->nslots; slotnum++)
+	{
+		pg_free(slot_arr->slots[slotnum].slotname);
+		pg_free(slot_arr->slots[slotnum].plugin);
+	}
+	pg_free(slot_arr->slots);
+	slot_arr->nslots = 0;
+}
 
 static void
 print_db_infos(DbInfoArr *db_arr)
@@ -660,3 +762,15 @@ print_rel_infos(RelInfoArr *rel_arr)
 			   rel_arr->rels[relnum].reloid,
 			   rel_arr->rels[relnum].tablespace);
 }
+
+static void
+print_slot_infos(LogicalSlotInfoArr *slot_arr)
+{
+	int			slotnum;
+
+	for (slotnum = 0; slotnum < slot_arr->nslots; slotnum++)
+		pg_log(PG_VERBOSE, "slotname: %s: plugin: %s: two_phase %d",
+			   slot_arr->slots[slotnum].slotname,
+			   slot_arr->slots[slotnum].plugin,
+			   slot_arr->slots[slotnum].two_phase);
+}
diff --git a/src/bin/pg_upgrade/meson.build b/src/bin/pg_upgrade/meson.build
index 12a97f84e2..228f29b688 100644
--- a/src/bin/pg_upgrade/meson.build
+++ b/src/bin/pg_upgrade/meson.build
@@ -42,6 +42,7 @@ tests += {
     'tests': [
       't/001_basic.pl',
       't/002_pg_upgrade.pl',
+      't/003_logical_replication_slots.pl',
     ],
     'test_kwargs': {'priority': 40}, # pg_upgrade tests are slow
   },
diff --git a/src/bin/pg_upgrade/option.c b/src/bin/pg_upgrade/option.c
index 8869b6b60d..f7b5ec9879 100644
--- a/src/bin/pg_upgrade/option.c
+++ b/src/bin/pg_upgrade/option.c
@@ -57,6 +57,7 @@ parseCommandLine(int argc, char *argv[])
 		{"verbose", no_argument, NULL, 'v'},
 		{"clone", no_argument, NULL, 1},
 		{"copy", no_argument, NULL, 2},
+		{"include-logical-replication-slots", no_argument, NULL, 3},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -199,6 +200,10 @@ parseCommandLine(int argc, char *argv[])
 				user_opts.transfer_mode = TRANSFER_MODE_COPY;
 				break;
 
+			case 3:
+				user_opts.include_logical_slots = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						os_info.progname);
@@ -289,6 +294,8 @@ usage(void)
 	printf(_("  -V, --version                 display version information, then exit\n"));
 	printf(_("  --clone                       clone instead of copying files to new cluster\n"));
 	printf(_("  --copy                        copy files to new cluster (default)\n"));
+	printf(_("  --include-logical-replication-slots\n"
+			 "                                upgrade logical replication slots\n"));
 	printf(_("  -?, --help                    show this help, then exit\n"));
 	printf(_("\n"
 			 "Before running pg_upgrade you must:\n"
diff --git a/src/bin/pg_upgrade/pg_upgrade.c b/src/bin/pg_upgrade/pg_upgrade.c
index 75bab0a04c..373a9ef490 100644
--- a/src/bin/pg_upgrade/pg_upgrade.c
+++ b/src/bin/pg_upgrade/pg_upgrade.c
@@ -59,6 +59,7 @@ static void copy_xact_xlog_xid(void);
 static void set_frozenxids(bool minmxid_only);
 static void make_outputdirs(char *pgdata);
 static void setup(char *argv0, bool *live_check);
+static void create_logical_replication_slots(void);
 
 ClusterInfo old_cluster,
 			new_cluster;
@@ -188,6 +189,19 @@ main(int argc, char **argv)
 			  new_cluster.pgdata);
 	check_ok();
 
+	/*
+	 * Create logical replication slots if requested.
+	 *
+	 * Note: This must be done after doing pg_resetwal command because the
+	 * command will remove required WALs.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		start_postmaster(&new_cluster, true);
+		create_logical_replication_slots();
+		stop_postmaster(false);
+	}
+
 	if (user_opts.do_sync)
 	{
 		prep_status("Sync data directory to disk");
@@ -860,3 +874,50 @@ set_frozenxids(bool minmxid_only)
 
 	check_ok();
 }
+
+/*
+ * create_logical_replication_slots()
+ *
+ * Similar to create_new_objects() but only restores logical replication slots.
+ */
+static void
+create_logical_replication_slots(void)
+{
+	int			dbnum;
+
+	prep_status_progress("Restoring logical replication slots in the new cluster");
+
+	for (dbnum = 0; dbnum < old_cluster.dbarr.ndbs; dbnum++)
+	{
+		char		slots_file_name[MAXPGPATH],
+					log_file_name[MAXPGPATH];
+		DbInfo	   *old_db = &old_cluster.dbarr.dbs[dbnum];
+
+		pg_log(PG_STATUS, "%s", old_db->db_name);
+
+		snprintf(slots_file_name, sizeof(slots_file_name),
+				 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+		snprintf(log_file_name, sizeof(log_file_name),
+				 DB_DUMP_LOG_FILE_MASK, old_db->db_oid);
+
+		parallel_exec_prog(log_file_name,
+						   NULL,
+						   "\"%s/psql\" %s --echo-queries --set ON_ERROR_STOP=on "
+						   "--no-psqlrc --dbname %s -f \"%s/%s\"",
+						   new_cluster.bindir,
+						   cluster_conn_opts(&new_cluster),
+						   old_db->db_name,
+						   log_opts.dumpdir,
+						   slots_file_name);
+	}
+
+	/* reap all children */
+	while (reap_child(true) == true)
+		;
+
+	end_progress_output();
+	check_ok();
+
+	/* update new_cluster info again */
+	get_logical_slot_infos(&new_cluster);
+}
diff --git a/src/bin/pg_upgrade/pg_upgrade.h b/src/bin/pg_upgrade/pg_upgrade.h
index 3eea0139c7..16d529668c 100644
--- a/src/bin/pg_upgrade/pg_upgrade.h
+++ b/src/bin/pg_upgrade/pg_upgrade.h
@@ -29,6 +29,7 @@
 /* contains both global db information and CREATE DATABASE commands */
 #define GLOBALS_DUMP_FILE	"pg_upgrade_dump_globals.sql"
 #define DB_DUMP_FILE_MASK	"pg_upgrade_dump_%u.custom"
+#define DB_DUMP_LOGICAL_SLOTS_FILE_MASK	"pg_upgrade_dump_%u_logical_slots.sql"
 
 /*
  * Base directories that include all the files generated internally, from the
@@ -150,6 +151,19 @@ typedef struct
 	int			nrels;
 } RelInfoArr;
 
+typedef struct
+{
+	char	   *slotname;		/* slot name */
+	char	   *plugin;			/* plugin */
+	bool		two_phase;		/* Can the slot decode 2PC? */
+} LogicalSlotInfo;
+
+typedef struct
+{
+	LogicalSlotInfo *slots;
+	int			nslots;
+} LogicalSlotInfoArr;
+
 /*
  * The following structure represents a relation mapping.
  */
@@ -176,6 +190,7 @@ typedef struct
 	char		db_tablespace[MAXPGPATH];	/* database default tablespace
 											 * path */
 	RelInfoArr	rel_arr;		/* array of all user relinfos */
+	LogicalSlotInfoArr slot_arr;	/* array of all logicalslotinfos */
 } DbInfo;
 
 /*
@@ -304,6 +319,8 @@ typedef struct
 	transferMode transfer_mode; /* copy files or link them? */
 	int			jobs;			/* number of processes/threads to use */
 	char	   *socketdir;		/* directory to use for Unix sockets */
+	bool		include_logical_slots;	/* true -> dump and restore logical
+										 * replication slots */
 } UserOpts;
 
 typedef struct
@@ -400,6 +417,7 @@ FileNameMap *gen_db_file_maps(DbInfo *old_db,
 							  DbInfo *new_db, int *nmaps, const char *old_pgdata,
 							  const char *new_pgdata);
 void		get_db_and_rel_infos(ClusterInfo *cluster);
+void		get_logical_slot_infos(ClusterInfo *cluster);
 
 /* option.c */
 
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
new file mode 100644
index 0000000000..e25bcd0142
--- /dev/null
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -0,0 +1,150 @@
+# Copyright (c) 2023, PostgreSQL Global Development Group
+
+# Tests for logical replication, especially for upgrading replication slots
+
+use strict;
+use warnings;
+
+use File::Path qw(rmtree);
+
+use PostgreSQL::Test::Cluster;
+use PostgreSQL::Test::Utils;
+use Test::More;
+
+# Can be changed to test the other modes
+my $mode = $ENV{PG_TEST_PG_UPGRADE_MODE} || '--copy';
+
+# Initialize old publisher node
+my $old_publisher = PostgreSQL::Test::Cluster->new('old_publisher');
+$old_publisher->init(allows_streaming => 'logical');
+$old_publisher->start;
+
+# Initialize subscriber node
+my $subscriber = PostgreSQL::Test::Cluster->new('subscriber');
+$subscriber->init(allows_streaming => 'logical');
+$subscriber->start;
+
+# Schema setup
+$old_publisher->safe_psql('postgres',
+	"CREATE TABLE tbl AS SELECT generate_series(1,10) AS a");
+$subscriber->safe_psql('postgres', "CREATE TABLE tbl (a int)");
+
+# Initialize new publisher node
+my $new_publisher = PostgreSQL::Test::Cluster->new('new_publisher');
+$new_publisher->init(allows_streaming => 1);
+
+my $bindir = $new_publisher->config_data('--bindir');
+
+$old_publisher->stop;
+
+# Cause a failure at the start of pg_upgrade because wal_level is replica
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
+		$mode,        '--include-logical-replication-slots',
+		'--check'
+	],
+	'run of pg_upgrade of old publisher with wrong wal_level');
+ok( -d $new_publisher->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_publisher->data_dir . "/pg_upgrade_output.d");
+
+$new_publisher->append_conf('postgresql.conf', "wal_level = 'logical'");
+$new_publisher->append_conf('postgresql.conf', "max_replication_slots = 0");
+
+# Cause a failure at the start of pg_upgrade because max_replication_slots is 0
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
+		$mode,        '--include-logical-replication-slots',
+		'--check'
+	],
+	'run of pg_upgrade of old publisher with wrong max_replication_slots');
+ok( -d $new_publisher->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_publisher->data_dir . "/pg_upgrade_output.d");
+$new_publisher->append_conf('postgresql.conf', "max_replication_slots = 10");
+
+$old_publisher->start;
+
+# Setup logical replication
+my $old_connstr = $old_publisher->connstr . ' dbname=postgres';
+$old_publisher->safe_psql('postgres',
+	"CREATE PUBLICATION pub FOR ALL TABLES");
+$subscriber->safe_psql('postgres',
+	"CREATE SUBSCRIPTION sub CONNECTION '$old_connstr' PUBLICATION pub");
+
+# Wait for initial table sync to finish
+$subscriber->wait_for_subscription_sync($old_publisher, 'sub');
+
+my $result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(10), 'check initial rows on subscriber');
+
+# Define another replication slot which allows to decode prepared transactions
+$old_publisher->safe_psql('postgres',
+	"SELECT pg_catalog.pg_create_logical_replication_slot('twophase_slot', 'pgoutput', false, true)");
+
+# Preparations for upgrading publisher
+$old_publisher->stop;
+$subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub DISABLE");
+
+# Actual run, pg_upgrade_output.d is removed at the end
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
+		$mode,        '--include-logical-replication-slots'
+	],
+	'run of pg_upgrade of old publisher');
+ok( !-d $new_publisher->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+# Check whether the replication slot is copied to new publisher
+$new_publisher->start;
+$result =
+  $new_publisher->safe_psql('postgres',
+	"SELECT slot_name, two_phase FROM pg_replication_slots");
+is($result, qq(sub|f
+twophase_slot|t), 'check the replication slot is copied to new publisher');
+
+# Change connection string and enable logical replication
+my $new_connstr = $new_publisher->connstr . ' dbname=postgres';
+
+$subscriber->safe_psql('postgres',
+	"ALTER SUBSCRIPTION sub CONNECTION '$new_connstr'");
+$subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub ENABLE");
+
+# Check whether changes on the new publisher get replicated to the subscriber
+$new_publisher->safe_psql('postgres',
+	"INSERT INTO tbl VALUES (generate_series(11, 20))");
+
+$new_publisher->wait_for_catchup('sub');
+
+$result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(20), 'check changes are shipped to subscriber');
+
+done_testing();
diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list
index b4058b88c3..5944cb34ea 100644
--- a/src/tools/pgindent/typedefs.list
+++ b/src/tools/pgindent/typedefs.list
@@ -1479,6 +1479,7 @@ LogicalRepBeginData
 LogicalRepCommitData
 LogicalRepCommitPreparedTxnData
 LogicalRepCtxStruct
+LogicalReplicationSlotInfo
 LogicalRepMode
 LogicalRepMsgType
 LogicalRepPartMapEntry
@@ -1492,6 +1493,8 @@ LogicalRepTupleData
 LogicalRepTyp
 LogicalRepWorker
 LogicalRewriteMappingData
+LogicalSlotInfo
+LogicalSlotInfoArr
 LogicalTape
 LogicalTapeSet
 LsnReadQueue
-- 
2.27.0

From b1b7b0d5eb56dda0263dab52578d9c1680ad2ccd Mon Sep 17 00:00:00 2001
From: Julien Rouhaud <julien.rouhaud@free.fr>
Date: Fri, 14 Apr 2023 13:49:09 +0800
Subject: [PATCH v9 2/4] Always persist to disk logical slots during a shutdown
 checkpoint.

It's entirely possible for a logical slot to have a confirmed_flush_lsn higher
than the last value saved on disk while not being marked as dirty.  It's
currently not a problem to lose that value during a clean shutdown / restart
cycle, but a later patch adding support for pg_upgrade of publications and
logical slots will rely on that value being properly persisted to disk.

Author: Julien Rouhaud
Reviewed-by: FIXME
Discussion: FIXME
---
 src/backend/access/transam/xlog.c |  2 +-
 src/backend/replication/slot.c    | 26 ++++++++++++++++----------
 src/include/replication/slot.h    |  2 +-
 3 files changed, 18 insertions(+), 12 deletions(-)

diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 63481d826f..07d775cf33 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7011,7 +7011,7 @@ static void
 CheckPointGuts(XLogRecPtr checkPointRedo, int flags)
 {
 	CheckPointRelationMap();
-	CheckPointReplicationSlots();
+	CheckPointReplicationSlots(flags & CHECKPOINT_IS_SHUTDOWN);
 	CheckPointSnapBuild();
 	CheckPointLogicalRewriteHeap();
 	CheckPointReplicationOrigin();
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index 8021aaa0a8..aeea6ffd1f 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -109,7 +109,8 @@ static void ReplicationSlotDropPtr(ReplicationSlot *slot);
 /* internal persistency functions */
 static void RestoreSlotFromDisk(const char *name);
 static void CreateSlotOnDisk(ReplicationSlot *slot);
-static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel);
+static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+						   bool is_shutdown);
 
 /*
  * Report shared-memory space needed by ReplicationSlotsShmemInit.
@@ -783,7 +784,7 @@ ReplicationSlotSave(void)
 	Assert(MyReplicationSlot != NULL);
 
 	sprintf(path, "pg_replslot/%s", NameStr(MyReplicationSlot->data.name));
-	SaveSlotToPath(MyReplicationSlot, path, ERROR);
+	SaveSlotToPath(MyReplicationSlot, path, ERROR, false);
 }
 
 /*
@@ -1565,11 +1566,10 @@ restart:
 /*
  * Flush all replication slots to disk.
  *
- * This needn't actually be part of a checkpoint, but it's a convenient
- * location.
+ * is_shutdown is true in case of a shutdown checkpoint.
  */
 void
-CheckPointReplicationSlots(void)
+CheckPointReplicationSlots(bool is_shutdown)
 {
 	int			i;
 
@@ -1594,7 +1594,7 @@ CheckPointReplicationSlots(void)
 
 		/* save the slot to disk, locking is handled in SaveSlotToPath() */
 		sprintf(path, "pg_replslot/%s", NameStr(s->data.name));
-		SaveSlotToPath(s, path, LOG);
+		SaveSlotToPath(s, path, LOG, is_shutdown);
 	}
 	LWLockRelease(ReplicationSlotAllocationLock);
 }
@@ -1700,7 +1700,7 @@ CreateSlotOnDisk(ReplicationSlot *slot)
 
 	/* Write the actual state file. */
 	slot->dirty = true;			/* signal that we really need to write */
-	SaveSlotToPath(slot, tmppath, ERROR);
+	SaveSlotToPath(slot, tmppath, ERROR, false);
 
 	/* Rename the directory into place. */
 	if (rename(tmppath, path) != 0)
@@ -1726,7 +1726,8 @@ CreateSlotOnDisk(ReplicationSlot *slot)
  * Shared functionality between saving and creating a replication slot.
  */
 static void
-SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
+SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+			   bool is_shutdown)
 {
 	char		tmppath[MAXPGPATH];
 	char		path[MAXPGPATH];
@@ -1740,8 +1741,13 @@ SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
 	slot->just_dirtied = false;
 	SpinLockRelease(&slot->mutex);
 
-	/* and don't do anything if there's nothing to write */
-	if (!was_dirty)
+	/*
+	 * and don't do anything if there's nothing to write, unless it's this is
+	 * called for a logical slot during a shutdown checkpoint, as we want to
+	 * persist the confirmed_flush_lsn in that case, even if that's the only
+	 * modification.
+	 */
+	if (!was_dirty && !is_shutdown && !SlotIsLogical(slot))
 		return;
 
 	LWLockAcquire(&slot->io_in_progress_lock, LW_EXCLUSIVE);
diff --git a/src/include/replication/slot.h b/src/include/replication/slot.h
index a8a89dc784..7ca37c9f70 100644
--- a/src/include/replication/slot.h
+++ b/src/include/replication/slot.h
@@ -241,7 +241,7 @@ extern void ReplicationSlotNameForTablesync(Oid suboid, Oid relid, char *syncslo
 extern void ReplicationSlotDropAtPubNode(WalReceiverConn *wrconn, char *slotname, bool missing_ok);
 
 extern void StartupReplicationSlots(void);
-extern void CheckPointReplicationSlots(void);
+extern void CheckPointReplicationSlots(bool is_shutdown);
 
 extern void CheckSlotRequirements(void);
 extern void CheckSlotPermissions(void);
-- 
2.27.0

From 1b189e20c2348aaddda568534268a0ac3d51507d Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Fri, 14 Apr 2023 08:59:03 +0000
Subject: [PATCH v9 3/4] pg_upgrade: Add check function for
 --include-logical-replication-slots option

---
 src/bin/pg_upgrade/check.c                    | 81 ++++++++++++++++++-
 .../t/003_logical_replication_slots.pl        | 38 ++++++++-
 2 files changed, 115 insertions(+), 4 deletions(-)

diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index 9674c1a2ce..294668e4dc 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -9,7 +9,10 @@
 
 #include "postgres_fe.h"
 
+#include "access/xlogrecord.h"
+#include "access/xlog_internal.h"
 #include "catalog/pg_authid_d.h"
+#include "catalog/pg_control.h"
 #include "catalog/pg_collation.h"
 #include "fe_utils/string_utils.h"
 #include "mb/pg_wchar.h"
@@ -31,7 +34,7 @@ static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
 static void check_for_parameter_settings(ClusterInfo *new_cluster);
-
+static void check_for_confirmed_flush_lsn(ClusterInfo *cluster);
 
 /*
  * fix_path_separator
@@ -105,6 +108,8 @@ check_and_dump_old_cluster(bool live_check)
 	check_for_composite_data_type_usage(&old_cluster);
 	check_for_reg_data_type_usage(&old_cluster);
 	check_for_isn_and_int8_passing_mismatch(&old_cluster);
+	if (user_opts.include_logical_slots)
+		check_for_confirmed_flush_lsn(&old_cluster);
 
 	/*
 	 * PG 16 increased the size of the 'aclitem' type, which breaks the on-disk
@@ -1436,6 +1441,10 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 	int			max_replication_slots;
 	char	   *wal_level;
 
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(new_cluster->major_version < 1600))
+		return;
+
 	prep_status("Checking for logical replication slots");
 
 	res = executeQueryOrDie(conn, "SHOW max_replication_slots;");
@@ -1459,3 +1468,73 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 
 	check_ok();
 }
+
+/*
+ * Verify that all logical replication slots consumed all WALs, except a
+ * CHECKPOINT_SHUTDOWN record.
+ */
+static void
+check_for_confirmed_flush_lsn(ClusterInfo *cluster)
+{
+	int			i,
+				ntups,
+				i_slotname;
+	bool		is_error = false;
+	PGresult   *res;
+	DbInfo	   *active_db = &cluster->dbarr.dbs[0];
+	PGconn	   *conn = connectToServer(cluster, active_db->db_name);
+
+	Assert(user_opts.include_logical_slots);
+
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(cluster->major_version < 1600))
+		return;
+
+	prep_status("Checking for logical replication slots");
+
+
+	/*
+	 * Check that all logical replication slots have reached the current WAL
+	 * position, except for the CHECKPOINT_SHUTDOWN record. Even if all WALs
+	 * are consumed before shutting down the node, the checkpointer generates
+	 * a CHECKPOINT_SHUTDOWN record at shutdown, which cannot be consumed by
+	 * any slots. Therefore, we must allow for a difference between
+	 * pg_current_wal_insert_lsn() and confirmed_flush_lsn.
+	 */
+#define SHUTDOWN_RECORD_SIZE  (SizeOfXLogRecord + \
+							   SizeOfXLogRecordDataHeaderShort + \
+							   sizeof(CheckPoint))
+
+	res = executeQueryOrDie(conn,
+							"SELECT slot_name FROM pg_catalog.pg_replication_slots "
+							"WHERE (pg_catalog.pg_current_wal_insert_lsn() - confirmed_flush_lsn) > %d "
+							"AND temporary = false AND wal_status IN ('reserved', 'extended');",
+							(int) (SizeOfXLogShortPHD + SHUTDOWN_RECORD_SIZE));
+
+#undef SHUTDOWN_RECORD_SIZE
+
+	ntups = PQntuples(res);
+	i_slotname = PQfnumber(res, "slot_name");
+
+	for (i = 0; i < ntups; i++)
+	{
+		char	   *slotname;
+
+		is_error = true;
+
+		slotname = PQgetvalue(res, i, i_slotname);
+
+		pg_log(PG_WARNING,
+			   "\nWARNING: logical replication slot \"%s\" has not consumed WALs yet.",
+			   slotname);
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	if (is_error)
+		pg_fatal("--include-logical-replication-slot requires that all "
+				 "logical replication slots consumed all the WALs");
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
index e25bcd0142..5a49ac96bb 100644
--- a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -85,6 +85,38 @@ $new_publisher->append_conf('postgresql.conf', "max_replication_slots = 10");
 
 $old_publisher->start;
 
+# Create a dummy slot on old publisher to fail the test
+$old_publisher->safe_psql('postgres',
+	"SELECT pg_create_logical_replication_slot('dropme_slot', 'pgoutput')");
+$old_publisher->safe_psql('postgres',
+	"INSERT INTO tbl VALUES (generate_series(11, 20))");
+$old_publisher->stop;
+
+# Cause a failure at the start of pg_upgrade because dropme_slot is not
+# advanced till the INSERT statement
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
+		$mode,        '--include-logical-replication-slots',
+		'--check'
+	],
+	'run of pg_upgrade of old publisher with idle replication slots');
+ok( -d $new_publisher->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_publisher->data_dir . "/pg_upgrade_output.d");
+$old_publisher->start;
+$old_publisher->safe_psql('postgres',
+	"SELECT pg_drop_replication_slot('dropme_slot')");
+
 # Setup logical replication
 my $old_connstr = $old_publisher->connstr . ' dbname=postgres';
 $old_publisher->safe_psql('postgres',
@@ -96,7 +128,7 @@ $subscriber->safe_psql('postgres',
 $subscriber->wait_for_subscription_sync($old_publisher, 'sub');
 
 my $result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
-is($result, qq(10), 'check initial rows on subscriber');
+is($result, qq(20), 'check initial rows on subscriber');
 
 # Define another replication slot which allows to decode prepared transactions
 $old_publisher->safe_psql('postgres',
@@ -140,11 +172,11 @@ $subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub ENABLE");
 
 # Check whether changes on the new publisher get replicated to the subscriber
 $new_publisher->safe_psql('postgres',
-	"INSERT INTO tbl VALUES (generate_series(11, 20))");
+	"INSERT INTO tbl VALUES (generate_series(21, 30))");
 
 $new_publisher->wait_for_catchup('sub');
 
 $result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
-is($result, qq(20), 'check changes are shipped to subscriber');
+is($result, qq(30), 'check changes are shipped to subscriber');
 
 done_testing();
-- 
2.27.0

From 4dd3cc2fcb51757b3e9a1e733e2efb334b3574e8 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Mon, 24 Apr 2023 11:03:28 +0000
Subject: [PATCH v9 4/4] Change the method used to check logical replication
 slots during the live check

When a live check is requested, there is a possibility of additional changes
occurring, which may cause the current WAL position to exceed the confirmed_flush_lsn
of the slot. As a result, we check the confirmed_flush_lsn of each logical slot
instead. This is sufficient as all the WAL records will be sent during thepublisher's
shutdown.
---
 src/bin/pg_upgrade/check.c                    | 68 ++++++++++++++++++-
 .../t/003_logical_replication_slots.pl        | 50 ++++++++++++--
 2 files changed, 111 insertions(+), 7 deletions(-)

diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index 294668e4dc..985614acab 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -35,6 +35,7 @@ static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
 static void check_for_parameter_settings(ClusterInfo *new_cluster);
 static void check_for_confirmed_flush_lsn(ClusterInfo *cluster);
+static void check_are_logical_slots_active(ClusterInfo *cluster);
 
 /*
  * fix_path_separator
@@ -109,7 +110,19 @@ check_and_dump_old_cluster(bool live_check)
 	check_for_reg_data_type_usage(&old_cluster);
 	check_for_isn_and_int8_passing_mismatch(&old_cluster);
 	if (user_opts.include_logical_slots)
-		check_for_confirmed_flush_lsn(&old_cluster);
+	{
+		/*
+		 * The method used to check logical replication slots is dependent on
+		 * the value of the live_check parameter. This change was implemented
+		 * because, during a live check, it is possible for additional changes
+		 * to occur at the old node, which could cause the current WAL position
+		 * to exceed the confirmed_flush_lsn of the slot.
+		 */
+		if (live_check)
+			check_are_logical_slots_active(&old_cluster);
+		else
+			check_for_confirmed_flush_lsn(&old_cluster);
+	}
 
 	/*
 	 * PG 16 increased the size of the 'aclitem' type, which breaks the on-disk
@@ -1469,6 +1482,59 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 	check_ok();
 }
 
+/*
+ * Verify that all logical replication slots are active
+ */
+static void
+check_are_logical_slots_active(ClusterInfo *cluster)
+{
+	int			i,
+				ntups,
+				i_slotname;
+	bool		is_error = false;
+	PGresult   *res;
+	DbInfo	   *active_db = &cluster->dbarr.dbs[0];
+	PGconn	   *conn = connectToServer(cluster, active_db->db_name);
+
+	Assert(user_opts.include_logical_slots);
+
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(cluster->major_version < 1600))
+		return;
+
+	prep_status("Checking for logical replication slots");
+
+	res = executeQueryOrDie(conn,
+							"SELECT slot_name FROM pg_catalog.pg_replication_slots "
+							"WHERE active IS FALSE "
+							"AND temporary = false AND wal_status IN ('reserved', 'extended');");
+
+	ntups = PQntuples(res);
+	i_slotname = PQfnumber(res, "slot_name");
+
+	for (i = 0; i < ntups; i++)
+	{
+		char	   *slotname;
+
+		is_error = true;
+
+		slotname = PQgetvalue(res, i, i_slotname);
+
+		pg_log(PG_WARNING,
+			   "\nWARNING: logical replication slot \"%s\" is not active.",
+			   slotname);
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	if (is_error)
+		pg_fatal("--include-logical-replication-slot with --check requires that "
+				 "all logical replication slots are active");
+
+	check_ok();
+}
+
 /*
  * Verify that all logical replication slots consumed all WALs, except a
  * CHECKPOINT_SHUTDOWN record.
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
index 5a49ac96bb..a3260e0c88 100644
--- a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -73,7 +73,6 @@ command_fails(
 		'-p',         $old_publisher->port,
 		'-P',         $new_publisher->port,
 		$mode,        '--include-logical-replication-slots',
-		'--check'
 	],
 	'run of pg_upgrade of old publisher with wrong max_replication_slots');
 ok( -d $new_publisher->data_dir . "/pg_upgrade_output.d",
@@ -105,7 +104,6 @@ command_fails(
 		'-p',         $old_publisher->port,
 		'-P',         $new_publisher->port,
 		$mode,        '--include-logical-replication-slots',
-		'--check'
 	],
 	'run of pg_upgrade of old publisher with idle replication slots');
 ok( -d $new_publisher->data_dir . "/pg_upgrade_output.d",
@@ -130,9 +128,49 @@ $subscriber->wait_for_subscription_sync($old_publisher, 'sub');
 my $result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
 is($result, qq(20), 'check initial rows on subscriber');
 
+# Start a background session and open a transaction (not committed yet)
+my $bsession = $old_publisher->background_psql('postgres');
+$bsession->query_safe(
+	q{
+BEGIN;
+INSERT INTO tbl VALUES (generate_series(21, 30))
+});
+
+$result = $old_publisher->safe_psql('postgres',
+	"SELECT count(*) FROM pg_replication_slots WHERE pg_current_wal_insert_lsn() > confirmed_flush_lsn"
+);
+is($result, qq(1),
+	'check the current WAL position exceeds confirmed_flush_lsn');
+
+
+# Run pg_upgrade --check. In the command the status of each logical slots will
+# be checked and then this will be succeeded.
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
+		$mode,        '--include-logical-replication-slots',
+		'--check'
+	],
+	'run of pg_upgrade of old publisher');
+ok( !-d $new_publisher->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+# Cleanup
+$bsession->query_safe("COMMIT");
+$bsession->quit;
+$old_publisher->wait_for_catchup('sub');
+
 # Define another replication slot which allows to decode prepared transactions
 $old_publisher->safe_psql('postgres',
-	"SELECT pg_catalog.pg_create_logical_replication_slot('twophase_slot', 'pgoutput', false, true)");
+	"SELECT pg_catalog.pg_create_logical_replication_slot('twophase_slot', 'pgoutput', false, true)"
+);
 
 # Preparations for upgrading publisher
 $old_publisher->stop;
@@ -160,7 +198,7 @@ $new_publisher->start;
 $result =
   $new_publisher->safe_psql('postgres',
 	"SELECT slot_name, two_phase FROM pg_replication_slots");
-is($result, qq(sub|f
+is( $result, qq(sub|f
 twophase_slot|t), 'check the replication slot is copied to new publisher');
 
 # Change connection string and enable logical replication
@@ -172,11 +210,11 @@ $subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub ENABLE");
 
 # Check whether changes on the new publisher get replicated to the subscriber
 $new_publisher->safe_psql('postgres',
-	"INSERT INTO tbl VALUES (generate_series(21, 30))");
+	"INSERT INTO tbl VALUES (generate_series(31, 40))");
 
 $new_publisher->wait_for_catchup('sub');
 
 $result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
-is($result, qq(30), 'check changes are shipped to subscriber');
+is($result, qq(40), 'check changes are shipped to subscriber');
 
 done_testing();
-- 
2.27.0

From eef7163f5b6b22043d4f8a9396df53965477554e Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Tue, 4 Apr 2023 05:49:34 +0000
Subject: [PATCH v10 1/4] pg_upgrade: Add --include-logical-replication-slots
 option

This commit introduces a new pg_upgrade option called "--include-logical-replication-slots".
This allows nodes with logical replication slots to be upgraded. The commit can
be divided into two parts: one for pg_dump and another for pg_upgrade.

For pg_dump this commit includes a new option called "--logical-replication-slots-only".
This option can be used to dump logical replication slots. When this option is
specified, the slot_name, plugin, and two_phase parameters are extracted from
pg_replication_slots. An SQL file is then generated which executes
pg_create_logical_replication_slot() with the extracted parameters.

For pg_upgrade, when '--include-logical-replication-slots' is specified, it executes
pg_dump with the new "--logical-replication-slots-only" option and restores from the
dump. Apart from restoring schema, pg_resetwal must not be called after restoring
replication slots. This is because the command discards WAL files and starts from a
new segment, even if they are required by replication slots. This leads to an ERROR:
"requested WAL segment XXX has already been removed". To avoid this, replication slots
are restored at a different time than other objects, after running pg_resetwal.

The significant advantage of this commit is that it makes it easy to continue
logical replication even after upgrading the publisher node. Previously, pg_upgrade
allowed copying publications to a new node. With this new commit, adjusting the
connection string to the new publisher will cause the apply worker on the subscriber
to connect to the new publisher automatically. This enables seamless continuation
of logical replication, even after an upgrade.

Author: Hayato Kuroda
Reviewed-by: Peter Smith, Julien Rouhaud, Vignesh C
---
 doc/src/sgml/ref/pg_dump.sgml                 |  10 ++
 doc/src/sgml/ref/pgupgrade.sgml               |  11 ++
 src/bin/pg_dump/pg_backup.h                   |   1 +
 src/bin/pg_dump/pg_dump.c                     | 150 ++++++++++++++++++
 src/bin/pg_dump/pg_dump.h                     |  17 +-
 src/bin/pg_dump/pg_dump_sort.c                |  11 +-
 src/bin/pg_upgrade/check.c                    |  57 +++++++
 src/bin/pg_upgrade/dump.c                     |  24 +++
 src/bin/pg_upgrade/info.c                     | 125 ++++++++++++++-
 src/bin/pg_upgrade/meson.build                |   1 +
 src/bin/pg_upgrade/option.c                   |   7 +
 src/bin/pg_upgrade/pg_upgrade.c               |  61 +++++++
 src/bin/pg_upgrade/pg_upgrade.h               |  18 +++
 .../t/003_logical_replication_slots.pl        | 109 +++++++++++++
 src/tools/pgindent/typedefs.list              |   3 +
 15 files changed, 601 insertions(+), 4 deletions(-)
 create mode 100644 src/bin/pg_upgrade/t/003_logical_replication_slots.pl

diff --git a/doc/src/sgml/ref/pg_dump.sgml b/doc/src/sgml/ref/pg_dump.sgml
index e81e35c13b..6e07f85281 100644
--- a/doc/src/sgml/ref/pg_dump.sgml
+++ b/doc/src/sgml/ref/pg_dump.sgml
@@ -1206,6 +1206,16 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--logical-replication-slots-only</option></term>
+      <listitem>
+       <para>
+        Dump only logical replication slots; not the schema (data definitions),
+        nor data. This is mainly used when upgrading nodes.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
        <term><option>-?</option></term>
        <term><option>--help</option></term>
diff --git a/doc/src/sgml/ref/pgupgrade.sgml b/doc/src/sgml/ref/pgupgrade.sgml
index 7816b4c685..94e90ff506 100644
--- a/doc/src/sgml/ref/pgupgrade.sgml
+++ b/doc/src/sgml/ref/pgupgrade.sgml
@@ -240,6 +240,17 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--include-logical-replication-slots</option></term>
+      <listitem>
+       <para>
+        Upgrade logical replication slots. Only permanent replication slots
+        are included. Note that pg_upgrade does not check the installation of
+        plugins.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
       <term><option>-?</option></term>
       <term><option>--help</option></term>
diff --git a/src/bin/pg_dump/pg_backup.h b/src/bin/pg_dump/pg_backup.h
index aba780ef4b..0a4e931f9b 100644
--- a/src/bin/pg_dump/pg_backup.h
+++ b/src/bin/pg_dump/pg_backup.h
@@ -187,6 +187,7 @@ typedef struct _dumpOptions
 	int			use_setsessauth;
 	int			enable_row_security;
 	int			load_via_partition_root;
+	int			logical_slots_only;
 
 	/* default, if no "inclusion" switches appear, is to dump everything */
 	bool		include_everything;
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index 058244cd17..ed98572cce 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -328,6 +328,9 @@ static void setupDumpWorker(Archive *AH);
 static TableInfo *getRootTableInfo(const TableInfo *tbinfo);
 static bool forcePartitionRootLoad(const TableInfo *tbinfo);
 
+static void getLogicalReplicationSlots(Archive *fout);
+static void dumpLogicalReplicationSlot(Archive *fout,
+									   const LogicalReplicationSlotInfo *slotinfo);
 
 int
 main(int argc, char **argv)
@@ -431,6 +434,7 @@ main(int argc, char **argv)
 		{"table-and-children", required_argument, NULL, 12},
 		{"exclude-table-and-children", required_argument, NULL, 13},
 		{"exclude-table-data-and-children", required_argument, NULL, 14},
+		{"logical-replication-slots-only", no_argument, NULL, 15},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -657,6 +661,10 @@ main(int argc, char **argv)
 										  optarg);
 				break;
 
+			case 15:			/* dump only replication slot(s) */
+				dopt.logical_slots_only = true;
+				break;
+
 			default:
 				/* getopt_long already emitted a complaint */
 				pg_log_error_hint("Try \"%s --help\" for more information.", progname);
@@ -714,6 +722,14 @@ main(int argc, char **argv)
 	if (dopt.do_nothing && dopt.dump_inserts == 0)
 		pg_fatal("option --on-conflict-do-nothing requires option --inserts, --rows-per-insert, or --column-inserts");
 
+	if (dopt.logical_slots_only && !dopt.binary_upgrade)
+		pg_fatal("options --logical-replication-slots-only requires option --binary-upgrade");
+
+	if (dopt.logical_slots_only && dopt.dataOnly)
+		pg_fatal("options --logical-replication-slots-only and -a/--data-only cannot be used together");
+	if (dopt.logical_slots_only && dopt.schemaOnly)
+		pg_fatal("options --logical-replication-slots-only and -s/--schema-only cannot be used together");
+
 	/* Identify archive format to emit */
 	archiveFormat = parseArchiveFormat(format, &archiveMode);
 
@@ -876,6 +892,16 @@ main(int argc, char **argv)
 			pg_fatal("no matching extensions were found");
 	}
 
+	/*
+	 * If dump logical-replication-slots-only was requested, dump only them
+	 * and skip everything else.
+	 */
+	if (dopt.logical_slots_only)
+	{
+		getLogicalReplicationSlots(fout);
+		goto dump;
+	}
+
 	/*
 	 * Dumping LOs is the default for dumps where an inclusion switch is not
 	 * used (an "include everything" dump).  -B can be used to exclude LOs
@@ -936,6 +962,8 @@ main(int argc, char **argv)
 	if (!dopt.no_security_labels)
 		collectSecLabels(fout);
 
+dump:
+
 	/* Lastly, create dummy objects to represent the section boundaries */
 	boundaryObjs = createBoundaryObjects();
 
@@ -1109,6 +1137,8 @@ help(const char *progname)
 			 "                               servers matching PATTERN\n"));
 	printf(_("  --inserts                    dump data as INSERT commands, rather than COPY\n"));
 	printf(_("  --load-via-partition-root    load partitions via the root table\n"));
+	printf(_("  --logical-replication-slots-only\n"
+			 "                               dump only logical replication slots, no schema or data\n"));
 	printf(_("  --no-comments                do not dump comments\n"));
 	printf(_("  --no-publications            do not dump publications\n"));
 	printf(_("  --no-security-labels         do not dump security label assignments\n"));
@@ -10252,6 +10282,10 @@ dumpDumpableObject(Archive *fout, DumpableObject *dobj)
 		case DO_SUBSCRIPTION:
 			dumpSubscription(fout, (const SubscriptionInfo *) dobj);
 			break;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			dumpLogicalReplicationSlot(fout,
+									   (const LogicalReplicationSlotInfo *) dobj);
+			break;
 		case DO_PRE_DATA_BOUNDARY:
 		case DO_POST_DATA_BOUNDARY:
 			/* never dumped, nothing to do */
@@ -18227,6 +18261,7 @@ addBoundaryDependencies(DumpableObject **dobjs, int numObjs,
 			case DO_PUBLICATION_REL:
 			case DO_PUBLICATION_TABLE_IN_SCHEMA:
 			case DO_SUBSCRIPTION:
+			case DO_LOGICAL_REPLICATION_SLOT:
 				/* Post-data objects: must come after the post-data boundary */
 				addObjectDependency(dobj, postDataBound->dumpId);
 				break;
@@ -18488,3 +18523,118 @@ appendReloptionsArrayAH(PQExpBuffer buffer, const char *reloptions,
 	if (!res)
 		pg_log_warning("could not parse %s array", "reloptions");
 }
+
+/*
+ * getLogicalReplicationSlots
+ *	  get information about replication slots
+ */
+static void
+getLogicalReplicationSlots(Archive *fout)
+{
+	PGresult   *res;
+	LogicalReplicationSlotInfo *slotinfo;
+	PQExpBuffer query;
+	DumpOptions *dopt = fout->dopt;
+
+	int			i_slotname;
+	int			i_plugin;
+	int			i_twophase;
+	int			i,
+				ntups;
+
+	/* Check whether we should dump or not */
+	if (fout->remoteVersion < 160000 || !dopt->logical_slots_only)
+		return;
+
+	query = createPQExpBuffer();
+
+	resetPQExpBuffer(query);
+
+	/*
+	 * Get replication slots.
+	 *
+	 * XXX: Which information must be extracted from old node? Currently three
+	 * attributes are extracted because they are used by
+	 * pg_create_logical_replication_slot().
+	 *
+	 * XXX: Do we have to support physical slots?
+	 */
+	appendPQExpBufferStr(query,
+						 "SELECT slot_name, plugin, two_phase "
+						 "FROM pg_catalog.pg_replication_slots "
+						 "WHERE database = current_database() AND temporary = false "
+						 "AND wal_status IN ('reserved', 'extended');");
+
+	res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
+
+	ntups = PQntuples(res);
+
+	i_slotname = PQfnumber(res, "slot_name");
+	i_plugin = PQfnumber(res, "plugin");
+	i_twophase = PQfnumber(res, "two_phase");
+
+	slotinfo = pg_malloc(ntups * sizeof(LogicalReplicationSlotInfo));
+
+	for (i = 0; i < ntups; i++)
+	{
+		slotinfo[i].dobj.objType = DO_LOGICAL_REPLICATION_SLOT;
+
+		slotinfo[i].dobj.catId.tableoid = InvalidOid;
+		slotinfo[i].dobj.catId.oid = InvalidOid;
+		AssignDumpId(&slotinfo[i].dobj);
+
+		slotinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_slotname));
+
+		slotinfo[i].plugin = pg_strdup(PQgetvalue(res, i, i_plugin));
+		slotinfo[i].twophase = (strcmp(PQgetvalue(res, i, i_twophase), "t") == 0);
+
+		/*
+		 * Note: Currently we do not have any options to include/exclude slots
+		 * in dumping, so all the slots must be selected.
+		 */
+		slotinfo[i].dobj.dump = DUMP_COMPONENT_ALL;
+	}
+	PQclear(res);
+
+	destroyPQExpBuffer(query);
+}
+
+/*
+ * dumpLogicalReplicationSlot
+ *	  dump creation functions for the given logical replication slots
+ */
+static void
+dumpLogicalReplicationSlot(Archive *fout,
+						   const LogicalReplicationSlotInfo *slotinfo)
+{
+	DumpOptions *dopt = fout->dopt;
+
+	if (!dopt->logical_slots_only)
+		return;
+
+	if (slotinfo->dobj.dump & DUMP_COMPONENT_DEFINITION)
+	{
+		PQExpBuffer query = createPQExpBuffer();
+		char	   *slotname = pg_strdup(slotinfo->dobj.name);
+
+		/*
+		 * XXX: For simplification, pg_create_logical_replication_slot() is
+		 * used. Is it sufficient?
+		 */
+		appendPQExpBuffer(query, "SELECT pg_catalog.pg_create_logical_replication_slot(");
+		appendStringLiteralAH(query, slotname, fout);
+		appendPQExpBuffer(query, ", ");
+		appendStringLiteralAH(query, slotinfo->plugin, fout);
+		appendPQExpBuffer(query, ", false, %s);",
+						  slotinfo->twophase ? "true" : "false");
+
+		ArchiveEntry(fout, slotinfo->dobj.catId, slotinfo->dobj.dumpId,
+					 ARCHIVE_OPTS(.tag = slotname,
+								  .description = "REPLICATION SLOT",
+								  .section = SECTION_POST_DATA,
+								  .createStmt = query->data));
+
+		free(slotname);
+		destroyPQExpBuffer(query);
+	}
+}
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index ed6ce41ad7..de081c35ae 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -82,7 +82,8 @@ typedef enum
 	DO_PUBLICATION,
 	DO_PUBLICATION_REL,
 	DO_PUBLICATION_TABLE_IN_SCHEMA,
-	DO_SUBSCRIPTION
+	DO_SUBSCRIPTION,
+	DO_LOGICAL_REPLICATION_SLOT
 } DumpableObjectType;
 
 /*
@@ -666,6 +667,20 @@ typedef struct _SubscriptionInfo
 	char	   *subpasswordrequired;
 } SubscriptionInfo;
 
+/*
+ * The LogicalReplicationSlotInfo struct is used to represent replication
+ * slots.
+ *
+ * XXX: add more attributes if needed
+ */
+typedef struct _LogicalReplicationSlotInfo
+{
+	DumpableObject dobj;
+	char	   *plugin;
+	char	   *slottype;
+	bool		twophase;
+} LogicalReplicationSlotInfo;
+
 /*
  *	common utility functions
  */
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index 745578d855..4e12e46dc5 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -93,6 +93,7 @@ enum dbObjectTypePriorities
 	PRIO_PUBLICATION_REL,
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,
 	PRIO_SUBSCRIPTION,
+	PRIO_LOGICAL_REPLICATION_SLOT,
 	PRIO_DEFAULT_ACL,			/* done in ACL pass */
 	PRIO_EVENT_TRIGGER,			/* must be next to last! */
 	PRIO_REFRESH_MATVIEW		/* must be last! */
@@ -146,10 +147,11 @@ static const int dbObjectTypePriority[] =
 	PRIO_PUBLICATION,			/* DO_PUBLICATION */
 	PRIO_PUBLICATION_REL,		/* DO_PUBLICATION_REL */
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,	/* DO_PUBLICATION_TABLE_IN_SCHEMA */
-	PRIO_SUBSCRIPTION			/* DO_SUBSCRIPTION */
+	PRIO_SUBSCRIPTION,			/* DO_SUBSCRIPTION */
+	PRIO_LOGICAL_REPLICATION_SLOT	/* DO_LOGICAL_REPLICATION_SLOT */
 };
 
-StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_SUBSCRIPTION + 1),
+StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_LOGICAL_REPLICATION_SLOT + 1),
 				 "array length mismatch");
 
 static DumpId preDataBoundId;
@@ -1498,6 +1500,11 @@ describeDumpableObject(DumpableObject *obj, char *buf, int bufsize)
 					 "SUBSCRIPTION (ID %d OID %u)",
 					 obj->dumpId, obj->catId.oid);
 			return;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			snprintf(buf, bufsize,
+					 "LOGICAL REPLICATION SLOT (ID %d NAME %s)",
+					 obj->dumpId, obj->name);
+			return;
 		case DO_PRE_DATA_BOUNDARY:
 			snprintf(buf, bufsize,
 					 "PRE-DATA BOUNDARY  (ID %d)",
diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index fea159689e..9674c1a2ce 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -30,6 +30,7 @@ static void check_for_jsonb_9_4_usage(ClusterInfo *cluster);
 static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
+static void check_for_parameter_settings(ClusterInfo *new_cluster);
 
 
 /*
@@ -88,6 +89,7 @@ check_and_dump_old_cluster(bool live_check)
 
 	/* Extract a list of databases and tables from the old cluster */
 	get_db_and_rel_infos(&old_cluster);
+	get_logical_slot_infos(&old_cluster);
 
 	init_tablespaces();
 
@@ -188,6 +190,7 @@ void
 check_new_cluster(void)
 {
 	get_db_and_rel_infos(&new_cluster);
+	get_logical_slot_infos(&new_cluster);
 
 	check_new_cluster_is_empty();
 
@@ -210,6 +213,9 @@ check_new_cluster(void)
 	check_for_prepared_transactions(&new_cluster);
 
 	check_for_new_tablespace_dir(&new_cluster);
+
+	if (user_opts.include_logical_slots)
+		check_for_parameter_settings(&new_cluster);
 }
 
 
@@ -364,6 +370,22 @@ check_new_cluster_is_empty(void)
 						 rel_arr->rels[relnum].nspname,
 						 rel_arr->rels[relnum].relname);
 		}
+
+		/*
+		 * If --include-logical-replication-slots is required, check the
+		 * existing of slots
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			LogicalSlotInfoArr *slot_arr = &new_cluster.dbarr.dbs[dbnum].slot_arr;
+
+			/* if nslots > 0, report just first entry and exit */
+			if (slot_arr->nslots)
+				pg_fatal("New cluster database \"%s\" is not empty: found logical replication slot \"%s\"",
+						 new_cluster.dbarr.dbs[dbnum].db_name,
+						 slot_arr->slots[0].slotname);
+		}
+
 	}
 }
 
@@ -1402,3 +1424,38 @@ check_for_user_defined_encoding_conversions(ClusterInfo *cluster)
 	else
 		check_ok();
 }
+
+/*
+ * Verify parameter settings for creating logical replication slots
+ */
+static void
+check_for_parameter_settings(ClusterInfo *new_cluster)
+{
+	PGresult   *res;
+	PGconn	   *conn = connectToServer(new_cluster, "template1");
+	int			max_replication_slots;
+	char	   *wal_level;
+
+	prep_status("Checking for logical replication slots");
+
+	res = executeQueryOrDie(conn, "SHOW max_replication_slots;");
+	max_replication_slots = atoi(PQgetvalue(res, 0, 0));
+
+	if (max_replication_slots == 0)
+		pg_fatal("max_replication_slots must be greater than 0");
+
+	PQclear(res);
+
+	res = executeQueryOrDie(conn, "SHOW wal_level;");
+	wal_level = PQgetvalue(res, 0, 0);
+
+	if (strcmp(wal_level, "logical") != 0)
+		pg_fatal("wal_level must be \"logical\", but set to \"%s\"",
+				 wal_level);
+
+	PQclear(res);
+
+	PQfinish(conn);
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/dump.c b/src/bin/pg_upgrade/dump.c
index 6c8c82dca8..e6b90864f5 100644
--- a/src/bin/pg_upgrade/dump.c
+++ b/src/bin/pg_upgrade/dump.c
@@ -59,6 +59,30 @@ generate_old_dump(void)
 						   log_opts.dumpdir,
 						   sql_file_name, escaped_connstr.data);
 
+		/*
+		 * Dump logical replication slots if needed.
+		 *
+		 * XXX We cannot dump replication slots at the same time as the schema
+		 * dump because we need to separate the timing of restoring
+		 * replication slots and other objects. Replication slots, in
+		 * particular, should not be restored before executing the pg_resetwal
+		 * command because it will remove WALs that are required by the slots.
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			char		slots_file_name[MAXPGPATH];
+
+			snprintf(slots_file_name, sizeof(slots_file_name),
+					 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+			parallel_exec_prog(log_file_name, NULL,
+							   "\"%s/pg_dump\" %s --logical-replication-slots-only "
+							   "--quote-all-identifiers --binary-upgrade %s "
+							   "--file=\"%s/%s\" %s",
+							   new_cluster.bindir, cluster_conn_opts(&old_cluster),
+							   log_opts.verbose ? "--verbose" : "",
+							   log_opts.dumpdir,
+							   slots_file_name, escaped_connstr.data);
+		}
 		termPQExpBuffer(&escaped_connstr);
 	}
 
diff --git a/src/bin/pg_upgrade/info.c b/src/bin/pg_upgrade/info.c
index 85ed15ae4a..8cc2e3b4ae 100644
--- a/src/bin/pg_upgrade/info.c
+++ b/src/bin/pg_upgrade/info.c
@@ -23,10 +23,12 @@ static void free_db_and_rel_infos(DbInfoArr *db_arr);
 static void get_template0_info(ClusterInfo *cluster);
 static void get_db_infos(ClusterInfo *cluster);
 static void get_rel_infos(ClusterInfo *cluster, DbInfo *dbinfo);
+static void get_logical_slot_infos_per_db(ClusterInfo *cluster, DbInfo *dbinfo);
 static void free_rel_infos(RelInfoArr *rel_arr);
+static void free_logical_slot_infos(LogicalSlotInfoArr *slot_arr);
 static void print_db_infos(DbInfoArr *db_arr);
 static void print_rel_infos(RelInfoArr *rel_arr);
-
+static void print_slot_infos(LogicalSlotInfoArr *slot_arr);
 
 /*
  * gen_db_file_maps()
@@ -283,8 +285,17 @@ get_db_and_rel_infos(ClusterInfo *cluster)
 	get_db_infos(cluster);
 
 	for (dbnum = 0; dbnum < cluster->dbarr.ndbs; dbnum++)
+	{
 		get_rel_infos(cluster, &cluster->dbarr.dbs[dbnum]);
 
+		/*
+		 * Additionally, slot_arr must be initialized because they will be
+		 * checked later.
+		 */
+		cluster->dbarr.dbs[dbnum].slot_arr.nslots = 0;
+		cluster->dbarr.dbs[dbnum].slot_arr.slots = NULL;
+	}
+
 	if (cluster == &old_cluster)
 		pg_log(PG_VERBOSE, "\nsource databases:");
 	else
@@ -600,6 +611,93 @@ get_rel_infos(ClusterInfo *cluster, DbInfo *dbinfo)
 	dbinfo->rel_arr.nrels = num_rels;
 }
 
+/*
+ * Higher level routine to generate LogicalSlotInfoArr for all databases.
+ */
+void
+get_logical_slot_infos(ClusterInfo *cluster)
+{
+	int			dbnum;
+
+	for (dbnum = 0; dbnum < cluster->dbarr.ndbs; dbnum++)
+	{
+		if (cluster->dbarr.dbs[dbnum].slot_arr.slots)
+			free_logical_slot_infos(&cluster->dbarr.dbs[dbnum].slot_arr);
+
+		get_logical_slot_infos_per_db(cluster, &cluster->dbarr.dbs[dbnum]);
+	}
+
+	if (cluster == &old_cluster)
+		pg_log(PG_VERBOSE, "\nsource databases:");
+	else
+		pg_log(PG_VERBOSE, "\ntarget databases:");
+
+	if (log_opts.verbose)
+	{
+		for (dbnum = 0; dbnum < cluster->dbarr.ndbs; dbnum++)
+		{
+			pg_log(PG_VERBOSE, "Database: %s", cluster->dbarr.dbs[dbnum].db_name);
+			print_slot_infos(&cluster->dbarr.dbs[dbnum].slot_arr);
+		}
+	}
+}
+
+/*
+ * get_logical_slot_infos_per_db()
+ *
+ * gets the LogicalSlotInfos for all the logical replication slots of the database
+ * referred to by "dbinfo".
+ */
+static void
+get_logical_slot_infos_per_db(ClusterInfo *cluster, DbInfo *dbinfo)
+{
+	PGconn	   *conn = connectToServer(cluster,
+									   dbinfo->db_name);
+	PGresult   *res;
+	LogicalSlotInfo *slotinfos;
+
+	int			ntups;
+	int			slotnum;
+	int			num_slots = 0;
+
+	int			i_slotname;
+	int			i_plugin;
+	int			i_twophase;
+
+	char		query[QUERY_ALLOC];
+
+	query[0] = '\0';			/* initialize query string to empty */
+
+	snprintf(query + strlen(query), sizeof(query) - strlen(query),
+			 "SELECT slot_name, plugin, two_phase "
+			 "FROM pg_catalog.pg_replication_slots "
+			 "WHERE database = current_database() AND temporary = false "
+			 "AND wal_status IN ('reserved', 'extended');");
+
+	res = executeQueryOrDie(conn, "%s", query);
+
+	ntups = PQntuples(res);
+
+	slotinfos = (LogicalSlotInfo *) pg_malloc(sizeof(LogicalSlotInfo) * ntups);
+
+	i_slotname = PQfnumber(res, "slot_name");
+	i_plugin = PQfnumber(res, "plugin");
+	i_twophase = PQfnumber(res, "two_phase");
+
+	for (slotnum = 0; slotnum < ntups; slotnum++)
+	{
+		LogicalSlotInfo *curr = &slotinfos[num_slots++];
+
+		curr->slotname = pg_strdup(PQgetvalue(res, slotnum, i_slotname));
+		curr->plugin = pg_strdup(PQgetvalue(res, slotnum, i_plugin));
+		curr->two_phase = (strcmp(PQgetvalue(res, slotnum, i_twophase), "t") == 0);
+	}
+
+	PQfinish(conn);
+
+	dbinfo->slot_arr.slots = slotinfos;
+	dbinfo->slot_arr.nslots = num_slots;
+}
 
 static void
 free_db_and_rel_infos(DbInfoArr *db_arr)
@@ -634,6 +732,19 @@ free_rel_infos(RelInfoArr *rel_arr)
 	rel_arr->nrels = 0;
 }
 
+static void
+free_logical_slot_infos(LogicalSlotInfoArr *slot_arr)
+{
+	int			slotnum;
+
+	for (slotnum = 0; slotnum < slot_arr->nslots; slotnum++)
+	{
+		pg_free(slot_arr->slots[slotnum].slotname);
+		pg_free(slot_arr->slots[slotnum].plugin);
+	}
+	pg_free(slot_arr->slots);
+	slot_arr->nslots = 0;
+}
 
 static void
 print_db_infos(DbInfoArr *db_arr)
@@ -660,3 +771,15 @@ print_rel_infos(RelInfoArr *rel_arr)
 			   rel_arr->rels[relnum].reloid,
 			   rel_arr->rels[relnum].tablespace);
 }
+
+static void
+print_slot_infos(LogicalSlotInfoArr *slot_arr)
+{
+	int			slotnum;
+
+	for (slotnum = 0; slotnum < slot_arr->nslots; slotnum++)
+		pg_log(PG_VERBOSE, "slotname: %s: plugin: %s: two_phase %d",
+			   slot_arr->slots[slotnum].slotname,
+			   slot_arr->slots[slotnum].plugin,
+			   slot_arr->slots[slotnum].two_phase);
+}
diff --git a/src/bin/pg_upgrade/meson.build b/src/bin/pg_upgrade/meson.build
index 12a97f84e2..228f29b688 100644
--- a/src/bin/pg_upgrade/meson.build
+++ b/src/bin/pg_upgrade/meson.build
@@ -42,6 +42,7 @@ tests += {
     'tests': [
       't/001_basic.pl',
       't/002_pg_upgrade.pl',
+      't/003_logical_replication_slots.pl',
     ],
     'test_kwargs': {'priority': 40}, # pg_upgrade tests are slow
   },
diff --git a/src/bin/pg_upgrade/option.c b/src/bin/pg_upgrade/option.c
index 8869b6b60d..f7b5ec9879 100644
--- a/src/bin/pg_upgrade/option.c
+++ b/src/bin/pg_upgrade/option.c
@@ -57,6 +57,7 @@ parseCommandLine(int argc, char *argv[])
 		{"verbose", no_argument, NULL, 'v'},
 		{"clone", no_argument, NULL, 1},
 		{"copy", no_argument, NULL, 2},
+		{"include-logical-replication-slots", no_argument, NULL, 3},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -199,6 +200,10 @@ parseCommandLine(int argc, char *argv[])
 				user_opts.transfer_mode = TRANSFER_MODE_COPY;
 				break;
 
+			case 3:
+				user_opts.include_logical_slots = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						os_info.progname);
@@ -289,6 +294,8 @@ usage(void)
 	printf(_("  -V, --version                 display version information, then exit\n"));
 	printf(_("  --clone                       clone instead of copying files to new cluster\n"));
 	printf(_("  --copy                        copy files to new cluster (default)\n"));
+	printf(_("  --include-logical-replication-slots\n"
+			 "                                upgrade logical replication slots\n"));
 	printf(_("  -?, --help                    show this help, then exit\n"));
 	printf(_("\n"
 			 "Before running pg_upgrade you must:\n"
diff --git a/src/bin/pg_upgrade/pg_upgrade.c b/src/bin/pg_upgrade/pg_upgrade.c
index 75bab0a04c..373a9ef490 100644
--- a/src/bin/pg_upgrade/pg_upgrade.c
+++ b/src/bin/pg_upgrade/pg_upgrade.c
@@ -59,6 +59,7 @@ static void copy_xact_xlog_xid(void);
 static void set_frozenxids(bool minmxid_only);
 static void make_outputdirs(char *pgdata);
 static void setup(char *argv0, bool *live_check);
+static void create_logical_replication_slots(void);
 
 ClusterInfo old_cluster,
 			new_cluster;
@@ -188,6 +189,19 @@ main(int argc, char **argv)
 			  new_cluster.pgdata);
 	check_ok();
 
+	/*
+	 * Create logical replication slots if requested.
+	 *
+	 * Note: This must be done after doing pg_resetwal command because the
+	 * command will remove required WALs.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		start_postmaster(&new_cluster, true);
+		create_logical_replication_slots();
+		stop_postmaster(false);
+	}
+
 	if (user_opts.do_sync)
 	{
 		prep_status("Sync data directory to disk");
@@ -860,3 +874,50 @@ set_frozenxids(bool minmxid_only)
 
 	check_ok();
 }
+
+/*
+ * create_logical_replication_slots()
+ *
+ * Similar to create_new_objects() but only restores logical replication slots.
+ */
+static void
+create_logical_replication_slots(void)
+{
+	int			dbnum;
+
+	prep_status_progress("Restoring logical replication slots in the new cluster");
+
+	for (dbnum = 0; dbnum < old_cluster.dbarr.ndbs; dbnum++)
+	{
+		char		slots_file_name[MAXPGPATH],
+					log_file_name[MAXPGPATH];
+		DbInfo	   *old_db = &old_cluster.dbarr.dbs[dbnum];
+
+		pg_log(PG_STATUS, "%s", old_db->db_name);
+
+		snprintf(slots_file_name, sizeof(slots_file_name),
+				 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+		snprintf(log_file_name, sizeof(log_file_name),
+				 DB_DUMP_LOG_FILE_MASK, old_db->db_oid);
+
+		parallel_exec_prog(log_file_name,
+						   NULL,
+						   "\"%s/psql\" %s --echo-queries --set ON_ERROR_STOP=on "
+						   "--no-psqlrc --dbname %s -f \"%s/%s\"",
+						   new_cluster.bindir,
+						   cluster_conn_opts(&new_cluster),
+						   old_db->db_name,
+						   log_opts.dumpdir,
+						   slots_file_name);
+	}
+
+	/* reap all children */
+	while (reap_child(true) == true)
+		;
+
+	end_progress_output();
+	check_ok();
+
+	/* update new_cluster info again */
+	get_logical_slot_infos(&new_cluster);
+}
diff --git a/src/bin/pg_upgrade/pg_upgrade.h b/src/bin/pg_upgrade/pg_upgrade.h
index 3eea0139c7..16d529668c 100644
--- a/src/bin/pg_upgrade/pg_upgrade.h
+++ b/src/bin/pg_upgrade/pg_upgrade.h
@@ -29,6 +29,7 @@
 /* contains both global db information and CREATE DATABASE commands */
 #define GLOBALS_DUMP_FILE	"pg_upgrade_dump_globals.sql"
 #define DB_DUMP_FILE_MASK	"pg_upgrade_dump_%u.custom"
+#define DB_DUMP_LOGICAL_SLOTS_FILE_MASK	"pg_upgrade_dump_%u_logical_slots.sql"
 
 /*
  * Base directories that include all the files generated internally, from the
@@ -150,6 +151,19 @@ typedef struct
 	int			nrels;
 } RelInfoArr;
 
+typedef struct
+{
+	char	   *slotname;		/* slot name */
+	char	   *plugin;			/* plugin */
+	bool		two_phase;		/* Can the slot decode 2PC? */
+} LogicalSlotInfo;
+
+typedef struct
+{
+	LogicalSlotInfo *slots;
+	int			nslots;
+} LogicalSlotInfoArr;
+
 /*
  * The following structure represents a relation mapping.
  */
@@ -176,6 +190,7 @@ typedef struct
 	char		db_tablespace[MAXPGPATH];	/* database default tablespace
 											 * path */
 	RelInfoArr	rel_arr;		/* array of all user relinfos */
+	LogicalSlotInfoArr slot_arr;	/* array of all logicalslotinfos */
 } DbInfo;
 
 /*
@@ -304,6 +319,8 @@ typedef struct
 	transferMode transfer_mode; /* copy files or link them? */
 	int			jobs;			/* number of processes/threads to use */
 	char	   *socketdir;		/* directory to use for Unix sockets */
+	bool		include_logical_slots;	/* true -> dump and restore logical
+										 * replication slots */
 } UserOpts;
 
 typedef struct
@@ -400,6 +417,7 @@ FileNameMap *gen_db_file_maps(DbInfo *old_db,
 							  DbInfo *new_db, int *nmaps, const char *old_pgdata,
 							  const char *new_pgdata);
 void		get_db_and_rel_infos(ClusterInfo *cluster);
+void		get_logical_slot_infos(ClusterInfo *cluster);
 
 /* option.c */
 
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
new file mode 100644
index 0000000000..6689a91fcc
--- /dev/null
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -0,0 +1,109 @@
+# Copyright (c) 2023, PostgreSQL Global Development Group
+
+# Tests for upgrading replication slots
+
+use strict;
+use warnings;
+
+use File::Path qw(rmtree);
+
+use PostgreSQL::Test::Cluster;
+use PostgreSQL::Test::Utils;
+use Test::More;
+
+# Can be changed to test the other modes
+my $mode = $ENV{PG_TEST_PG_UPGRADE_MODE} || '--copy';
+
+# Initialize old node
+my $old_node = PostgreSQL::Test::Cluster->new('old_node');
+$old_node->init(allows_streaming => 'logical');
+$old_node->start;
+
+# Initialize new node
+my $new_node = PostgreSQL::Test::Cluster->new('new_node');
+$new_node->init(allows_streaming => 1);
+
+my $bindir = $new_node->config_data('--bindir');
+
+$old_node->stop;
+
+# Cause a failure at the start of pg_upgrade because wal_level is replica
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with wrong wal_level');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+$new_node->append_conf('postgresql.conf', "wal_level = 'logical'");
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 0");
+
+# Cause a failure at the start of pg_upgrade because max_replication_slots is 0
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with wrong max_replication_slots');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 10");
+
+# Create a slot on old node, and generate WALs
+$old_node->start;
+$old_node->safe_psql(
+	'postgres', qq[
+	SELECT pg_create_logical_replication_slot('test_slot', 'test_decoding', false, true);
+	CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a;
+]);
+
+my $result = $old_node->safe_psql('postgres',
+	"SELECT count (*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)"
+);
+is($result, qq(12), 'ensure WALs are not consumed yet');
+$old_node->stop;
+
+# Actual run, pg_upgrade_output.d is removed at the end
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots'
+	],
+	'run of pg_upgrade of old node');
+ok( !-d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+$new_node->start;
+$result = $new_node->safe_psql('postgres',
+	"SELECT slot_name, two_phase FROM pg_replication_slots");
+is($result, qq(test_slot|t), 'check the slot exists on new node');
+
+done_testing();
diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list
index b4058b88c3..5944cb34ea 100644
--- a/src/tools/pgindent/typedefs.list
+++ b/src/tools/pgindent/typedefs.list
@@ -1479,6 +1479,7 @@ LogicalRepBeginData
 LogicalRepCommitData
 LogicalRepCommitPreparedTxnData
 LogicalRepCtxStruct
+LogicalReplicationSlotInfo
 LogicalRepMode
 LogicalRepMsgType
 LogicalRepPartMapEntry
@@ -1492,6 +1493,8 @@ LogicalRepTupleData
 LogicalRepTyp
 LogicalRepWorker
 LogicalRewriteMappingData
+LogicalSlotInfo
+LogicalSlotInfoArr
 LogicalTape
 LogicalTapeSet
 LsnReadQueue
-- 
2.27.0

From ff0fb35aa1ce896f1a5a786733a98f81538c5254 Mon Sep 17 00:00:00 2001
From: Julien Rouhaud <julien.rouhaud@free.fr>
Date: Fri, 14 Apr 2023 13:49:09 +0800
Subject: [PATCH v10 2/4] Always persist to disk logical slots during a
 shutdown checkpoint.

It's entirely possible for a logical slot to have a confirmed_flush_lsn higher
than the last value saved on disk while not being marked as dirty.  It's
currently not a problem to lose that value during a clean shutdown / restart
cycle, but a later patch adding support for pg_upgrade of publications and
logical slots will rely on that value being properly persisted to disk.

Author: Julien Rouhaud
Reviewed-by: FIXME
Discussion: FIXME
---
 src/backend/access/transam/xlog.c |  2 +-
 src/backend/replication/slot.c    | 26 ++++++++++++++++----------
 src/include/replication/slot.h    |  2 +-
 3 files changed, 18 insertions(+), 12 deletions(-)

diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 63481d826f..07d775cf33 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7011,7 +7011,7 @@ static void
 CheckPointGuts(XLogRecPtr checkPointRedo, int flags)
 {
 	CheckPointRelationMap();
-	CheckPointReplicationSlots();
+	CheckPointReplicationSlots(flags & CHECKPOINT_IS_SHUTDOWN);
 	CheckPointSnapBuild();
 	CheckPointLogicalRewriteHeap();
 	CheckPointReplicationOrigin();
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index 8021aaa0a8..aeea6ffd1f 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -109,7 +109,8 @@ static void ReplicationSlotDropPtr(ReplicationSlot *slot);
 /* internal persistency functions */
 static void RestoreSlotFromDisk(const char *name);
 static void CreateSlotOnDisk(ReplicationSlot *slot);
-static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel);
+static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+						   bool is_shutdown);
 
 /*
  * Report shared-memory space needed by ReplicationSlotsShmemInit.
@@ -783,7 +784,7 @@ ReplicationSlotSave(void)
 	Assert(MyReplicationSlot != NULL);
 
 	sprintf(path, "pg_replslot/%s", NameStr(MyReplicationSlot->data.name));
-	SaveSlotToPath(MyReplicationSlot, path, ERROR);
+	SaveSlotToPath(MyReplicationSlot, path, ERROR, false);
 }
 
 /*
@@ -1565,11 +1566,10 @@ restart:
 /*
  * Flush all replication slots to disk.
  *
- * This needn't actually be part of a checkpoint, but it's a convenient
- * location.
+ * is_shutdown is true in case of a shutdown checkpoint.
  */
 void
-CheckPointReplicationSlots(void)
+CheckPointReplicationSlots(bool is_shutdown)
 {
 	int			i;
 
@@ -1594,7 +1594,7 @@ CheckPointReplicationSlots(void)
 
 		/* save the slot to disk, locking is handled in SaveSlotToPath() */
 		sprintf(path, "pg_replslot/%s", NameStr(s->data.name));
-		SaveSlotToPath(s, path, LOG);
+		SaveSlotToPath(s, path, LOG, is_shutdown);
 	}
 	LWLockRelease(ReplicationSlotAllocationLock);
 }
@@ -1700,7 +1700,7 @@ CreateSlotOnDisk(ReplicationSlot *slot)
 
 	/* Write the actual state file. */
 	slot->dirty = true;			/* signal that we really need to write */
-	SaveSlotToPath(slot, tmppath, ERROR);
+	SaveSlotToPath(slot, tmppath, ERROR, false);
 
 	/* Rename the directory into place. */
 	if (rename(tmppath, path) != 0)
@@ -1726,7 +1726,8 @@ CreateSlotOnDisk(ReplicationSlot *slot)
  * Shared functionality between saving and creating a replication slot.
  */
 static void
-SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
+SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+			   bool is_shutdown)
 {
 	char		tmppath[MAXPGPATH];
 	char		path[MAXPGPATH];
@@ -1740,8 +1741,13 @@ SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
 	slot->just_dirtied = false;
 	SpinLockRelease(&slot->mutex);
 
-	/* and don't do anything if there's nothing to write */
-	if (!was_dirty)
+	/*
+	 * and don't do anything if there's nothing to write, unless it's this is
+	 * called for a logical slot during a shutdown checkpoint, as we want to
+	 * persist the confirmed_flush_lsn in that case, even if that's the only
+	 * modification.
+	 */
+	if (!was_dirty && !is_shutdown && !SlotIsLogical(slot))
 		return;
 
 	LWLockAcquire(&slot->io_in_progress_lock, LW_EXCLUSIVE);
diff --git a/src/include/replication/slot.h b/src/include/replication/slot.h
index a8a89dc784..7ca37c9f70 100644
--- a/src/include/replication/slot.h
+++ b/src/include/replication/slot.h
@@ -241,7 +241,7 @@ extern void ReplicationSlotNameForTablesync(Oid suboid, Oid relid, char *syncslo
 extern void ReplicationSlotDropAtPubNode(WalReceiverConn *wrconn, char *slotname, bool missing_ok);
 
 extern void StartupReplicationSlots(void);
-extern void CheckPointReplicationSlots(void);
+extern void CheckPointReplicationSlots(bool is_shutdown);
 
 extern void CheckSlotRequirements(void);
 extern void CheckSlotPermissions(void);
-- 
2.27.0

From d6cc01a8ad04536243f6e3072c0c13f8a3ca4ebf Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Fri, 14 Apr 2023 08:59:03 +0000
Subject: [PATCH v10 3/4] pg_upgrade: Add check function for
 --include-logical-replication-slots option

---
 src/bin/pg_upgrade/check.c                    | 80 ++++++++++++++++++-
 .../t/003_logical_replication_slots.pl        | 30 ++++++-
 2 files changed, 108 insertions(+), 2 deletions(-)

diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index 9674c1a2ce..946431f059 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -9,7 +9,10 @@
 
 #include "postgres_fe.h"
 
+#include "access/xlogrecord.h"
+#include "access/xlog_internal.h"
 #include "catalog/pg_authid_d.h"
+#include "catalog/pg_control.h"
 #include "catalog/pg_collation.h"
 #include "fe_utils/string_utils.h"
 #include "mb/pg_wchar.h"
@@ -31,7 +34,7 @@ static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
 static void check_for_parameter_settings(ClusterInfo *new_cluster);
-
+static void check_for_confirmed_flush_lsn(ClusterInfo *cluster);
 
 /*
  * fix_path_separator
@@ -105,6 +108,8 @@ check_and_dump_old_cluster(bool live_check)
 	check_for_composite_data_type_usage(&old_cluster);
 	check_for_reg_data_type_usage(&old_cluster);
 	check_for_isn_and_int8_passing_mismatch(&old_cluster);
+	if (user_opts.include_logical_slots)
+		check_for_confirmed_flush_lsn(&old_cluster);
 
 	/*
 	 * PG 16 increased the size of the 'aclitem' type, which breaks the on-disk
@@ -1436,6 +1441,10 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 	int			max_replication_slots;
 	char	   *wal_level;
 
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(new_cluster->major_version < 1600))
+		return;
+
 	prep_status("Checking for logical replication slots");
 
 	res = executeQueryOrDie(conn, "SHOW max_replication_slots;");
@@ -1459,3 +1468,72 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 
 	check_ok();
 }
+
+/*
+ * Verify that all logical replication slots consumed all WALs, except a
+ * CHECKPOINT_SHUTDOWN record.
+ */
+static void
+check_for_confirmed_flush_lsn(ClusterInfo *cluster)
+{
+	int			i,
+				ntups,
+				i_slotname;
+	bool		is_error = false;
+	PGresult   *res;
+	DbInfo	   *active_db = &cluster->dbarr.dbs[0];
+	PGconn	   *conn = connectToServer(cluster, active_db->db_name);
+
+	Assert(user_opts.include_logical_slots);
+
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(cluster->major_version < 1600))
+		return;
+
+	prep_status("Checking for logical replication slots");
+
+	/*
+	 * Check that all logical replication slots have reached the current WAL
+	 * position, except for the CHECKPOINT_SHUTDOWN record. Even if all WALs
+	 * are consumed before shutting down the node, the checkpointer generates
+	 * a CHECKPOINT_SHUTDOWN record at shutdown, which cannot be consumed by
+	 * any slots. Therefore, we must allow for a difference between
+	 * pg_current_wal_insert_lsn() and confirmed_flush_lsn.
+	 */
+#define SHUTDOWN_RECORD_SIZE  (SizeOfXLogRecord + \
+							   SizeOfXLogRecordDataHeaderShort + \
+							   sizeof(CheckPoint))
+
+	res = executeQueryOrDie(conn,
+							"SELECT slot_name FROM pg_catalog.pg_replication_slots "
+							"WHERE (pg_catalog.pg_current_wal_insert_lsn() - confirmed_flush_lsn) > %d "
+							"AND temporary = false AND wal_status IN ('reserved', 'extended');",
+							(int) (SizeOfXLogLongPHD + SHUTDOWN_RECORD_SIZE));
+
+#undef SHUTDOWN_RECORD_SIZE
+
+	ntups = PQntuples(res);
+	i_slotname = PQfnumber(res, "slot_name");
+
+	for (i = 0; i < ntups; i++)
+	{
+		char	   *slotname;
+
+		is_error = true;
+
+		slotname = PQgetvalue(res, i, i_slotname);
+
+		pg_log(PG_WARNING,
+			   "\nWARNING: logical replication slot \"%s\" has not consumed WALs yet",
+			   slotname);
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	if (is_error)
+		pg_fatal("--include-logical-replication-slots requires that all "
+				 "logical replication slots consumed all the WALs");
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
index 6689a91fcc..378e2bfb6e 100644
--- a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -79,11 +79,39 @@ $old_node->safe_psql(
 ]);
 
 my $result = $old_node->safe_psql('postgres',
-	"SELECT count (*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)"
+	"SELECT count(*) FROM pg_logical_slot_peek_changes('test_slot', NULL, NULL)"
 );
+
 is($result, qq(12), 'ensure WALs are not consumed yet');
 $old_node->stop;
 
+# Cause a failure at the start of pg_upgrade because test_slot does not
+# finish consuming all the WALs
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with idle replication slots');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+$old_node->start;
+$old_node->safe_psql('postgres',
+	"SELECT count (*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)"
+);
+$old_node->stop;
+
 # Actual run, pg_upgrade_output.d is removed at the end
 command_ok(
 	[
-- 
2.27.0

From 077ab6ebe26e604873bc48a11990cecad6badf11 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Mon, 24 Apr 2023 11:03:28 +0000
Subject: [PATCH v10 4/4] Change the method used to check logical replication
 slots during the live check

When a live check is requested, there is a possibility of additional changes
occurring, which may cause the current WAL position to exceed the confirmed_flush_lsn
of the slot. As a result, we check the confirmed_flush_lsn of each logical slot
instead. This is sufficient as all the WAL records will be sent during thepublisher's
shutdown.
---
 src/bin/pg_upgrade/check.c                    | 68 ++++++++++++++++++-
 .../t/003_logical_replication_slots.pl        | 66 +++++++++++++++++-
 2 files changed, 130 insertions(+), 4 deletions(-)

diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index 946431f059..247f8cb5f3 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -35,6 +35,7 @@ static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
 static void check_for_parameter_settings(ClusterInfo *new_cluster);
 static void check_for_confirmed_flush_lsn(ClusterInfo *cluster);
+static void check_are_logical_slots_active(ClusterInfo *cluster);
 
 /*
  * fix_path_separator
@@ -109,7 +110,19 @@ check_and_dump_old_cluster(bool live_check)
 	check_for_reg_data_type_usage(&old_cluster);
 	check_for_isn_and_int8_passing_mismatch(&old_cluster);
 	if (user_opts.include_logical_slots)
-		check_for_confirmed_flush_lsn(&old_cluster);
+	{
+		/*
+		 * The method used to check logical replication slots is dependent on
+		 * the value of the live_check parameter. This change was implemented
+		 * because, during a live check, it is possible for additional changes
+		 * to occur at the old node, which could cause the current WAL position
+		 * to exceed the confirmed_flush_lsn of the slot.
+		 */
+		if (live_check)
+			check_are_logical_slots_active(&old_cluster);
+		else
+			check_for_confirmed_flush_lsn(&old_cluster);
+	}
 
 	/*
 	 * PG 16 increased the size of the 'aclitem' type, which breaks the on-disk
@@ -1469,6 +1482,59 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 	check_ok();
 }
 
+/*
+ * Verify that all logical replication slots are active
+ */
+static void
+check_are_logical_slots_active(ClusterInfo *cluster)
+{
+	int			i,
+				ntups,
+				i_slotname;
+	bool		is_error = false;
+	PGresult   *res;
+	DbInfo	   *active_db = &cluster->dbarr.dbs[0];
+	PGconn	   *conn = connectToServer(cluster, active_db->db_name);
+
+	Assert(user_opts.include_logical_slots);
+
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(cluster->major_version < 1600))
+		return;
+
+	prep_status("Checking for logical replication slots");
+
+	res = executeQueryOrDie(conn,
+							"SELECT slot_name FROM pg_catalog.pg_replication_slots "
+							"WHERE active IS FALSE "
+							"AND temporary = false AND wal_status IN ('reserved', 'extended');");
+
+	ntups = PQntuples(res);
+	i_slotname = PQfnumber(res, "slot_name");
+
+	for (i = 0; i < ntups; i++)
+	{
+		char	   *slotname;
+
+		is_error = true;
+
+		slotname = PQgetvalue(res, i, i_slotname);
+
+		pg_log(PG_WARNING,
+			   "\nWARNING: logical replication slot \"%s\" is not active",
+			   slotname);
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	if (is_error)
+		pg_fatal("--include-logical-replication-slots with --check requires that "
+				 "all logical replication slots are active");
+
+	check_ok();
+}
+
 /*
  * Verify that all logical replication slots consumed all WALs, except a
  * CHECKPOINT_SHUTDOWN record.
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
index 378e2bfb6e..aa52445b08 100644
--- a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -19,6 +19,10 @@ my $old_node = PostgreSQL::Test::Cluster->new('old_node');
 $old_node->init(allows_streaming => 'logical');
 $old_node->start;
 
+# Initialize subscriber, which will be used only for --check
+my $subscriber = PostgreSQL::Test::Cluster->new('subscriber');
+$subscriber->init(allows_streaming => 'logical');
+
 # Initialize new node
 my $new_node = PostgreSQL::Test::Cluster->new('new_node');
 $new_node->init(allows_streaming => 1);
@@ -70,15 +74,71 @@ ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
 rmtree($new_node->data_dir . "/pg_upgrade_output.d");
 $new_node->append_conf('postgresql.conf', "max_replication_slots = 10");
 
-# Create a slot on old node, and generate WALs
+# Setup logical replication
 $old_node->start;
+$old_node->safe_psql('postgres',
+	"CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a");
+$old_node->safe_psql('postgres', "CREATE PUBLICATION pub FOR ALL TABLES");
+my $old_connstr = $old_node->connstr . ' dbname=postgres';
+
+$subscriber->start;
+$subscriber->safe_psql('postgres', "CREATE TABLE tbl (a int)");
+$subscriber->safe_psql('postgres',
+	"CREATE SUBSCRIPTION sub CONNECTION '$old_connstr' PUBLICATION pub WITH (copy_data = true)"
+);
+
+# Wait for initial table sync to finish
+$subscriber->wait_for_subscription_sync($old_node, 'sub');
+
+my $result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(10), 'check initial rows on subscriber');
+
+# Start a background session and open a transaction (not committed yet)
+my $bsession = $old_node->background_psql('postgres');
+$bsession->query_safe(
+	q{
+BEGIN;
+INSERT INTO tbl VALUES (generate_series(11, 20))
+});
+
+$result = $old_node->safe_psql('postgres',
+	"SELECT count(*) FROM pg_replication_slots WHERE pg_current_wal_insert_lsn() > confirmed_flush_lsn"
+);
+is($result, qq(1),
+	'check the current WAL position exceeds confirmed_flush_lsn');
+
+# Run pg_upgrade --check. In the command the status of each logical slots will
+# be checked and then this will be succeeded.
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+		'--check'
+	],
+	'run of pg_upgrade of old node');
+ok( !-d $old_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+# Cleanup
+$bsession->query_safe("ABORT");
+$bsession->quit;
+$subscriber->safe_psql('postgres', "DROP SUBSCRIPTION sub");
+
+# Create a slot on old node, and generate WALs
 $old_node->safe_psql(
 	'postgres', qq[
 	SELECT pg_create_logical_replication_slot('test_slot', 'test_decoding', false, true);
-	CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a;
+	INSERT INTO tbl VALUES (generate_series(11, 20));
 ]);
 
-my $result = $old_node->safe_psql('postgres',
+$result = $old_node->safe_psql('postgres',
 	"SELECT count(*) FROM pg_logical_slot_peek_changes('test_slot', NULL, NULL)"
 );
 
-- 
2.27.0

From c27ef8c9dc6288cb5833e4d4d4f36d0f9af9464d Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Tue, 4 Apr 2023 05:49:34 +0000
Subject: [PATCH v11 1/4] pg_upgrade: Add --include-logical-replication-slots
 option

This commit introduces a new pg_upgrade option called "--include-logical-replication-slots".
This allows nodes with logical replication slots to be upgraded. The commit can
be divided into two parts: one for pg_dump and another for pg_upgrade.

For pg_dump this commit includes a new option called "--logical-replication-slots-only".
This option can be used to dump logical replication slots. When this option is
specified, the slot_name, plugin, and two_phase parameters are extracted from
pg_replication_slots. An SQL file is then generated which executes
pg_create_logical_replication_slot() with the extracted parameters.

For pg_upgrade, when '--include-logical-replication-slots' is specified, it executes
pg_dump with the new "--logical-replication-slots-only" option and restores from the
dump. Note that we cannot dump replication slots at the same time as the schema
dump because we need to separate the timing of restoring replication slots and
other objects. Replication slots, in  particular, should not be restored before
executing the pg_resetwal command because it will remove WALs that are required
by the slots.

The significant advantage of this commit is that it makes it easy to continue
logical replication even after upgrading the publisher node. Previously, pg_upgrade
allowed copying publications to a new node. With this new commit, adjusting the
connection string to the new publisher will cause the apply worker on the subscriber
to connect to the new publisher automatically. This enables seamless continuation
of logical replication, even after an upgrade.

Author: Hayato Kuroda
Reviewed-by: Peter Smith, Julien Rouhaud, Vignesh C
---
 doc/src/sgml/ref/pgupgrade.sgml               |  11 ++
 src/bin/pg_dump/pg_backup.h                   |   1 +
 src/bin/pg_dump/pg_dump.c                     | 150 ++++++++++++++++++
 src/bin/pg_dump/pg_dump.h                     |  17 +-
 src/bin/pg_dump/pg_dump_sort.c                |  11 +-
 src/bin/pg_upgrade/Makefile                   |   3 +
 src/bin/pg_upgrade/check.c                    |  60 +++++++
 src/bin/pg_upgrade/dump.c                     |  24 +++
 src/bin/pg_upgrade/info.c                     | 119 +++++++++++++-
 src/bin/pg_upgrade/meson.build                |   1 +
 src/bin/pg_upgrade/option.c                   |   7 +
 src/bin/pg_upgrade/pg_upgrade.c               |  61 +++++++
 src/bin/pg_upgrade/pg_upgrade.h               |  21 +++
 .../t/003_logical_replication_slots.pl        | 115 ++++++++++++++
 src/tools/pgindent/typedefs.list              |   3 +
 15 files changed, 599 insertions(+), 5 deletions(-)
 create mode 100644 src/bin/pg_upgrade/t/003_logical_replication_slots.pl

diff --git a/doc/src/sgml/ref/pgupgrade.sgml b/doc/src/sgml/ref/pgupgrade.sgml
index 7816b4c685..94e90ff506 100644
--- a/doc/src/sgml/ref/pgupgrade.sgml
+++ b/doc/src/sgml/ref/pgupgrade.sgml
@@ -240,6 +240,17 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--include-logical-replication-slots</option></term>
+      <listitem>
+       <para>
+        Upgrade logical replication slots. Only permanent replication slots
+        are included. Note that pg_upgrade does not check the installation of
+        plugins.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
       <term><option>-?</option></term>
       <term><option>--help</option></term>
diff --git a/src/bin/pg_dump/pg_backup.h b/src/bin/pg_dump/pg_backup.h
index aba780ef4b..0a4e931f9b 100644
--- a/src/bin/pg_dump/pg_backup.h
+++ b/src/bin/pg_dump/pg_backup.h
@@ -187,6 +187,7 @@ typedef struct _dumpOptions
 	int			use_setsessauth;
 	int			enable_row_security;
 	int			load_via_partition_root;
+	int			logical_slots_only;
 
 	/* default, if no "inclusion" switches appear, is to dump everything */
 	bool		include_everything;
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index 41a51ec5cd..906f9a9541 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -328,6 +328,9 @@ static void setupDumpWorker(Archive *AH);
 static TableInfo *getRootTableInfo(const TableInfo *tbinfo);
 static bool forcePartitionRootLoad(const TableInfo *tbinfo);
 
+static void getLogicalReplicationSlots(Archive *fout);
+static void dumpLogicalReplicationSlot(Archive *fout,
+									   const LogicalReplicationSlotInfo *slotinfo);
 
 int
 main(int argc, char **argv)
@@ -431,6 +434,7 @@ main(int argc, char **argv)
 		{"table-and-children", required_argument, NULL, 12},
 		{"exclude-table-and-children", required_argument, NULL, 13},
 		{"exclude-table-data-and-children", required_argument, NULL, 14},
+		{"logical-replication-slots-only", no_argument, NULL, 15},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -657,6 +661,10 @@ main(int argc, char **argv)
 										  optarg);
 				break;
 
+			case 15:			/* dump only replication slot(s) */
+				dopt.logical_slots_only = true;
+				break;
+
 			default:
 				/* getopt_long already emitted a complaint */
 				pg_log_error_hint("Try \"%s --help\" for more information.", progname);
@@ -714,6 +722,18 @@ main(int argc, char **argv)
 	if (dopt.do_nothing && dopt.dump_inserts == 0)
 		pg_fatal("option --on-conflict-do-nothing requires option --inserts, --rows-per-insert, or --column-inserts");
 
+	if (dopt.logical_slots_only)
+	{
+		if (!dopt.binary_upgrade)
+			pg_fatal("options --logical-replication-slots-only requires option --binary-upgrade");
+
+		if (dopt.dataOnly)
+			pg_fatal("options --logical-replication-slots-only and -a/--data-only cannot be used together");
+
+		if (dopt.schemaOnly)
+			pg_fatal("options --logical-replication-slots-only and -s/--schema-only cannot be used together");
+	}
+
 	/* Identify archive format to emit */
 	archiveFormat = parseArchiveFormat(format, &archiveMode);
 
@@ -876,6 +896,16 @@ main(int argc, char **argv)
 			pg_fatal("no matching extensions were found");
 	}
 
+	/*
+	 * If dump logical-replication-slots-only was requested, dump only them
+	 * and skip everything else.
+	 */
+	if (dopt.logical_slots_only)
+	{
+		getLogicalReplicationSlots(fout);
+		goto dump;
+	}
+
 	/*
 	 * Dumping LOs is the default for dumps where an inclusion switch is not
 	 * used (an "include everything" dump).  -B can be used to exclude LOs
@@ -936,6 +966,8 @@ main(int argc, char **argv)
 	if (!dopt.no_security_labels)
 		collectSecLabels(fout);
 
+dump:
+
 	/* Lastly, create dummy objects to represent the section boundaries */
 	boundaryObjs = createBoundaryObjects();
 
@@ -1109,6 +1141,8 @@ help(const char *progname)
 			 "                               servers matching PATTERN\n"));
 	printf(_("  --inserts                    dump data as INSERT commands, rather than COPY\n"));
 	printf(_("  --load-via-partition-root    load partitions via the root table\n"));
+	printf(_("  --logical-replication-slots-only\n"
+			 "                               dump only logical replication slots, no schema or data\n"));
 	printf(_("  --no-comments                do not dump comments\n"));
 	printf(_("  --no-publications            do not dump publications\n"));
 	printf(_("  --no-security-labels         do not dump security label assignments\n"));
@@ -10252,6 +10286,10 @@ dumpDumpableObject(Archive *fout, DumpableObject *dobj)
 		case DO_SUBSCRIPTION:
 			dumpSubscription(fout, (const SubscriptionInfo *) dobj);
 			break;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			dumpLogicalReplicationSlot(fout,
+									   (const LogicalReplicationSlotInfo *) dobj);
+			break;
 		case DO_PRE_DATA_BOUNDARY:
 		case DO_POST_DATA_BOUNDARY:
 			/* never dumped, nothing to do */
@@ -18227,6 +18265,7 @@ addBoundaryDependencies(DumpableObject **dobjs, int numObjs,
 			case DO_PUBLICATION_REL:
 			case DO_PUBLICATION_TABLE_IN_SCHEMA:
 			case DO_SUBSCRIPTION:
+			case DO_LOGICAL_REPLICATION_SLOT:
 				/* Post-data objects: must come after the post-data boundary */
 				addObjectDependency(dobj, postDataBound->dumpId);
 				break;
@@ -18488,3 +18527,114 @@ appendReloptionsArrayAH(PQExpBuffer buffer, const char *reloptions,
 	if (!res)
 		pg_log_warning("could not parse %s array", "reloptions");
 }
+
+/*
+ * getLogicalReplicationSlots
+ *	  get information about replication slots
+ */
+static void
+getLogicalReplicationSlots(Archive *fout)
+{
+	PGresult   *res;
+	LogicalReplicationSlotInfo *slotinfo;
+	PQExpBuffer query;
+
+	int			i_slotname;
+	int			i_plugin;
+	int			i_twophase;
+	int			i,
+				ntups;
+
+	/* Check whether we should dump or not */
+	if (fout->remoteVersion < 160000)
+		return;
+
+	Assert(fout->dopt->logical_slots_only);
+
+	query = createPQExpBuffer();
+
+	resetPQExpBuffer(query);
+
+	/*
+	 * Get replication slots.
+	 *
+	 * XXX: Which information must be extracted from old node? Currently three
+	 * attributes are extracted because they are used by
+	 * pg_create_logical_replication_slot().
+	 *
+	 * XXX: Do we have to support physical slots?
+	 */
+	appendPQExpBufferStr(query,
+						 "SELECT slot_name, plugin, two_phase "
+						 "FROM pg_catalog.pg_replication_slots "
+						 "WHERE database = current_database() AND temporary = false "
+						 "AND wal_status IN ('reserved', 'extended');");
+
+	res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
+
+	ntups = PQntuples(res);
+
+	i_slotname = PQfnumber(res, "slot_name");
+	i_plugin = PQfnumber(res, "plugin");
+	i_twophase = PQfnumber(res, "two_phase");
+
+	slotinfo = pg_malloc(ntups * sizeof(LogicalReplicationSlotInfo));
+
+	for (i = 0; i < ntups; i++)
+	{
+		slotinfo[i].dobj.objType = DO_LOGICAL_REPLICATION_SLOT;
+
+		slotinfo[i].dobj.catId.tableoid = InvalidOid;
+		slotinfo[i].dobj.catId.oid = InvalidOid;
+		AssignDumpId(&slotinfo[i].dobj);
+
+		slotinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_slotname));
+
+		slotinfo[i].plugin = pg_strdup(PQgetvalue(res, i, i_plugin));
+		slotinfo[i].twophase = (strcmp(PQgetvalue(res, i, i_twophase), "t") == 0);
+
+		/*
+		 * Note: Currently we do not have any options to include/exclude slots
+		 * in dumping, so all the slots must be selected.
+		 */
+		slotinfo[i].dobj.dump = DUMP_COMPONENT_ALL;
+	}
+	PQclear(res);
+
+	destroyPQExpBuffer(query);
+}
+
+/*
+ * dumpLogicalReplicationSlot
+ *	  dump creation functions for the given logical replication slots
+ */
+static void
+dumpLogicalReplicationSlot(Archive *fout,
+						   const LogicalReplicationSlotInfo *slotinfo)
+{
+	Assert(fout->dopt->logical_slots_only);
+
+	if (slotinfo->dobj.dump & DUMP_COMPONENT_DEFINITION)
+	{
+		PQExpBuffer query = createPQExpBuffer();
+
+		/*
+		 * XXX: For simplification, pg_create_logical_replication_slot() is
+		 * used. Is it sufficient?
+		 */
+		appendPQExpBuffer(query, "SELECT pg_catalog.pg_create_logical_replication_slot(");
+		appendStringLiteralAH(query, slotinfo->dobj.name, fout);
+		appendPQExpBuffer(query, ", ");
+		appendStringLiteralAH(query, slotinfo->plugin, fout);
+		appendPQExpBuffer(query, ", false, %s);",
+						  slotinfo->twophase ? "true" : "false");
+
+		ArchiveEntry(fout, slotinfo->dobj.catId, slotinfo->dobj.dumpId,
+					 ARCHIVE_OPTS(.tag = slotinfo->dobj.name,
+								  .description = "REPLICATION SLOT",
+								  .section = SECTION_POST_DATA,
+								  .createStmt = query->data));
+
+		destroyPQExpBuffer(query);
+	}
+}
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index ed6ce41ad7..de081c35ae 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -82,7 +82,8 @@ typedef enum
 	DO_PUBLICATION,
 	DO_PUBLICATION_REL,
 	DO_PUBLICATION_TABLE_IN_SCHEMA,
-	DO_SUBSCRIPTION
+	DO_SUBSCRIPTION,
+	DO_LOGICAL_REPLICATION_SLOT
 } DumpableObjectType;
 
 /*
@@ -666,6 +667,20 @@ typedef struct _SubscriptionInfo
 	char	   *subpasswordrequired;
 } SubscriptionInfo;
 
+/*
+ * The LogicalReplicationSlotInfo struct is used to represent replication
+ * slots.
+ *
+ * XXX: add more attributes if needed
+ */
+typedef struct _LogicalReplicationSlotInfo
+{
+	DumpableObject dobj;
+	char	   *plugin;
+	char	   *slottype;
+	bool		twophase;
+} LogicalReplicationSlotInfo;
+
 /*
  *	common utility functions
  */
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index 745578d855..4e12e46dc5 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -93,6 +93,7 @@ enum dbObjectTypePriorities
 	PRIO_PUBLICATION_REL,
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,
 	PRIO_SUBSCRIPTION,
+	PRIO_LOGICAL_REPLICATION_SLOT,
 	PRIO_DEFAULT_ACL,			/* done in ACL pass */
 	PRIO_EVENT_TRIGGER,			/* must be next to last! */
 	PRIO_REFRESH_MATVIEW		/* must be last! */
@@ -146,10 +147,11 @@ static const int dbObjectTypePriority[] =
 	PRIO_PUBLICATION,			/* DO_PUBLICATION */
 	PRIO_PUBLICATION_REL,		/* DO_PUBLICATION_REL */
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,	/* DO_PUBLICATION_TABLE_IN_SCHEMA */
-	PRIO_SUBSCRIPTION			/* DO_SUBSCRIPTION */
+	PRIO_SUBSCRIPTION,			/* DO_SUBSCRIPTION */
+	PRIO_LOGICAL_REPLICATION_SLOT	/* DO_LOGICAL_REPLICATION_SLOT */
 };
 
-StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_SUBSCRIPTION + 1),
+StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_LOGICAL_REPLICATION_SLOT + 1),
 				 "array length mismatch");
 
 static DumpId preDataBoundId;
@@ -1498,6 +1500,11 @@ describeDumpableObject(DumpableObject *obj, char *buf, int bufsize)
 					 "SUBSCRIPTION (ID %d OID %u)",
 					 obj->dumpId, obj->catId.oid);
 			return;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			snprintf(buf, bufsize,
+					 "LOGICAL REPLICATION SLOT (ID %d NAME %s)",
+					 obj->dumpId, obj->name);
+			return;
 		case DO_PRE_DATA_BOUNDARY:
 			snprintf(buf, bufsize,
 					 "PRE-DATA BOUNDARY  (ID %d)",
diff --git a/src/bin/pg_upgrade/Makefile b/src/bin/pg_upgrade/Makefile
index 5834513add..815d1a7ca1 100644
--- a/src/bin/pg_upgrade/Makefile
+++ b/src/bin/pg_upgrade/Makefile
@@ -3,6 +3,9 @@
 PGFILEDESC = "pg_upgrade - an in-place binary upgrade utility"
 PGAPPICON = win32
 
+# required for 003_logical_replication_slots.pl
+EXTRA_INSTALL=contrib/test_decoding
+
 subdir = src/bin/pg_upgrade
 top_builddir = ../../..
 include $(top_builddir)/src/Makefile.global
diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index fea159689e..1802a30fe6 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -30,6 +30,7 @@ static void check_for_jsonb_9_4_usage(ClusterInfo *cluster);
 static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
+static void check_for_parameter_settings(ClusterInfo *new_cluster);
 
 
 /*
@@ -89,6 +90,10 @@ check_and_dump_old_cluster(bool live_check)
 	/* Extract a list of databases and tables from the old cluster */
 	get_db_and_rel_infos(&old_cluster);
 
+	/* Additionally, extract a list of logical replication slots if required */
+	if (user_opts.include_logical_slots)
+		get_logical_slot_infos(&old_cluster);
+
 	init_tablespaces();
 
 	get_loadable_libraries();
@@ -188,6 +193,7 @@ void
 check_new_cluster(void)
 {
 	get_db_and_rel_infos(&new_cluster);
+	get_logical_slot_infos(&new_cluster);
 
 	check_new_cluster_is_empty();
 
@@ -210,6 +216,9 @@ check_new_cluster(void)
 	check_for_prepared_transactions(&new_cluster);
 
 	check_for_new_tablespace_dir(&new_cluster);
+
+	if (user_opts.include_logical_slots)
+		check_for_parameter_settings(&new_cluster);
 }
 
 
@@ -364,6 +373,22 @@ check_new_cluster_is_empty(void)
 						 rel_arr->rels[relnum].nspname,
 						 rel_arr->rels[relnum].relname);
 		}
+
+		/*
+		 * If --include-logical-replication-slots is required, check the
+		 * existence of slots
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			LogicalSlotInfoArr *slot_arr = &new_cluster.dbarr.dbs[dbnum].slot_arr;
+
+			/* if nslots > 0, report just first entry and exit */
+			if (slot_arr->nslots)
+				pg_fatal("New cluster database \"%s\" is not empty: found logical replication slot \"%s\"",
+						 new_cluster.dbarr.dbs[dbnum].db_name,
+						 slot_arr->slots[0].slotname);
+		}
+
 	}
 }
 
@@ -1402,3 +1427,38 @@ check_for_user_defined_encoding_conversions(ClusterInfo *cluster)
 	else
 		check_ok();
 }
+
+/*
+ * Verify parameter settings for creating logical replication slots
+ */
+static void
+check_for_parameter_settings(ClusterInfo *new_cluster)
+{
+	PGresult   *res;
+	PGconn	   *conn = connectToServer(new_cluster, "template1");
+	int			max_replication_slots;
+	char	   *wal_level;
+
+	prep_status("Checking for logical replication slots");
+
+	res = executeQueryOrDie(conn, "SHOW max_replication_slots;");
+	max_replication_slots = atoi(PQgetvalue(res, 0, 0));
+
+	if (max_replication_slots == 0)
+		pg_fatal("max_replication_slots must be greater than 0");
+
+	PQclear(res);
+
+	res = executeQueryOrDie(conn, "SHOW wal_level;");
+	wal_level = PQgetvalue(res, 0, 0);
+
+	if (strcmp(wal_level, "logical") != 0)
+		pg_fatal("wal_level must be \"logical\", but is set to \"%s\"",
+				 wal_level);
+
+	PQclear(res);
+
+	PQfinish(conn);
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/dump.c b/src/bin/pg_upgrade/dump.c
index 6c8c82dca8..e6b90864f5 100644
--- a/src/bin/pg_upgrade/dump.c
+++ b/src/bin/pg_upgrade/dump.c
@@ -59,6 +59,30 @@ generate_old_dump(void)
 						   log_opts.dumpdir,
 						   sql_file_name, escaped_connstr.data);
 
+		/*
+		 * Dump logical replication slots if needed.
+		 *
+		 * XXX We cannot dump replication slots at the same time as the schema
+		 * dump because we need to separate the timing of restoring
+		 * replication slots and other objects. Replication slots, in
+		 * particular, should not be restored before executing the pg_resetwal
+		 * command because it will remove WALs that are required by the slots.
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			char		slots_file_name[MAXPGPATH];
+
+			snprintf(slots_file_name, sizeof(slots_file_name),
+					 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+			parallel_exec_prog(log_file_name, NULL,
+							   "\"%s/pg_dump\" %s --logical-replication-slots-only "
+							   "--quote-all-identifiers --binary-upgrade %s "
+							   "--file=\"%s/%s\" %s",
+							   new_cluster.bindir, cluster_conn_opts(&old_cluster),
+							   log_opts.verbose ? "--verbose" : "",
+							   log_opts.dumpdir,
+							   slots_file_name, escaped_connstr.data);
+		}
 		termPQExpBuffer(&escaped_connstr);
 	}
 
diff --git a/src/bin/pg_upgrade/info.c b/src/bin/pg_upgrade/info.c
index 85ed15ae4a..9679941217 100644
--- a/src/bin/pg_upgrade/info.c
+++ b/src/bin/pg_upgrade/info.c
@@ -23,10 +23,12 @@ static void free_db_and_rel_infos(DbInfoArr *db_arr);
 static void get_template0_info(ClusterInfo *cluster);
 static void get_db_infos(ClusterInfo *cluster);
 static void get_rel_infos(ClusterInfo *cluster, DbInfo *dbinfo);
+static void get_logical_slot_infos_per_db(ClusterInfo *cluster, DbInfo *dbinfo);
 static void free_rel_infos(RelInfoArr *rel_arr);
+static void free_logical_slot_infos(LogicalSlotInfoArr *slot_arr);
 static void print_db_infos(DbInfoArr *db_arr);
 static void print_rel_infos(RelInfoArr *rel_arr);
-
+static void print_slot_infos(LogicalSlotInfoArr *slot_arr);
 
 /*
  * gen_db_file_maps()
@@ -394,7 +396,7 @@ get_db_infos(ClusterInfo *cluster)
 	i_spclocation = PQfnumber(res, "spclocation");
 
 	ntups = PQntuples(res);
-	dbinfos = (DbInfo *) pg_malloc(sizeof(DbInfo) * ntups);
+	dbinfos = (DbInfo *) pg_malloc0(sizeof(DbInfo) * ntups);
 
 	for (tupnum = 0; tupnum < ntups; tupnum++)
 	{
@@ -600,6 +602,94 @@ get_rel_infos(ClusterInfo *cluster, DbInfo *dbinfo)
 	dbinfo->rel_arr.nrels = num_rels;
 }
 
+/*
+ * get_logical_slot_infos()
+ *
+ * Higher level routine to generate LogicalSlotInfoArr for all databases.
+ */
+void
+get_logical_slot_infos(ClusterInfo *cluster)
+{
+	int			dbnum;
+
+	if (cluster == &old_cluster)
+		pg_log(PG_VERBOSE, "\nsource databases:");
+	else
+		pg_log(PG_VERBOSE, "\ntarget databases:");
+
+	for (dbnum = 0; dbnum < cluster->dbarr.ndbs; dbnum++)
+	{
+		DbInfo *pDbInfo = &cluster->dbarr.dbs[dbnum];
+
+		if (pDbInfo->slot_arr.slots)
+			free_logical_slot_infos(&pDbInfo->slot_arr);
+
+		get_logical_slot_infos_per_db(cluster, pDbInfo);
+
+		if (log_opts.verbose)
+		{
+			pg_log(PG_VERBOSE, "Database: %s", pDbInfo->db_name);
+			print_slot_infos(&pDbInfo->slot_arr);
+		}
+	}
+}
+
+/*
+ * get_logical_slot_infos_per_db()
+ *
+ * gets the LogicalSlotInfos for all the logical replication slots of the database
+ * referred to by "dbinfo".
+ */
+static void
+get_logical_slot_infos_per_db(ClusterInfo *cluster, DbInfo *dbinfo)
+{
+	PGconn	   *conn = connectToServer(cluster,
+									   dbinfo->db_name);
+	PGresult   *res;
+	LogicalSlotInfo *slotinfos;
+
+	int			ntups;
+	int			slotnum;
+	int			num_slots = 0;
+
+	int			i_slotname;
+	int			i_plugin;
+	int			i_twophase;
+
+	char		query[QUERY_ALLOC];
+
+	query[0] = '\0';			/* initialize query string to empty */
+
+	snprintf(query, sizeof(query),
+			 "SELECT slot_name, plugin, two_phase "
+			 "FROM pg_catalog.pg_replication_slots "
+			 "WHERE database = current_database() AND temporary = false "
+			 "AND wal_status IN ('reserved', 'extended');");
+
+	res = executeQueryOrDie(conn, "%s", query);
+
+	ntups = PQntuples(res);
+
+	slotinfos = (LogicalSlotInfo *) pg_malloc(sizeof(LogicalSlotInfo) * ntups);
+
+	i_slotname = PQfnumber(res, "slot_name");
+	i_plugin = PQfnumber(res, "plugin");
+	i_twophase = PQfnumber(res, "two_phase");
+
+	for (slotnum = 0; slotnum < ntups; slotnum++)
+	{
+		LogicalSlotInfo *curr = &slotinfos[num_slots++];
+
+		curr->slotname = pg_strdup(PQgetvalue(res, slotnum, i_slotname));
+		curr->plugin = pg_strdup(PQgetvalue(res, slotnum, i_plugin));
+		curr->two_phase = (strcmp(PQgetvalue(res, slotnum, i_twophase), "t") == 0);
+	}
+
+	PQfinish(conn);
+
+	dbinfo->slot_arr.slots = slotinfos;
+	dbinfo->slot_arr.nslots = num_slots;
+}
 
 static void
 free_db_and_rel_infos(DbInfoArr *db_arr)
@@ -634,6 +724,19 @@ free_rel_infos(RelInfoArr *rel_arr)
 	rel_arr->nrels = 0;
 }
 
+static void
+free_logical_slot_infos(LogicalSlotInfoArr *slot_arr)
+{
+	int			slotnum;
+
+	for (slotnum = 0; slotnum < slot_arr->nslots; slotnum++)
+	{
+		pg_free(slot_arr->slots[slotnum].slotname);
+		pg_free(slot_arr->slots[slotnum].plugin);
+	}
+	pg_free(slot_arr->slots);
+	slot_arr->nslots = 0;
+}
 
 static void
 print_db_infos(DbInfoArr *db_arr)
@@ -660,3 +763,15 @@ print_rel_infos(RelInfoArr *rel_arr)
 			   rel_arr->rels[relnum].reloid,
 			   rel_arr->rels[relnum].tablespace);
 }
+
+static void
+print_slot_infos(LogicalSlotInfoArr *slot_arr)
+{
+	int			slotnum;
+
+	for (slotnum = 0; slotnum < slot_arr->nslots; slotnum++)
+		pg_log(PG_VERBOSE, "slotname: %s, plugin: %s, two_phase: %d",
+			   slot_arr->slots[slotnum].slotname,
+			   slot_arr->slots[slotnum].plugin,
+			   slot_arr->slots[slotnum].two_phase);
+}
diff --git a/src/bin/pg_upgrade/meson.build b/src/bin/pg_upgrade/meson.build
index 12a97f84e2..228f29b688 100644
--- a/src/bin/pg_upgrade/meson.build
+++ b/src/bin/pg_upgrade/meson.build
@@ -42,6 +42,7 @@ tests += {
     'tests': [
       't/001_basic.pl',
       't/002_pg_upgrade.pl',
+      't/003_logical_replication_slots.pl',
     ],
     'test_kwargs': {'priority': 40}, # pg_upgrade tests are slow
   },
diff --git a/src/bin/pg_upgrade/option.c b/src/bin/pg_upgrade/option.c
index 640361009e..df66a5ffe6 100644
--- a/src/bin/pg_upgrade/option.c
+++ b/src/bin/pg_upgrade/option.c
@@ -57,6 +57,7 @@ parseCommandLine(int argc, char *argv[])
 		{"verbose", no_argument, NULL, 'v'},
 		{"clone", no_argument, NULL, 1},
 		{"copy", no_argument, NULL, 2},
+		{"include-logical-replication-slots", no_argument, NULL, 3},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -199,6 +200,10 @@ parseCommandLine(int argc, char *argv[])
 				user_opts.transfer_mode = TRANSFER_MODE_COPY;
 				break;
 
+			case 3:
+				user_opts.include_logical_slots = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						os_info.progname);
@@ -289,6 +294,8 @@ usage(void)
 	printf(_("  -V, --version                 display version information, then exit\n"));
 	printf(_("  --clone                       clone instead of copying files to new cluster\n"));
 	printf(_("  --copy                        copy files to new cluster (default)\n"));
+	printf(_("  --include-logical-replication-slots\n"
+			 "                                upgrade logical replication slots\n"));
 	printf(_("  -?, --help                    show this help, then exit\n"));
 	printf(_("\n"
 			 "Before running pg_upgrade you must:\n"
diff --git a/src/bin/pg_upgrade/pg_upgrade.c b/src/bin/pg_upgrade/pg_upgrade.c
index 75bab0a04c..373a9ef490 100644
--- a/src/bin/pg_upgrade/pg_upgrade.c
+++ b/src/bin/pg_upgrade/pg_upgrade.c
@@ -59,6 +59,7 @@ static void copy_xact_xlog_xid(void);
 static void set_frozenxids(bool minmxid_only);
 static void make_outputdirs(char *pgdata);
 static void setup(char *argv0, bool *live_check);
+static void create_logical_replication_slots(void);
 
 ClusterInfo old_cluster,
 			new_cluster;
@@ -188,6 +189,19 @@ main(int argc, char **argv)
 			  new_cluster.pgdata);
 	check_ok();
 
+	/*
+	 * Create logical replication slots if requested.
+	 *
+	 * Note: This must be done after doing pg_resetwal command because the
+	 * command will remove required WALs.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		start_postmaster(&new_cluster, true);
+		create_logical_replication_slots();
+		stop_postmaster(false);
+	}
+
 	if (user_opts.do_sync)
 	{
 		prep_status("Sync data directory to disk");
@@ -860,3 +874,50 @@ set_frozenxids(bool minmxid_only)
 
 	check_ok();
 }
+
+/*
+ * create_logical_replication_slots()
+ *
+ * Similar to create_new_objects() but only restores logical replication slots.
+ */
+static void
+create_logical_replication_slots(void)
+{
+	int			dbnum;
+
+	prep_status_progress("Restoring logical replication slots in the new cluster");
+
+	for (dbnum = 0; dbnum < old_cluster.dbarr.ndbs; dbnum++)
+	{
+		char		slots_file_name[MAXPGPATH],
+					log_file_name[MAXPGPATH];
+		DbInfo	   *old_db = &old_cluster.dbarr.dbs[dbnum];
+
+		pg_log(PG_STATUS, "%s", old_db->db_name);
+
+		snprintf(slots_file_name, sizeof(slots_file_name),
+				 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+		snprintf(log_file_name, sizeof(log_file_name),
+				 DB_DUMP_LOG_FILE_MASK, old_db->db_oid);
+
+		parallel_exec_prog(log_file_name,
+						   NULL,
+						   "\"%s/psql\" %s --echo-queries --set ON_ERROR_STOP=on "
+						   "--no-psqlrc --dbname %s -f \"%s/%s\"",
+						   new_cluster.bindir,
+						   cluster_conn_opts(&new_cluster),
+						   old_db->db_name,
+						   log_opts.dumpdir,
+						   slots_file_name);
+	}
+
+	/* reap all children */
+	while (reap_child(true) == true)
+		;
+
+	end_progress_output();
+	check_ok();
+
+	/* update new_cluster info again */
+	get_logical_slot_infos(&new_cluster);
+}
diff --git a/src/bin/pg_upgrade/pg_upgrade.h b/src/bin/pg_upgrade/pg_upgrade.h
index 3eea0139c7..7adbb50807 100644
--- a/src/bin/pg_upgrade/pg_upgrade.h
+++ b/src/bin/pg_upgrade/pg_upgrade.h
@@ -29,6 +29,7 @@
 /* contains both global db information and CREATE DATABASE commands */
 #define GLOBALS_DUMP_FILE	"pg_upgrade_dump_globals.sql"
 #define DB_DUMP_FILE_MASK	"pg_upgrade_dump_%u.custom"
+#define DB_DUMP_LOGICAL_SLOTS_FILE_MASK	"pg_upgrade_dump_%u_logical_slots.sql"
 
 /*
  * Base directories that include all the files generated internally, from the
@@ -150,6 +151,22 @@ typedef struct
 	int			nrels;
 } RelInfoArr;
 
+/*
+ * Structure to store logical replication slot information
+ */
+typedef struct
+{
+	char	   *slotname;		/* slot name */
+	char	   *plugin;			/* plugin */
+	bool		two_phase;		/* Can the slot decode 2PC? */
+} LogicalSlotInfo;
+
+typedef struct
+{
+	LogicalSlotInfo *slots;
+	int			nslots;
+} LogicalSlotInfoArr;
+
 /*
  * The following structure represents a relation mapping.
  */
@@ -176,6 +193,7 @@ typedef struct
 	char		db_tablespace[MAXPGPATH];	/* database default tablespace
 											 * path */
 	RelInfoArr	rel_arr;		/* array of all user relinfos */
+	LogicalSlotInfoArr slot_arr;	/* array of all LogicalSlotInfo */
 } DbInfo;
 
 /*
@@ -304,6 +322,8 @@ typedef struct
 	transferMode transfer_mode; /* copy files or link them? */
 	int			jobs;			/* number of processes/threads to use */
 	char	   *socketdir;		/* directory to use for Unix sockets */
+	bool		include_logical_slots;	/* true -> dump and restore logical
+										 * replication slots */
 } UserOpts;
 
 typedef struct
@@ -400,6 +420,7 @@ FileNameMap *gen_db_file_maps(DbInfo *old_db,
 							  DbInfo *new_db, int *nmaps, const char *old_pgdata,
 							  const char *new_pgdata);
 void		get_db_and_rel_infos(ClusterInfo *cluster);
+void		get_logical_slot_infos(ClusterInfo *cluster);
 
 /* option.c */
 
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
new file mode 100644
index 0000000000..3430c641aa
--- /dev/null
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -0,0 +1,115 @@
+# Copyright (c) 2023, PostgreSQL Global Development Group
+
+# Tests for upgrading replication slots
+
+use strict;
+use warnings;
+
+use File::Path qw(rmtree);
+
+use PostgreSQL::Test::Cluster;
+use PostgreSQL::Test::Utils;
+use Test::More;
+
+# Can be changed to test the other modes
+my $mode = $ENV{PG_TEST_PG_UPGRADE_MODE} || '--copy';
+
+# Initialize old node
+my $old_node = PostgreSQL::Test::Cluster->new('old_node');
+$old_node->init(allows_streaming => 'logical');
+$old_node->start;
+
+# Initialize new node
+my $new_node = PostgreSQL::Test::Cluster->new('new_node');
+$new_node->init(allows_streaming => 1);
+
+my $bindir = $new_node->config_data('--bindir');
+
+$old_node->stop;
+
+# Cause a failure at the start of pg_upgrade because wal_level is replica
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with wrong wal_level');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+# Preparations for the subsequent test. The case max_replication_slots is set
+# to 0 is prohibit.
+$new_node->append_conf('postgresql.conf', "wal_level = 'logical'");
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 0");
+
+# Cause a failure at the start of pg_upgrade because max_replication_slots is 0
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with wrong max_replication_slots');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+# Preparations for the subsequent test. max_replication_slots is set to
+# non-zero value to succeed the pg_upgrade
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 10");
+
+# Create a slot on old node, and generate WALs
+$old_node->start;
+$old_node->safe_psql(
+	'postgres', qq[
+	SELECT pg_create_logical_replication_slot('test_slot', 'test_decoding', false, true);
+	CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a;
+]);
+
+my $result = $old_node->safe_psql('postgres',
+	"SELECT count (*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)"
+);
+is($result, qq(12), 'ensure WALs are not consumed yet');
+$old_node->stop;
+
+# Actual run, pg_upgrade_output.d is removed at the end
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots'
+	],
+	'run of pg_upgrade of old node');
+ok( !-d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+$new_node->start;
+$result = $new_node->safe_psql('postgres',
+	"SELECT slot_name, two_phase FROM pg_replication_slots");
+is($result, qq(test_slot|t), 'check the slot exists on new node');
+
+done_testing();
diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list
index b4058b88c3..5944cb34ea 100644
--- a/src/tools/pgindent/typedefs.list
+++ b/src/tools/pgindent/typedefs.list
@@ -1479,6 +1479,7 @@ LogicalRepBeginData
 LogicalRepCommitData
 LogicalRepCommitPreparedTxnData
 LogicalRepCtxStruct
+LogicalReplicationSlotInfo
 LogicalRepMode
 LogicalRepMsgType
 LogicalRepPartMapEntry
@@ -1492,6 +1493,8 @@ LogicalRepTupleData
 LogicalRepTyp
 LogicalRepWorker
 LogicalRewriteMappingData
+LogicalSlotInfo
+LogicalSlotInfoArr
 LogicalTape
 LogicalTapeSet
 LsnReadQueue
-- 
2.27.0

From d388af3dc8aea721d27f16dd0c535004d5def0c8 Mon Sep 17 00:00:00 2001
From: Julien Rouhaud <julien.rouhaud@free.fr>
Date: Fri, 14 Apr 2023 13:49:09 +0800
Subject: [PATCH v11 2/4] Always persist to disk logical slots during a
 shutdown checkpoint.

It's entirely possible for a logical slot to have a confirmed_flush_lsn higher
than the last value saved on disk while not being marked as dirty.  It's
currently not a problem to lose that value during a clean shutdown / restart
cycle, but a later patch adding support for pg_upgrade of publications and
logical slots will rely on that value being properly persisted to disk.

Author: Julien Rouhaud
Reviewed-by: FIXME
Discussion: FIXME
---
 src/backend/access/transam/xlog.c |  2 +-
 src/backend/replication/slot.c    | 26 ++++++++++++++++----------
 src/include/replication/slot.h    |  2 +-
 3 files changed, 18 insertions(+), 12 deletions(-)

diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index bc5a8e0569..78b4528f2c 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7011,7 +7011,7 @@ static void
 CheckPointGuts(XLogRecPtr checkPointRedo, int flags)
 {
 	CheckPointRelationMap();
-	CheckPointReplicationSlots();
+	CheckPointReplicationSlots(flags & CHECKPOINT_IS_SHUTDOWN);
 	CheckPointSnapBuild();
 	CheckPointLogicalRewriteHeap();
 	CheckPointReplicationOrigin();
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index 8021aaa0a8..aeea6ffd1f 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -109,7 +109,8 @@ static void ReplicationSlotDropPtr(ReplicationSlot *slot);
 /* internal persistency functions */
 static void RestoreSlotFromDisk(const char *name);
 static void CreateSlotOnDisk(ReplicationSlot *slot);
-static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel);
+static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+						   bool is_shutdown);
 
 /*
  * Report shared-memory space needed by ReplicationSlotsShmemInit.
@@ -783,7 +784,7 @@ ReplicationSlotSave(void)
 	Assert(MyReplicationSlot != NULL);
 
 	sprintf(path, "pg_replslot/%s", NameStr(MyReplicationSlot->data.name));
-	SaveSlotToPath(MyReplicationSlot, path, ERROR);
+	SaveSlotToPath(MyReplicationSlot, path, ERROR, false);
 }
 
 /*
@@ -1565,11 +1566,10 @@ restart:
 /*
  * Flush all replication slots to disk.
  *
- * This needn't actually be part of a checkpoint, but it's a convenient
- * location.
+ * is_shutdown is true in case of a shutdown checkpoint.
  */
 void
-CheckPointReplicationSlots(void)
+CheckPointReplicationSlots(bool is_shutdown)
 {
 	int			i;
 
@@ -1594,7 +1594,7 @@ CheckPointReplicationSlots(void)
 
 		/* save the slot to disk, locking is handled in SaveSlotToPath() */
 		sprintf(path, "pg_replslot/%s", NameStr(s->data.name));
-		SaveSlotToPath(s, path, LOG);
+		SaveSlotToPath(s, path, LOG, is_shutdown);
 	}
 	LWLockRelease(ReplicationSlotAllocationLock);
 }
@@ -1700,7 +1700,7 @@ CreateSlotOnDisk(ReplicationSlot *slot)
 
 	/* Write the actual state file. */
 	slot->dirty = true;			/* signal that we really need to write */
-	SaveSlotToPath(slot, tmppath, ERROR);
+	SaveSlotToPath(slot, tmppath, ERROR, false);
 
 	/* Rename the directory into place. */
 	if (rename(tmppath, path) != 0)
@@ -1726,7 +1726,8 @@ CreateSlotOnDisk(ReplicationSlot *slot)
  * Shared functionality between saving and creating a replication slot.
  */
 static void
-SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
+SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+			   bool is_shutdown)
 {
 	char		tmppath[MAXPGPATH];
 	char		path[MAXPGPATH];
@@ -1740,8 +1741,13 @@ SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
 	slot->just_dirtied = false;
 	SpinLockRelease(&slot->mutex);
 
-	/* and don't do anything if there's nothing to write */
-	if (!was_dirty)
+	/*
+	 * and don't do anything if there's nothing to write, unless it's this is
+	 * called for a logical slot during a shutdown checkpoint, as we want to
+	 * persist the confirmed_flush_lsn in that case, even if that's the only
+	 * modification.
+	 */
+	if (!was_dirty && !is_shutdown && !SlotIsLogical(slot))
 		return;
 
 	LWLockAcquire(&slot->io_in_progress_lock, LW_EXCLUSIVE);
diff --git a/src/include/replication/slot.h b/src/include/replication/slot.h
index a8a89dc784..7ca37c9f70 100644
--- a/src/include/replication/slot.h
+++ b/src/include/replication/slot.h
@@ -241,7 +241,7 @@ extern void ReplicationSlotNameForTablesync(Oid suboid, Oid relid, char *syncslo
 extern void ReplicationSlotDropAtPubNode(WalReceiverConn *wrconn, char *slotname, bool missing_ok);
 
 extern void StartupReplicationSlots(void);
-extern void CheckPointReplicationSlots(void);
+extern void CheckPointReplicationSlots(bool is_shutdown);
 
 extern void CheckSlotRequirements(void);
 extern void CheckSlotPermissions(void);
-- 
2.27.0

From 369128a910d704294cf8a7bfabaa34b42671be8d Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Fri, 14 Apr 2023 08:59:03 +0000
Subject: [PATCH v11 3/4] pg_upgrade: Add check function for
 --include-logical-replication-slots option

---
 src/bin/pg_upgrade/check.c                    | 80 ++++++++++++++++++-
 .../t/003_logical_replication_slots.pl        | 30 ++++++-
 2 files changed, 108 insertions(+), 2 deletions(-)

diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index 1802a30fe6..678867d64d 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -9,7 +9,10 @@
 
 #include "postgres_fe.h"
 
+#include "access/xlogrecord.h"
+#include "access/xlog_internal.h"
 #include "catalog/pg_authid_d.h"
+#include "catalog/pg_control.h"
 #include "catalog/pg_collation.h"
 #include "fe_utils/string_utils.h"
 #include "mb/pg_wchar.h"
@@ -31,7 +34,7 @@ static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
 static void check_for_parameter_settings(ClusterInfo *new_cluster);
-
+static void check_for_confirmed_flush_lsn(ClusterInfo *cluster);
 
 /*
  * fix_path_separator
@@ -108,6 +111,8 @@ check_and_dump_old_cluster(bool live_check)
 	check_for_composite_data_type_usage(&old_cluster);
 	check_for_reg_data_type_usage(&old_cluster);
 	check_for_isn_and_int8_passing_mismatch(&old_cluster);
+	if (user_opts.include_logical_slots)
+		check_for_confirmed_flush_lsn(&old_cluster);
 
 	/*
 	 * PG 16 increased the size of the 'aclitem' type, which breaks the on-disk
@@ -1439,6 +1444,10 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 	int			max_replication_slots;
 	char	   *wal_level;
 
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(new_cluster->major_version < 1600))
+		return;
+
 	prep_status("Checking for logical replication slots");
 
 	res = executeQueryOrDie(conn, "SHOW max_replication_slots;");
@@ -1462,3 +1471,72 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 
 	check_ok();
 }
+
+/*
+ * Verify that all logical replication slots consumed all WALs, except a
+ * CHECKPOINT_SHUTDOWN record.
+ */
+static void
+check_for_confirmed_flush_lsn(ClusterInfo *cluster)
+{
+	int			i,
+				ntups,
+				i_slotname;
+	bool		is_error = false;
+	PGresult   *res;
+	DbInfo	   *active_db = &cluster->dbarr.dbs[0];
+	PGconn	   *conn = connectToServer(cluster, active_db->db_name);
+
+	Assert(user_opts.include_logical_slots);
+
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(cluster->major_version < 1600))
+		return;
+
+	prep_status("Checking for logical replication slots");
+
+	/*
+	 * Check that all logical replication slots have reached the current WAL
+	 * position, except for the CHECKPOINT_SHUTDOWN record. Even if all WALs
+	 * are consumed before shutting down the node, the checkpointer generates
+	 * a CHECKPOINT_SHUTDOWN record at shutdown, which cannot be consumed by
+	 * any slots. Therefore, we must allow for a difference between
+	 * pg_current_wal_insert_lsn() and confirmed_flush_lsn.
+	 */
+#define SHUTDOWN_RECORD_SIZE  (SizeOfXLogRecord + \
+							   SizeOfXLogRecordDataHeaderShort + \
+							   sizeof(CheckPoint))
+
+	res = executeQueryOrDie(conn,
+							"SELECT slot_name FROM pg_catalog.pg_replication_slots "
+							"WHERE (pg_catalog.pg_current_wal_insert_lsn() - confirmed_flush_lsn) > %d "
+							"AND temporary = false AND wal_status IN ('reserved', 'extended');",
+							(int) (SizeOfXLogLongPHD + SHUTDOWN_RECORD_SIZE));
+
+#undef SHUTDOWN_RECORD_SIZE
+
+	ntups = PQntuples(res);
+	i_slotname = PQfnumber(res, "slot_name");
+
+	for (i = 0; i < ntups; i++)
+	{
+		char	   *slotname;
+
+		is_error = true;
+
+		slotname = PQgetvalue(res, i, i_slotname);
+
+		pg_log(PG_WARNING,
+			   "\nWARNING: logical replication slot \"%s\" has not consumed WALs yet",
+			   slotname);
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	if (is_error)
+		pg_fatal("--include-logical-replication-slots requires that all "
+				 "logical replication slots consumed all the WALs");
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
index 3430c641aa..e0e68df1ca 100644
--- a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -85,11 +85,39 @@ $old_node->safe_psql(
 ]);
 
 my $result = $old_node->safe_psql('postgres',
-	"SELECT count (*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)"
+	"SELECT count(*) FROM pg_logical_slot_peek_changes('test_slot', NULL, NULL)"
 );
+
 is($result, qq(12), 'ensure WALs are not consumed yet');
 $old_node->stop;
 
+# Cause a failure at the start of pg_upgrade because test_slot does not
+# finish consuming all the WALs
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with idle replication slots');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+$old_node->start;
+$old_node->safe_psql('postgres',
+	"SELECT count (*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)"
+);
+$old_node->stop;
+
 # Actual run, pg_upgrade_output.d is removed at the end
 command_ok(
 	[
-- 
2.27.0

From fcfd691eb1a2b8621ee6ef9c0aab8a2d91ee5049 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Mon, 24 Apr 2023 11:03:28 +0000
Subject: [PATCH v11 4/4] Change the method used to check logical replication
 slots during the live check

When a live check is requested, there is a possibility of additional changes
occurring, which may cause the current WAL position to exceed the confirmed_flush_lsn
of the slot. As a result, we check the confirmed_flush_lsn of each logical slot
instead. This is sufficient as all the WAL records will be sent during thepublisher's
shutdown.
---
 src/bin/pg_upgrade/check.c                    | 68 ++++++++++++++++++-
 .../t/003_logical_replication_slots.pl        | 66 +++++++++++++++++-
 2 files changed, 130 insertions(+), 4 deletions(-)

diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index 678867d64d..eef44706f4 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -35,6 +35,7 @@ static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
 static void check_for_parameter_settings(ClusterInfo *new_cluster);
 static void check_for_confirmed_flush_lsn(ClusterInfo *cluster);
+static void check_are_logical_slots_active(ClusterInfo *cluster);
 
 /*
  * fix_path_separator
@@ -112,7 +113,19 @@ check_and_dump_old_cluster(bool live_check)
 	check_for_reg_data_type_usage(&old_cluster);
 	check_for_isn_and_int8_passing_mismatch(&old_cluster);
 	if (user_opts.include_logical_slots)
-		check_for_confirmed_flush_lsn(&old_cluster);
+	{
+		/*
+		 * The method used to check logical replication slots is dependent on
+		 * the value of the live_check parameter. This change was implemented
+		 * because, during a live check, it is possible for additional changes
+		 * to occur at the old node, which could cause the current WAL position
+		 * to exceed the confirmed_flush_lsn of the slot.
+		 */
+		if (live_check)
+			check_are_logical_slots_active(&old_cluster);
+		else
+			check_for_confirmed_flush_lsn(&old_cluster);
+	}
 
 	/*
 	 * PG 16 increased the size of the 'aclitem' type, which breaks the on-disk
@@ -1472,6 +1485,59 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 	check_ok();
 }
 
+/*
+ * Verify that all logical replication slots are active
+ */
+static void
+check_are_logical_slots_active(ClusterInfo *cluster)
+{
+	int			i,
+				ntups,
+				i_slotname;
+	bool		is_error = false;
+	PGresult   *res;
+	DbInfo	   *active_db = &cluster->dbarr.dbs[0];
+	PGconn	   *conn = connectToServer(cluster, active_db->db_name);
+
+	Assert(user_opts.include_logical_slots);
+
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(cluster->major_version < 1600))
+		return;
+
+	prep_status("Checking for logical replication slots");
+
+	res = executeQueryOrDie(conn,
+							"SELECT slot_name FROM pg_catalog.pg_replication_slots "
+							"WHERE active IS FALSE "
+							"AND temporary = false AND wal_status IN ('reserved', 'extended');");
+
+	ntups = PQntuples(res);
+	i_slotname = PQfnumber(res, "slot_name");
+
+	for (i = 0; i < ntups; i++)
+	{
+		char	   *slotname;
+
+		is_error = true;
+
+		slotname = PQgetvalue(res, i, i_slotname);
+
+		pg_log(PG_WARNING,
+			   "\nWARNING: logical replication slot \"%s\" is not active",
+			   slotname);
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	if (is_error)
+		pg_fatal("--include-logical-replication-slots with --check requires that "
+				 "all logical replication slots are active");
+
+	check_ok();
+}
+
 /*
  * Verify that all logical replication slots consumed all WALs, except a
  * CHECKPOINT_SHUTDOWN record.
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
index e0e68df1ca..cf58f735f9 100644
--- a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -19,6 +19,10 @@ my $old_node = PostgreSQL::Test::Cluster->new('old_node');
 $old_node->init(allows_streaming => 'logical');
 $old_node->start;
 
+# Initialize subscriber, which will be used only for --check
+my $subscriber = PostgreSQL::Test::Cluster->new('subscriber');
+$subscriber->init(allows_streaming => 'logical');
+
 # Initialize new node
 my $new_node = PostgreSQL::Test::Cluster->new('new_node');
 $new_node->init(allows_streaming => 1);
@@ -76,15 +80,71 @@ rmtree($new_node->data_dir . "/pg_upgrade_output.d");
 # non-zero value to succeed the pg_upgrade
 $new_node->append_conf('postgresql.conf', "max_replication_slots = 10");
 
-# Create a slot on old node, and generate WALs
+# Setup logical replication
 $old_node->start;
+$old_node->safe_psql('postgres',
+	"CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a");
+$old_node->safe_psql('postgres', "CREATE PUBLICATION pub FOR ALL TABLES");
+my $old_connstr = $old_node->connstr . ' dbname=postgres';
+
+$subscriber->start;
+$subscriber->safe_psql('postgres', "CREATE TABLE tbl (a int)");
+$subscriber->safe_psql('postgres',
+	"CREATE SUBSCRIPTION sub CONNECTION '$old_connstr' PUBLICATION pub WITH (copy_data = true)"
+);
+
+# Wait for initial table sync to finish
+$subscriber->wait_for_subscription_sync($old_node, 'sub');
+
+my $result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(10), 'check initial rows on subscriber');
+
+# Start a background session and open a transaction (not committed yet)
+my $bsession = $old_node->background_psql('postgres');
+$bsession->query_safe(
+	q{
+BEGIN;
+INSERT INTO tbl VALUES (generate_series(11, 20))
+});
+
+$result = $old_node->safe_psql('postgres',
+	"SELECT count(*) FROM pg_replication_slots WHERE pg_current_wal_insert_lsn() > confirmed_flush_lsn"
+);
+is($result, qq(1),
+	'check the current WAL position exceeds confirmed_flush_lsn');
+
+# Run pg_upgrade --check. In the command the status of each logical slots will
+# be checked and then this will be succeeded.
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+		'--check'
+	],
+	'run of pg_upgrade of old node');
+ok( !-d $old_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+# Cleanup
+$bsession->query_safe("ABORT");
+$bsession->quit;
+$subscriber->safe_psql('postgres', "DROP SUBSCRIPTION sub");
+
+# Create a slot on old node, and generate WALs
 $old_node->safe_psql(
 	'postgres', qq[
 	SELECT pg_create_logical_replication_slot('test_slot', 'test_decoding', false, true);
-	CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a;
+	INSERT INTO tbl VALUES (generate_series(11, 20));
 ]);
 
-my $result = $old_node->safe_psql('postgres',
+$result = $old_node->safe_psql('postgres',
 	"SELECT count(*) FROM pg_logical_slot_peek_changes('test_slot', NULL, NULL)"
 );
 
-- 
2.27.0

From ec1981d9c2d6a5fe1e1924d4b8ee8d7447be0297 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Tue, 4 Apr 2023 05:49:34 +0000
Subject: [PATCH v12 1/4] pg_upgrade: Add --include-logical-replication-slots
 option

This commit introduces a new pg_upgrade option called "--include-logical-replication-slots".
This allows nodes with logical replication slots to be upgraded. The commit can
be divided into two parts: one for pg_dump and another for pg_upgrade.

For pg_dump this commit includes a new option called "--logical-replication-slots-only".
This option can be used to dump logical replication slots. When this option is
specified, the slot_name, plugin, and two_phase parameters are extracted from
pg_replication_slots. An SQL file is then generated which executes
pg_create_logical_replication_slot() with the extracted parameters.

For pg_upgrade, when '--include-logical-replication-slots' is specified, it executes
pg_dump with the new "--logical-replication-slots-only" option and restores from the
dump. Note that we cannot dump replication slots at the same time as the schema
dump because we need to separate the timing of restoring replication slots and
other objects. Replication slots, in  particular, should not be restored before
executing the pg_resetwal command because it will remove WALs that are required
by the slots.

The significant advantage of this commit is that it makes it easy to continue
logical replication even after upgrading the publisher node. Previously, pg_upgrade
allowed copying publications to a new node. With this new commit, adjusting the
connection string to the new publisher will cause the apply worker on the subscriber
to connect to the new publisher automatically. This enables seamless continuation
of logical replication, even after an upgrade.

Author: Hayato Kuroda
Reviewed-by: Peter Smith, Julien Rouhaud, Vignesh C
---
 doc/src/sgml/ref/pgupgrade.sgml               |  11 ++
 src/bin/pg_dump/pg_backup.h                   |   1 +
 src/bin/pg_dump/pg_dump.c                     | 155 ++++++++++++++++++
 src/bin/pg_dump/pg_dump.h                     |  17 +-
 src/bin/pg_dump/pg_dump_sort.c                |  11 +-
 src/bin/pg_upgrade/Makefile                   |   3 +
 src/bin/pg_upgrade/check.c                    |  62 +++++++
 src/bin/pg_upgrade/dump.c                     |  24 +++
 src/bin/pg_upgrade/info.c                     | 108 +++++++++++-
 src/bin/pg_upgrade/meson.build                |   1 +
 src/bin/pg_upgrade/option.c                   |   7 +
 src/bin/pg_upgrade/pg_upgrade.c               |  61 +++++++
 src/bin/pg_upgrade/pg_upgrade.h               |  21 +++
 .../t/003_logical_replication_slots.pl        | 115 +++++++++++++
 src/tools/pgindent/typedefs.list              |   3 +
 15 files changed, 595 insertions(+), 5 deletions(-)
 create mode 100644 src/bin/pg_upgrade/t/003_logical_replication_slots.pl

diff --git a/doc/src/sgml/ref/pgupgrade.sgml b/doc/src/sgml/ref/pgupgrade.sgml
index 7816b4c685..94e90ff506 100644
--- a/doc/src/sgml/ref/pgupgrade.sgml
+++ b/doc/src/sgml/ref/pgupgrade.sgml
@@ -240,6 +240,17 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--include-logical-replication-slots</option></term>
+      <listitem>
+       <para>
+        Upgrade logical replication slots. Only permanent replication slots
+        are included. Note that pg_upgrade does not check the installation of
+        plugins.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
       <term><option>-?</option></term>
       <term><option>--help</option></term>
diff --git a/src/bin/pg_dump/pg_backup.h b/src/bin/pg_dump/pg_backup.h
index aba780ef4b..0a4e931f9b 100644
--- a/src/bin/pg_dump/pg_backup.h
+++ b/src/bin/pg_dump/pg_backup.h
@@ -187,6 +187,7 @@ typedef struct _dumpOptions
 	int			use_setsessauth;
 	int			enable_row_security;
 	int			load_via_partition_root;
+	int			logical_slots_only;
 
 	/* default, if no "inclusion" switches appear, is to dump everything */
 	bool		include_everything;
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index 41a51ec5cd..0ff503a736 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -328,6 +328,9 @@ static void setupDumpWorker(Archive *AH);
 static TableInfo *getRootTableInfo(const TableInfo *tbinfo);
 static bool forcePartitionRootLoad(const TableInfo *tbinfo);
 
+static void getLogicalReplicationSlots(Archive *fout);
+static void dumpLogicalReplicationSlot(Archive *fout,
+									   const LogicalReplicationSlotInfo *slotinfo);
 
 int
 main(int argc, char **argv)
@@ -431,6 +434,7 @@ main(int argc, char **argv)
 		{"table-and-children", required_argument, NULL, 12},
 		{"exclude-table-and-children", required_argument, NULL, 13},
 		{"exclude-table-data-and-children", required_argument, NULL, 14},
+		{"logical-replication-slots-only", no_argument, NULL, 15},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -657,6 +661,10 @@ main(int argc, char **argv)
 										  optarg);
 				break;
 
+			case 15:			/* dump only replication slot(s) */
+				dopt.logical_slots_only = true;
+				break;
+
 			default:
 				/* getopt_long already emitted a complaint */
 				pg_log_error_hint("Try \"%s --help\" for more information.", progname);
@@ -714,6 +722,21 @@ main(int argc, char **argv)
 	if (dopt.do_nothing && dopt.dump_inserts == 0)
 		pg_fatal("option --on-conflict-do-nothing requires option --inserts, --rows-per-insert, or --column-inserts");
 
+	if (dopt.logical_slots_only)
+	{
+		if (!dopt.binary_upgrade)
+			pg_fatal("options --logical-replication-slots-only requires option --binary-upgrade");
+
+		if (dopt.dataOnly)
+			pg_fatal("options --logical-replication-slots-only and -a/--data-only cannot be used together");
+
+		if (dopt.schemaOnly)
+			pg_fatal("options --logical-replication-slots-only and -s/--schema-only cannot be used together");
+
+		if (dopt.outputClean)
+			pg_fatal("options --logical-replication-slots-only and -c/--clean cannot be used together");
+	}
+
 	/* Identify archive format to emit */
 	archiveFormat = parseArchiveFormat(format, &archiveMode);
 
@@ -876,6 +899,16 @@ main(int argc, char **argv)
 			pg_fatal("no matching extensions were found");
 	}
 
+	/*
+	 * If dump logical-replication-slots-only was requested, dump only them
+	 * and skip everything else.
+	 */
+	if (dopt.logical_slots_only)
+	{
+		getLogicalReplicationSlots(fout);
+		goto dump;
+	}
+
 	/*
 	 * Dumping LOs is the default for dumps where an inclusion switch is not
 	 * used (an "include everything" dump).  -B can be used to exclude LOs
@@ -936,6 +969,8 @@ main(int argc, char **argv)
 	if (!dopt.no_security_labels)
 		collectSecLabels(fout);
 
+dump:
+
 	/* Lastly, create dummy objects to represent the section boundaries */
 	boundaryObjs = createBoundaryObjects();
 
@@ -1109,6 +1144,10 @@ help(const char *progname)
 			 "                               servers matching PATTERN\n"));
 	printf(_("  --inserts                    dump data as INSERT commands, rather than COPY\n"));
 	printf(_("  --load-via-partition-root    load partitions via the root table\n"));
+	/*
+	 * The option --logical-replication-slots-only is used only by pg_upgrade
+	 * and should not be called by users, which is why it is not listed.
+	 */
 	printf(_("  --no-comments                do not dump comments\n"));
 	printf(_("  --no-publications            do not dump publications\n"));
 	printf(_("  --no-security-labels         do not dump security label assignments\n"));
@@ -10252,6 +10291,10 @@ dumpDumpableObject(Archive *fout, DumpableObject *dobj)
 		case DO_SUBSCRIPTION:
 			dumpSubscription(fout, (const SubscriptionInfo *) dobj);
 			break;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			dumpLogicalReplicationSlot(fout,
+									   (const LogicalReplicationSlotInfo *) dobj);
+			break;
 		case DO_PRE_DATA_BOUNDARY:
 		case DO_POST_DATA_BOUNDARY:
 			/* never dumped, nothing to do */
@@ -18227,6 +18270,7 @@ addBoundaryDependencies(DumpableObject **dobjs, int numObjs,
 			case DO_PUBLICATION_REL:
 			case DO_PUBLICATION_TABLE_IN_SCHEMA:
 			case DO_SUBSCRIPTION:
+			case DO_LOGICAL_REPLICATION_SLOT:
 				/* Post-data objects: must come after the post-data boundary */
 				addObjectDependency(dobj, postDataBound->dumpId);
 				break;
@@ -18488,3 +18532,114 @@ appendReloptionsArrayAH(PQExpBuffer buffer, const char *reloptions,
 	if (!res)
 		pg_log_warning("could not parse %s array", "reloptions");
 }
+
+/*
+ * getLogicalReplicationSlots
+ *	  get information about replication slots
+ */
+static void
+getLogicalReplicationSlots(Archive *fout)
+{
+	PGresult   *res;
+	LogicalReplicationSlotInfo *slotinfo;
+	PQExpBuffer query;
+
+	int			i_slotname;
+	int			i_plugin;
+	int			i_twophase;
+	int			i,
+				ntups;
+
+	/* Check whether we should dump or not */
+	if (fout->remoteVersion < 160000)
+		return;
+
+	Assert(fout->dopt->logical_slots_only);
+
+	query = createPQExpBuffer();
+
+	resetPQExpBuffer(query);
+
+	/*
+	 * Get replication slots.
+	 *
+	 * XXX: Which information must be extracted from old node? Currently three
+	 * attributes are extracted because they are used by
+	 * pg_create_logical_replication_slot().
+	 *
+	 * XXX: Do we have to support physical slots?
+	 */
+	appendPQExpBufferStr(query,
+						 "SELECT slot_name, plugin, two_phase "
+						 "FROM pg_catalog.pg_replication_slots "
+						 "WHERE database = current_database() AND temporary = false "
+						 "AND wal_status IN ('reserved', 'extended');");
+
+	res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
+
+	ntups = PQntuples(res);
+
+	i_slotname = PQfnumber(res, "slot_name");
+	i_plugin = PQfnumber(res, "plugin");
+	i_twophase = PQfnumber(res, "two_phase");
+
+	slotinfo = pg_malloc(ntups * sizeof(LogicalReplicationSlotInfo));
+
+	for (i = 0; i < ntups; i++)
+	{
+		slotinfo[i].dobj.objType = DO_LOGICAL_REPLICATION_SLOT;
+
+		slotinfo[i].dobj.catId.tableoid = InvalidOid;
+		slotinfo[i].dobj.catId.oid = InvalidOid;
+		AssignDumpId(&slotinfo[i].dobj);
+
+		slotinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_slotname));
+
+		slotinfo[i].plugin = pg_strdup(PQgetvalue(res, i, i_plugin));
+		slotinfo[i].twophase = (strcmp(PQgetvalue(res, i, i_twophase), "t") == 0);
+
+		/*
+		 * Note: Currently we do not have any options to include/exclude slots
+		 * in dumping, so all the slots must be selected.
+		 */
+		slotinfo[i].dobj.dump = DUMP_COMPONENT_DEFINITION;
+	}
+	PQclear(res);
+
+	destroyPQExpBuffer(query);
+}
+
+/*
+ * dumpLogicalReplicationSlot
+ *	  dump creation functions for the given logical replication slots
+ */
+static void
+dumpLogicalReplicationSlot(Archive *fout,
+						   const LogicalReplicationSlotInfo *slotinfo)
+{
+	Assert(fout->dopt->logical_slots_only);
+
+	if (slotinfo->dobj.dump & DUMP_COMPONENT_DEFINITION)
+	{
+		PQExpBuffer query = createPQExpBuffer();
+
+		/*
+		 * XXX: For simplification, pg_create_logical_replication_slot() is
+		 * used. Is it sufficient?
+		 */
+		appendPQExpBuffer(query, "SELECT pg_catalog.pg_create_logical_replication_slot(");
+		appendStringLiteralAH(query, slotinfo->dobj.name, fout);
+		appendPQExpBuffer(query, ", ");
+		appendStringLiteralAH(query, slotinfo->plugin, fout);
+		appendPQExpBuffer(query, ", false, %s);",
+						  slotinfo->twophase ? "true" : "false");
+
+		ArchiveEntry(fout, slotinfo->dobj.catId, slotinfo->dobj.dumpId,
+					 ARCHIVE_OPTS(.tag = slotinfo->dobj.name,
+								  .description = "REPLICATION SLOT",
+								  .section = SECTION_POST_DATA,
+								  .createStmt = query->data));
+
+		destroyPQExpBuffer(query);
+	}
+}
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index ed6ce41ad7..de081c35ae 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -82,7 +82,8 @@ typedef enum
 	DO_PUBLICATION,
 	DO_PUBLICATION_REL,
 	DO_PUBLICATION_TABLE_IN_SCHEMA,
-	DO_SUBSCRIPTION
+	DO_SUBSCRIPTION,
+	DO_LOGICAL_REPLICATION_SLOT
 } DumpableObjectType;
 
 /*
@@ -666,6 +667,20 @@ typedef struct _SubscriptionInfo
 	char	   *subpasswordrequired;
 } SubscriptionInfo;
 
+/*
+ * The LogicalReplicationSlotInfo struct is used to represent replication
+ * slots.
+ *
+ * XXX: add more attributes if needed
+ */
+typedef struct _LogicalReplicationSlotInfo
+{
+	DumpableObject dobj;
+	char	   *plugin;
+	char	   *slottype;
+	bool		twophase;
+} LogicalReplicationSlotInfo;
+
 /*
  *	common utility functions
  */
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index 745578d855..4e12e46dc5 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -93,6 +93,7 @@ enum dbObjectTypePriorities
 	PRIO_PUBLICATION_REL,
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,
 	PRIO_SUBSCRIPTION,
+	PRIO_LOGICAL_REPLICATION_SLOT,
 	PRIO_DEFAULT_ACL,			/* done in ACL pass */
 	PRIO_EVENT_TRIGGER,			/* must be next to last! */
 	PRIO_REFRESH_MATVIEW		/* must be last! */
@@ -146,10 +147,11 @@ static const int dbObjectTypePriority[] =
 	PRIO_PUBLICATION,			/* DO_PUBLICATION */
 	PRIO_PUBLICATION_REL,		/* DO_PUBLICATION_REL */
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,	/* DO_PUBLICATION_TABLE_IN_SCHEMA */
-	PRIO_SUBSCRIPTION			/* DO_SUBSCRIPTION */
+	PRIO_SUBSCRIPTION,			/* DO_SUBSCRIPTION */
+	PRIO_LOGICAL_REPLICATION_SLOT	/* DO_LOGICAL_REPLICATION_SLOT */
 };
 
-StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_SUBSCRIPTION + 1),
+StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_LOGICAL_REPLICATION_SLOT + 1),
 				 "array length mismatch");
 
 static DumpId preDataBoundId;
@@ -1498,6 +1500,11 @@ describeDumpableObject(DumpableObject *obj, char *buf, int bufsize)
 					 "SUBSCRIPTION (ID %d OID %u)",
 					 obj->dumpId, obj->catId.oid);
 			return;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			snprintf(buf, bufsize,
+					 "LOGICAL REPLICATION SLOT (ID %d NAME %s)",
+					 obj->dumpId, obj->name);
+			return;
 		case DO_PRE_DATA_BOUNDARY:
 			snprintf(buf, bufsize,
 					 "PRE-DATA BOUNDARY  (ID %d)",
diff --git a/src/bin/pg_upgrade/Makefile b/src/bin/pg_upgrade/Makefile
index 5834513add..815d1a7ca1 100644
--- a/src/bin/pg_upgrade/Makefile
+++ b/src/bin/pg_upgrade/Makefile
@@ -3,6 +3,9 @@
 PGFILEDESC = "pg_upgrade - an in-place binary upgrade utility"
 PGAPPICON = win32
 
+# required for 003_logical_replication_slots.pl
+EXTRA_INSTALL=contrib/test_decoding
+
 subdir = src/bin/pg_upgrade
 top_builddir = ../../..
 include $(top_builddir)/src/Makefile.global
diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index fea159689e..c7bed668ac 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -30,6 +30,7 @@ static void check_for_jsonb_9_4_usage(ClusterInfo *cluster);
 static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
+static void check_for_parameter_settings(ClusterInfo *new_cluster);
 
 
 /*
@@ -89,6 +90,10 @@ check_and_dump_old_cluster(bool live_check)
 	/* Extract a list of databases and tables from the old cluster */
 	get_db_and_rel_infos(&old_cluster);
 
+	/* Additionally, extract a list of logical replication slots if required */
+	if (user_opts.include_logical_slots)
+		get_logical_slot_infos(&old_cluster);
+
 	init_tablespaces();
 
 	get_loadable_libraries();
@@ -189,6 +194,12 @@ check_new_cluster(void)
 {
 	get_db_and_rel_infos(&new_cluster);
 
+	if (user_opts.include_logical_slots)
+	{
+		get_logical_slot_infos(&new_cluster);
+		check_for_parameter_settings(&new_cluster);
+	}
+
 	check_new_cluster_is_empty();
 
 	check_loadable_libraries();
@@ -364,6 +375,22 @@ check_new_cluster_is_empty(void)
 						 rel_arr->rels[relnum].nspname,
 						 rel_arr->rels[relnum].relname);
 		}
+
+		/*
+		 * If --include-logical-replication-slots is required, check the
+		 * existence of slots
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			LogicalSlotInfoArr *slot_arr = &new_cluster.dbarr.dbs[dbnum].slot_arr;
+
+			/* if nslots > 0, report just first entry and exit */
+			if (slot_arr->nslots)
+				pg_fatal("New cluster database \"%s\" is not empty: found logical replication slot \"%s\"",
+						 new_cluster.dbarr.dbs[dbnum].db_name,
+						 slot_arr->slots[0].slotname);
+		}
+
 	}
 }
 
@@ -1402,3 +1429,38 @@ check_for_user_defined_encoding_conversions(ClusterInfo *cluster)
 	else
 		check_ok();
 }
+
+/*
+ * Verify parameter settings for creating logical replication slots
+ */
+static void
+check_for_parameter_settings(ClusterInfo *new_cluster)
+{
+	PGresult   *res;
+	PGconn	   *conn = connectToServer(new_cluster, "template1");
+	int			max_replication_slots;
+	char	   *wal_level;
+
+	prep_status("Checking for logical replication slots");
+
+	res = executeQueryOrDie(conn, "SHOW max_replication_slots;");
+	max_replication_slots = atoi(PQgetvalue(res, 0, 0));
+
+	if (max_replication_slots == 0)
+		pg_fatal("max_replication_slots must be greater than 0");
+
+	PQclear(res);
+
+	res = executeQueryOrDie(conn, "SHOW wal_level;");
+	wal_level = PQgetvalue(res, 0, 0);
+
+	if (strcmp(wal_level, "logical") != 0)
+		pg_fatal("wal_level must be \"logical\", but is set to \"%s\"",
+				 wal_level);
+
+	PQclear(res);
+
+	PQfinish(conn);
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/dump.c b/src/bin/pg_upgrade/dump.c
index 6c8c82dca8..e6b90864f5 100644
--- a/src/bin/pg_upgrade/dump.c
+++ b/src/bin/pg_upgrade/dump.c
@@ -59,6 +59,30 @@ generate_old_dump(void)
 						   log_opts.dumpdir,
 						   sql_file_name, escaped_connstr.data);
 
+		/*
+		 * Dump logical replication slots if needed.
+		 *
+		 * XXX We cannot dump replication slots at the same time as the schema
+		 * dump because we need to separate the timing of restoring
+		 * replication slots and other objects. Replication slots, in
+		 * particular, should not be restored before executing the pg_resetwal
+		 * command because it will remove WALs that are required by the slots.
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			char		slots_file_name[MAXPGPATH];
+
+			snprintf(slots_file_name, sizeof(slots_file_name),
+					 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+			parallel_exec_prog(log_file_name, NULL,
+							   "\"%s/pg_dump\" %s --logical-replication-slots-only "
+							   "--quote-all-identifiers --binary-upgrade %s "
+							   "--file=\"%s/%s\" %s",
+							   new_cluster.bindir, cluster_conn_opts(&old_cluster),
+							   log_opts.verbose ? "--verbose" : "",
+							   log_opts.dumpdir,
+							   slots_file_name, escaped_connstr.data);
+		}
 		termPQExpBuffer(&escaped_connstr);
 	}
 
diff --git a/src/bin/pg_upgrade/info.c b/src/bin/pg_upgrade/info.c
index 85ed15ae4a..6916cb23b4 100644
--- a/src/bin/pg_upgrade/info.c
+++ b/src/bin/pg_upgrade/info.c
@@ -23,10 +23,11 @@ static void free_db_and_rel_infos(DbInfoArr *db_arr);
 static void get_template0_info(ClusterInfo *cluster);
 static void get_db_infos(ClusterInfo *cluster);
 static void get_rel_infos(ClusterInfo *cluster, DbInfo *dbinfo);
+static void get_logical_slot_infos_per_db(ClusterInfo *cluster, DbInfo *dbinfo);
 static void free_rel_infos(RelInfoArr *rel_arr);
 static void print_db_infos(DbInfoArr *db_arr);
 static void print_rel_infos(RelInfoArr *rel_arr);
-
+static void print_slot_infos(LogicalSlotInfoArr *slot_arr);
 
 /*
  * gen_db_file_maps()
@@ -600,6 +601,98 @@ get_rel_infos(ClusterInfo *cluster, DbInfo *dbinfo)
 	dbinfo->rel_arr.nrels = num_rels;
 }
 
+/*
+ * get_logical_slot_infos()
+ *
+ * Higher level routine to generate LogicalSlotInfoArr for all databases.
+ */
+void
+get_logical_slot_infos(ClusterInfo *cluster)
+{
+	int			dbnum;
+
+	if (cluster == &old_cluster)
+		pg_log(PG_VERBOSE, "\nsource databases:");
+	else
+		pg_log(PG_VERBOSE, "\ntarget databases:");
+
+	for (dbnum = 0; dbnum < cluster->dbarr.ndbs; dbnum++)
+	{
+		DbInfo *pDbInfo = &cluster->dbarr.dbs[dbnum];
+
+		get_logical_slot_infos_per_db(cluster, pDbInfo);
+
+		if (log_opts.verbose)
+		{
+			pg_log(PG_VERBOSE, "Database: %s", pDbInfo->db_name);
+			print_slot_infos(&pDbInfo->slot_arr);
+		}
+	}
+}
+
+/*
+ * get_logical_slot_infos_per_db()
+ *
+ * gets the LogicalSlotInfos for all the logical replication slots of the database
+ * referred to by "dbinfo".
+ */
+static void
+get_logical_slot_infos_per_db(ClusterInfo *cluster, DbInfo *dbinfo)
+{
+	PGconn	   *conn = connectToServer(cluster,
+									   dbinfo->db_name);
+	PGresult   *res;
+	LogicalSlotInfo *slotinfos;
+
+	int			ntups;
+	int			slotnum;
+	int			num_slots = 0;
+
+	int			i_slotname;
+	int			i_plugin;
+	int			i_twophase;
+
+	char		query[QUERY_ALLOC];
+
+	query[0] = '\0';			/* initialize query string to empty */
+
+	snprintf(query, sizeof(query),
+			 "SELECT slot_name, plugin, two_phase "
+			 "FROM pg_catalog.pg_replication_slots "
+			 "WHERE database = current_database() AND temporary = false "
+			 "AND wal_status IN ('reserved', 'extended');");
+
+	res = executeQueryOrDie(conn, "%s", query);
+
+	ntups = PQntuples(res);
+
+	if (ntups)
+		slotinfos = (LogicalSlotInfo *) pg_malloc(sizeof(LogicalSlotInfo) * ntups);
+	else
+	{
+		slotinfos = NULL;
+		goto cleanup;
+	}
+
+	i_slotname = PQfnumber(res, "slot_name");
+	i_plugin = PQfnumber(res, "plugin");
+	i_twophase = PQfnumber(res, "two_phase");
+
+	for (slotnum = 0; slotnum < ntups; slotnum++)
+	{
+		LogicalSlotInfo *curr = &slotinfos[num_slots++];
+
+		curr->slotname = pg_strdup(PQgetvalue(res, slotnum, i_slotname));
+		curr->plugin = pg_strdup(PQgetvalue(res, slotnum, i_plugin));
+		curr->two_phase = (strcmp(PQgetvalue(res, slotnum, i_twophase), "t") == 0);
+	}
+
+cleanup:
+	PQfinish(conn);
+
+	dbinfo->slot_arr.slots = slotinfos;
+	dbinfo->slot_arr.nslots = num_slots;
+}
 
 static void
 free_db_and_rel_infos(DbInfoArr *db_arr)
@@ -634,7 +727,6 @@ free_rel_infos(RelInfoArr *rel_arr)
 	rel_arr->nrels = 0;
 }
 
-
 static void
 print_db_infos(DbInfoArr *db_arr)
 {
@@ -660,3 +752,15 @@ print_rel_infos(RelInfoArr *rel_arr)
 			   rel_arr->rels[relnum].reloid,
 			   rel_arr->rels[relnum].tablespace);
 }
+
+static void
+print_slot_infos(LogicalSlotInfoArr *slot_arr)
+{
+	int			slotnum;
+
+	for (slotnum = 0; slotnum < slot_arr->nslots; slotnum++)
+		pg_log(PG_VERBOSE, "slotname: %s, plugin: %s, two_phase: %d",
+			   slot_arr->slots[slotnum].slotname,
+			   slot_arr->slots[slotnum].plugin,
+			   slot_arr->slots[slotnum].two_phase);
+}
diff --git a/src/bin/pg_upgrade/meson.build b/src/bin/pg_upgrade/meson.build
index 12a97f84e2..228f29b688 100644
--- a/src/bin/pg_upgrade/meson.build
+++ b/src/bin/pg_upgrade/meson.build
@@ -42,6 +42,7 @@ tests += {
     'tests': [
       't/001_basic.pl',
       't/002_pg_upgrade.pl',
+      't/003_logical_replication_slots.pl',
     ],
     'test_kwargs': {'priority': 40}, # pg_upgrade tests are slow
   },
diff --git a/src/bin/pg_upgrade/option.c b/src/bin/pg_upgrade/option.c
index 640361009e..df66a5ffe6 100644
--- a/src/bin/pg_upgrade/option.c
+++ b/src/bin/pg_upgrade/option.c
@@ -57,6 +57,7 @@ parseCommandLine(int argc, char *argv[])
 		{"verbose", no_argument, NULL, 'v'},
 		{"clone", no_argument, NULL, 1},
 		{"copy", no_argument, NULL, 2},
+		{"include-logical-replication-slots", no_argument, NULL, 3},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -199,6 +200,10 @@ parseCommandLine(int argc, char *argv[])
 				user_opts.transfer_mode = TRANSFER_MODE_COPY;
 				break;
 
+			case 3:
+				user_opts.include_logical_slots = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						os_info.progname);
@@ -289,6 +294,8 @@ usage(void)
 	printf(_("  -V, --version                 display version information, then exit\n"));
 	printf(_("  --clone                       clone instead of copying files to new cluster\n"));
 	printf(_("  --copy                        copy files to new cluster (default)\n"));
+	printf(_("  --include-logical-replication-slots\n"
+			 "                                upgrade logical replication slots\n"));
 	printf(_("  -?, --help                    show this help, then exit\n"));
 	printf(_("\n"
 			 "Before running pg_upgrade you must:\n"
diff --git a/src/bin/pg_upgrade/pg_upgrade.c b/src/bin/pg_upgrade/pg_upgrade.c
index 75bab0a04c..373a9ef490 100644
--- a/src/bin/pg_upgrade/pg_upgrade.c
+++ b/src/bin/pg_upgrade/pg_upgrade.c
@@ -59,6 +59,7 @@ static void copy_xact_xlog_xid(void);
 static void set_frozenxids(bool minmxid_only);
 static void make_outputdirs(char *pgdata);
 static void setup(char *argv0, bool *live_check);
+static void create_logical_replication_slots(void);
 
 ClusterInfo old_cluster,
 			new_cluster;
@@ -188,6 +189,19 @@ main(int argc, char **argv)
 			  new_cluster.pgdata);
 	check_ok();
 
+	/*
+	 * Create logical replication slots if requested.
+	 *
+	 * Note: This must be done after doing pg_resetwal command because the
+	 * command will remove required WALs.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		start_postmaster(&new_cluster, true);
+		create_logical_replication_slots();
+		stop_postmaster(false);
+	}
+
 	if (user_opts.do_sync)
 	{
 		prep_status("Sync data directory to disk");
@@ -860,3 +874,50 @@ set_frozenxids(bool minmxid_only)
 
 	check_ok();
 }
+
+/*
+ * create_logical_replication_slots()
+ *
+ * Similar to create_new_objects() but only restores logical replication slots.
+ */
+static void
+create_logical_replication_slots(void)
+{
+	int			dbnum;
+
+	prep_status_progress("Restoring logical replication slots in the new cluster");
+
+	for (dbnum = 0; dbnum < old_cluster.dbarr.ndbs; dbnum++)
+	{
+		char		slots_file_name[MAXPGPATH],
+					log_file_name[MAXPGPATH];
+		DbInfo	   *old_db = &old_cluster.dbarr.dbs[dbnum];
+
+		pg_log(PG_STATUS, "%s", old_db->db_name);
+
+		snprintf(slots_file_name, sizeof(slots_file_name),
+				 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+		snprintf(log_file_name, sizeof(log_file_name),
+				 DB_DUMP_LOG_FILE_MASK, old_db->db_oid);
+
+		parallel_exec_prog(log_file_name,
+						   NULL,
+						   "\"%s/psql\" %s --echo-queries --set ON_ERROR_STOP=on "
+						   "--no-psqlrc --dbname %s -f \"%s/%s\"",
+						   new_cluster.bindir,
+						   cluster_conn_opts(&new_cluster),
+						   old_db->db_name,
+						   log_opts.dumpdir,
+						   slots_file_name);
+	}
+
+	/* reap all children */
+	while (reap_child(true) == true)
+		;
+
+	end_progress_output();
+	check_ok();
+
+	/* update new_cluster info again */
+	get_logical_slot_infos(&new_cluster);
+}
diff --git a/src/bin/pg_upgrade/pg_upgrade.h b/src/bin/pg_upgrade/pg_upgrade.h
index 3eea0139c7..7adbb50807 100644
--- a/src/bin/pg_upgrade/pg_upgrade.h
+++ b/src/bin/pg_upgrade/pg_upgrade.h
@@ -29,6 +29,7 @@
 /* contains both global db information and CREATE DATABASE commands */
 #define GLOBALS_DUMP_FILE	"pg_upgrade_dump_globals.sql"
 #define DB_DUMP_FILE_MASK	"pg_upgrade_dump_%u.custom"
+#define DB_DUMP_LOGICAL_SLOTS_FILE_MASK	"pg_upgrade_dump_%u_logical_slots.sql"
 
 /*
  * Base directories that include all the files generated internally, from the
@@ -150,6 +151,22 @@ typedef struct
 	int			nrels;
 } RelInfoArr;
 
+/*
+ * Structure to store logical replication slot information
+ */
+typedef struct
+{
+	char	   *slotname;		/* slot name */
+	char	   *plugin;			/* plugin */
+	bool		two_phase;		/* Can the slot decode 2PC? */
+} LogicalSlotInfo;
+
+typedef struct
+{
+	LogicalSlotInfo *slots;
+	int			nslots;
+} LogicalSlotInfoArr;
+
 /*
  * The following structure represents a relation mapping.
  */
@@ -176,6 +193,7 @@ typedef struct
 	char		db_tablespace[MAXPGPATH];	/* database default tablespace
 											 * path */
 	RelInfoArr	rel_arr;		/* array of all user relinfos */
+	LogicalSlotInfoArr slot_arr;	/* array of all LogicalSlotInfo */
 } DbInfo;
 
 /*
@@ -304,6 +322,8 @@ typedef struct
 	transferMode transfer_mode; /* copy files or link them? */
 	int			jobs;			/* number of processes/threads to use */
 	char	   *socketdir;		/* directory to use for Unix sockets */
+	bool		include_logical_slots;	/* true -> dump and restore logical
+										 * replication slots */
 } UserOpts;
 
 typedef struct
@@ -400,6 +420,7 @@ FileNameMap *gen_db_file_maps(DbInfo *old_db,
 							  DbInfo *new_db, int *nmaps, const char *old_pgdata,
 							  const char *new_pgdata);
 void		get_db_and_rel_infos(ClusterInfo *cluster);
+void		get_logical_slot_infos(ClusterInfo *cluster);
 
 /* option.c */
 
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
new file mode 100644
index 0000000000..525a7704cf
--- /dev/null
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -0,0 +1,115 @@
+# Copyright (c) 2023, PostgreSQL Global Development Group
+
+# Tests for upgrading replication slots
+
+use strict;
+use warnings;
+
+use File::Path qw(rmtree);
+
+use PostgreSQL::Test::Cluster;
+use PostgreSQL::Test::Utils;
+use Test::More;
+
+# Can be changed to test the other modes
+my $mode = $ENV{PG_TEST_PG_UPGRADE_MODE} || '--copy';
+
+# Initialize old node
+my $old_node = PostgreSQL::Test::Cluster->new('old_node');
+$old_node->init(allows_streaming => 'logical');
+$old_node->start;
+
+# Initialize new node
+my $new_node = PostgreSQL::Test::Cluster->new('new_node');
+$new_node->init(allows_streaming => 1);
+
+my $bindir = $new_node->config_data('--bindir');
+
+$old_node->stop;
+
+# Cause a failure at the start of pg_upgrade because wal_level is replica
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with wrong wal_level');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+# Preparations for the subsequent test. The case max_replication_slots is set
+# to 0 is prohibited.
+$new_node->append_conf('postgresql.conf', "wal_level = 'logical'");
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 0");
+
+# Cause a failure at the start of pg_upgrade because max_replication_slots is 0
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with wrong max_replication_slots');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+# Preparations for the subsequent test. max_replication_slots is set to
+# non-zero value to succeed the pg_upgrade
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 10");
+
+# Create a slot on old node, and generate WALs
+$old_node->start;
+$old_node->safe_psql(
+	'postgres', qq[
+	SELECT pg_create_logical_replication_slot('test_slot', 'test_decoding', false, true);
+	CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a;
+]);
+
+my $result = $old_node->safe_psql('postgres',
+	"SELECT count (*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)"
+);
+is($result, qq(12), 'ensure WALs are not consumed yet');
+$old_node->stop;
+
+# Actual run, pg_upgrade_output.d is removed at the end
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots'
+	],
+	'run of pg_upgrade of old node');
+ok( !-d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+$new_node->start;
+$result = $new_node->safe_psql('postgres',
+	"SELECT slot_name, two_phase FROM pg_replication_slots");
+is($result, qq(test_slot|t), 'check the slot exists on new node');
+
+done_testing();
diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list
index b4058b88c3..5944cb34ea 100644
--- a/src/tools/pgindent/typedefs.list
+++ b/src/tools/pgindent/typedefs.list
@@ -1479,6 +1479,7 @@ LogicalRepBeginData
 LogicalRepCommitData
 LogicalRepCommitPreparedTxnData
 LogicalRepCtxStruct
+LogicalReplicationSlotInfo
 LogicalRepMode
 LogicalRepMsgType
 LogicalRepPartMapEntry
@@ -1492,6 +1493,8 @@ LogicalRepTupleData
 LogicalRepTyp
 LogicalRepWorker
 LogicalRewriteMappingData
+LogicalSlotInfo
+LogicalSlotInfoArr
 LogicalTape
 LogicalTapeSet
 LsnReadQueue
-- 
2.27.0

From 35bdeb9ba30f37c235fd9f592df494847921562f Mon Sep 17 00:00:00 2001
From: Julien Rouhaud <julien.rouhaud@free.fr>
Date: Fri, 14 Apr 2023 13:49:09 +0800
Subject: [PATCH v12 2/4] Always persist to disk logical slots during a
 shutdown checkpoint.

It's entirely possible for a logical slot to have a confirmed_flush_lsn higher
than the last value saved on disk while not being marked as dirty.  It's
currently not a problem to lose that value during a clean shutdown / restart
cycle, but a later patch adding support for pg_upgrade of publications and
logical slots will rely on that value being properly persisted to disk.

Author: Julien Rouhaud
Reviewed-by: FIXME
Discussion: FIXME
---
 src/backend/access/transam/xlog.c |  2 +-
 src/backend/replication/slot.c    | 26 ++++++++++++++++----------
 src/include/replication/slot.h    |  2 +-
 3 files changed, 18 insertions(+), 12 deletions(-)

diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index bc5a8e0569..78b4528f2c 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7011,7 +7011,7 @@ static void
 CheckPointGuts(XLogRecPtr checkPointRedo, int flags)
 {
 	CheckPointRelationMap();
-	CheckPointReplicationSlots();
+	CheckPointReplicationSlots(flags & CHECKPOINT_IS_SHUTDOWN);
 	CheckPointSnapBuild();
 	CheckPointLogicalRewriteHeap();
 	CheckPointReplicationOrigin();
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index 8021aaa0a8..aeea6ffd1f 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -109,7 +109,8 @@ static void ReplicationSlotDropPtr(ReplicationSlot *slot);
 /* internal persistency functions */
 static void RestoreSlotFromDisk(const char *name);
 static void CreateSlotOnDisk(ReplicationSlot *slot);
-static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel);
+static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+						   bool is_shutdown);
 
 /*
  * Report shared-memory space needed by ReplicationSlotsShmemInit.
@@ -783,7 +784,7 @@ ReplicationSlotSave(void)
 	Assert(MyReplicationSlot != NULL);
 
 	sprintf(path, "pg_replslot/%s", NameStr(MyReplicationSlot->data.name));
-	SaveSlotToPath(MyReplicationSlot, path, ERROR);
+	SaveSlotToPath(MyReplicationSlot, path, ERROR, false);
 }
 
 /*
@@ -1565,11 +1566,10 @@ restart:
 /*
  * Flush all replication slots to disk.
  *
- * This needn't actually be part of a checkpoint, but it's a convenient
- * location.
+ * is_shutdown is true in case of a shutdown checkpoint.
  */
 void
-CheckPointReplicationSlots(void)
+CheckPointReplicationSlots(bool is_shutdown)
 {
 	int			i;
 
@@ -1594,7 +1594,7 @@ CheckPointReplicationSlots(void)
 
 		/* save the slot to disk, locking is handled in SaveSlotToPath() */
 		sprintf(path, "pg_replslot/%s", NameStr(s->data.name));
-		SaveSlotToPath(s, path, LOG);
+		SaveSlotToPath(s, path, LOG, is_shutdown);
 	}
 	LWLockRelease(ReplicationSlotAllocationLock);
 }
@@ -1700,7 +1700,7 @@ CreateSlotOnDisk(ReplicationSlot *slot)
 
 	/* Write the actual state file. */
 	slot->dirty = true;			/* signal that we really need to write */
-	SaveSlotToPath(slot, tmppath, ERROR);
+	SaveSlotToPath(slot, tmppath, ERROR, false);
 
 	/* Rename the directory into place. */
 	if (rename(tmppath, path) != 0)
@@ -1726,7 +1726,8 @@ CreateSlotOnDisk(ReplicationSlot *slot)
  * Shared functionality between saving and creating a replication slot.
  */
 static void
-SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
+SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+			   bool is_shutdown)
 {
 	char		tmppath[MAXPGPATH];
 	char		path[MAXPGPATH];
@@ -1740,8 +1741,13 @@ SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
 	slot->just_dirtied = false;
 	SpinLockRelease(&slot->mutex);
 
-	/* and don't do anything if there's nothing to write */
-	if (!was_dirty)
+	/*
+	 * and don't do anything if there's nothing to write, unless it's this is
+	 * called for a logical slot during a shutdown checkpoint, as we want to
+	 * persist the confirmed_flush_lsn in that case, even if that's the only
+	 * modification.
+	 */
+	if (!was_dirty && !is_shutdown && !SlotIsLogical(slot))
 		return;
 
 	LWLockAcquire(&slot->io_in_progress_lock, LW_EXCLUSIVE);
diff --git a/src/include/replication/slot.h b/src/include/replication/slot.h
index a8a89dc784..7ca37c9f70 100644
--- a/src/include/replication/slot.h
+++ b/src/include/replication/slot.h
@@ -241,7 +241,7 @@ extern void ReplicationSlotNameForTablesync(Oid suboid, Oid relid, char *syncslo
 extern void ReplicationSlotDropAtPubNode(WalReceiverConn *wrconn, char *slotname, bool missing_ok);
 
 extern void StartupReplicationSlots(void);
-extern void CheckPointReplicationSlots(void);
+extern void CheckPointReplicationSlots(bool is_shutdown);
 
 extern void CheckSlotRequirements(void);
 extern void CheckSlotPermissions(void);
-- 
2.27.0

From 1398560d8c9cb316576375b7ceeec7a66b706f28 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Fri, 14 Apr 2023 08:59:03 +0000
Subject: [PATCH v12 3/4] pg_upgrade: Add check function for
 --include-logical-replication-slots option

---
 src/bin/pg_upgrade/check.c                    | 80 ++++++++++++++++++-
 .../t/003_logical_replication_slots.pl        | 30 ++++++-
 2 files changed, 108 insertions(+), 2 deletions(-)

diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index c7bed668ac..2e62eabe0a 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -9,7 +9,10 @@
 
 #include "postgres_fe.h"
 
+#include "access/xlogrecord.h"
+#include "access/xlog_internal.h"
 #include "catalog/pg_authid_d.h"
+#include "catalog/pg_control.h"
 #include "catalog/pg_collation.h"
 #include "fe_utils/string_utils.h"
 #include "mb/pg_wchar.h"
@@ -31,7 +34,7 @@ static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
 static void check_for_parameter_settings(ClusterInfo *new_cluster);
-
+static void check_for_confirmed_flush_lsn(ClusterInfo *cluster);
 
 /*
  * fix_path_separator
@@ -108,6 +111,8 @@ check_and_dump_old_cluster(bool live_check)
 	check_for_composite_data_type_usage(&old_cluster);
 	check_for_reg_data_type_usage(&old_cluster);
 	check_for_isn_and_int8_passing_mismatch(&old_cluster);
+	if (user_opts.include_logical_slots)
+		check_for_confirmed_flush_lsn(&old_cluster);
 
 	/*
 	 * PG 16 increased the size of the 'aclitem' type, which breaks the on-disk
@@ -1441,6 +1446,10 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 	int			max_replication_slots;
 	char	   *wal_level;
 
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(new_cluster->major_version < 1600))
+		return;
+
 	prep_status("Checking for logical replication slots");
 
 	res = executeQueryOrDie(conn, "SHOW max_replication_slots;");
@@ -1464,3 +1473,72 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 
 	check_ok();
 }
+
+/*
+ * Verify that all logical replication slots consumed all WALs, except a
+ * CHECKPOINT_SHUTDOWN record.
+ */
+static void
+check_for_confirmed_flush_lsn(ClusterInfo *cluster)
+{
+	int			i,
+				ntups,
+				i_slotname;
+	bool		is_error = false;
+	PGresult   *res;
+	DbInfo	   *active_db = &cluster->dbarr.dbs[0];
+	PGconn	   *conn = connectToServer(cluster, active_db->db_name);
+
+	Assert(user_opts.include_logical_slots);
+
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(cluster->major_version < 1600))
+		return;
+
+	prep_status("Checking for logical replication slots");
+
+	/*
+	 * Check that all logical replication slots have reached the current WAL
+	 * position, except for the CHECKPOINT_SHUTDOWN record. Even if all WALs
+	 * are consumed before shutting down the node, the checkpointer generates
+	 * a CHECKPOINT_SHUTDOWN record at shutdown, which cannot be consumed by
+	 * any slots. Therefore, we must allow for a difference between
+	 * pg_current_wal_insert_lsn() and confirmed_flush_lsn.
+	 */
+#define SHUTDOWN_RECORD_SIZE  (SizeOfXLogRecord + \
+							   SizeOfXLogRecordDataHeaderShort + \
+							   sizeof(CheckPoint))
+
+	res = executeQueryOrDie(conn,
+							"SELECT slot_name FROM pg_catalog.pg_replication_slots "
+							"WHERE (pg_catalog.pg_current_wal_insert_lsn() - confirmed_flush_lsn) > %d "
+							"AND temporary = false AND wal_status IN ('reserved', 'extended');",
+							(int) (SizeOfXLogLongPHD + SHUTDOWN_RECORD_SIZE));
+
+#undef SHUTDOWN_RECORD_SIZE
+
+	ntups = PQntuples(res);
+	i_slotname = PQfnumber(res, "slot_name");
+
+	for (i = 0; i < ntups; i++)
+	{
+		char	   *slotname;
+
+		is_error = true;
+
+		slotname = PQgetvalue(res, i, i_slotname);
+
+		pg_log(PG_WARNING,
+			   "\nWARNING: logical replication slot \"%s\" has not consumed WALs yet",
+			   slotname);
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	if (is_error)
+		pg_fatal("--include-logical-replication-slots requires that all "
+				 "logical replication slots consumed all the WALs");
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
index 525a7704cf..21fefca084 100644
--- a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -85,11 +85,39 @@ $old_node->safe_psql(
 ]);
 
 my $result = $old_node->safe_psql('postgres',
-	"SELECT count (*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)"
+	"SELECT count(*) FROM pg_logical_slot_peek_changes('test_slot', NULL, NULL)"
 );
+
 is($result, qq(12), 'ensure WALs are not consumed yet');
 $old_node->stop;
 
+# Cause a failure at the start of pg_upgrade because test_slot does not
+# finish consuming all the WALs
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with idle replication slots');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+$old_node->start;
+$old_node->safe_psql('postgres',
+	"SELECT count (*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)"
+);
+$old_node->stop;
+
 # Actual run, pg_upgrade_output.d is removed at the end
 command_ok(
 	[
-- 
2.27.0

From 4a4ecb107f7794edd59986826b794917df6f3210 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Mon, 24 Apr 2023 11:03:28 +0000
Subject: [PATCH v12 4/4] Change the method used to check logical replication
 slots during the live check

When a live check is requested, there is a possibility of additional changes
occurring, which may cause the current WAL position to exceed the confirmed_flush_lsn
of the slot. As a result, we check the confirmed_flush_lsn of each logical slot
instead. This is sufficient as all the WAL records will be sent during thepublisher's
shutdown.
---
 src/bin/pg_upgrade/check.c                    | 68 ++++++++++++++++++-
 .../t/003_logical_replication_slots.pl        | 66 +++++++++++++++++-
 2 files changed, 130 insertions(+), 4 deletions(-)

diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index 2e62eabe0a..f589af9eba 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -35,6 +35,7 @@ static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
 static void check_for_parameter_settings(ClusterInfo *new_cluster);
 static void check_for_confirmed_flush_lsn(ClusterInfo *cluster);
+static void check_are_logical_slots_active(ClusterInfo *cluster);
 
 /*
  * fix_path_separator
@@ -112,7 +113,19 @@ check_and_dump_old_cluster(bool live_check)
 	check_for_reg_data_type_usage(&old_cluster);
 	check_for_isn_and_int8_passing_mismatch(&old_cluster);
 	if (user_opts.include_logical_slots)
-		check_for_confirmed_flush_lsn(&old_cluster);
+	{
+		/*
+		 * The method used to check logical replication slots is dependent on
+		 * the value of the live_check parameter. This change was implemented
+		 * because, during a live check, it is possible for additional changes
+		 * to occur at the old node, which could cause the current WAL position
+		 * to exceed the confirmed_flush_lsn of the slot.
+		 */
+		if (live_check)
+			check_are_logical_slots_active(&old_cluster);
+		else
+			check_for_confirmed_flush_lsn(&old_cluster);
+	}
 
 	/*
 	 * PG 16 increased the size of the 'aclitem' type, which breaks the on-disk
@@ -1474,6 +1487,59 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 	check_ok();
 }
 
+/*
+ * Verify that all logical replication slots are active
+ */
+static void
+check_are_logical_slots_active(ClusterInfo *cluster)
+{
+	int			i,
+				ntups,
+				i_slotname;
+	bool		is_error = false;
+	PGresult   *res;
+	DbInfo	   *active_db = &cluster->dbarr.dbs[0];
+	PGconn	   *conn = connectToServer(cluster, active_db->db_name);
+
+	Assert(user_opts.include_logical_slots);
+
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(cluster->major_version < 1600))
+		return;
+
+	prep_status("Checking for logical replication slots");
+
+	res = executeQueryOrDie(conn,
+							"SELECT slot_name FROM pg_catalog.pg_replication_slots "
+							"WHERE active IS FALSE "
+							"AND temporary = false AND wal_status IN ('reserved', 'extended');");
+
+	ntups = PQntuples(res);
+	i_slotname = PQfnumber(res, "slot_name");
+
+	for (i = 0; i < ntups; i++)
+	{
+		char	   *slotname;
+
+		is_error = true;
+
+		slotname = PQgetvalue(res, i, i_slotname);
+
+		pg_log(PG_WARNING,
+			   "\nWARNING: logical replication slot \"%s\" is not active",
+			   slotname);
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	if (is_error)
+		pg_fatal("--include-logical-replication-slots with --check requires that "
+				 "all logical replication slots are active");
+
+	check_ok();
+}
+
 /*
  * Verify that all logical replication slots consumed all WALs, except a
  * CHECKPOINT_SHUTDOWN record.
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
index 21fefca084..d7fd864bd7 100644
--- a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -19,6 +19,10 @@ my $old_node = PostgreSQL::Test::Cluster->new('old_node');
 $old_node->init(allows_streaming => 'logical');
 $old_node->start;
 
+# Initialize subscriber, which will be used only for --check
+my $subscriber = PostgreSQL::Test::Cluster->new('subscriber');
+$subscriber->init(allows_streaming => 'logical');
+
 # Initialize new node
 my $new_node = PostgreSQL::Test::Cluster->new('new_node');
 $new_node->init(allows_streaming => 1);
@@ -76,15 +80,71 @@ rmtree($new_node->data_dir . "/pg_upgrade_output.d");
 # non-zero value to succeed the pg_upgrade
 $new_node->append_conf('postgresql.conf', "max_replication_slots = 10");
 
-# Create a slot on old node, and generate WALs
+# Setup logical replication
 $old_node->start;
+$old_node->safe_psql('postgres',
+	"CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a");
+$old_node->safe_psql('postgres', "CREATE PUBLICATION pub FOR ALL TABLES");
+my $old_connstr = $old_node->connstr . ' dbname=postgres';
+
+$subscriber->start;
+$subscriber->safe_psql('postgres', "CREATE TABLE tbl (a int)");
+$subscriber->safe_psql('postgres',
+	"CREATE SUBSCRIPTION sub CONNECTION '$old_connstr' PUBLICATION pub WITH (copy_data = true)"
+);
+
+# Wait for initial table sync to finish
+$subscriber->wait_for_subscription_sync($old_node, 'sub');
+
+my $result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(10), 'check initial rows on subscriber');
+
+# Start a background session and open a transaction (not committed yet)
+my $bsession = $old_node->background_psql('postgres');
+$bsession->query_safe(
+	q{
+BEGIN;
+INSERT INTO tbl VALUES (generate_series(11, 20))
+});
+
+$result = $old_node->safe_psql('postgres',
+	"SELECT count(*) FROM pg_replication_slots WHERE pg_current_wal_insert_lsn() > confirmed_flush_lsn"
+);
+is($result, qq(1),
+	'check the current WAL position exceeds confirmed_flush_lsn');
+
+# Run pg_upgrade --check. In the command the status of each logical slots will
+# be checked and then this will be succeeded.
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+		'--check'
+	],
+	'run of pg_upgrade of old node');
+ok( !-d $old_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+# Cleanup
+$bsession->query_safe("ABORT");
+$bsession->quit;
+$subscriber->safe_psql('postgres', "DROP SUBSCRIPTION sub");
+
+# Create a slot on old node, and generate WALs
 $old_node->safe_psql(
 	'postgres', qq[
 	SELECT pg_create_logical_replication_slot('test_slot', 'test_decoding', false, true);
-	CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a;
+	INSERT INTO tbl VALUES (generate_series(11, 20));
 ]);
 
-my $result = $old_node->safe_psql('postgres',
+$result = $old_node->safe_psql('postgres',
 	"SELECT count(*) FROM pg_logical_slot_peek_changes('test_slot', NULL, NULL)"
 );
 
-- 
2.27.0

From 45230a4af7c2819be44b1227ef19f425a075d6a7 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Tue, 4 Apr 2023 05:49:34 +0000
Subject: [PATCH v13 1/4] pg_upgrade: Add --include-logical-replication-slots
 option

This commit introduces a new pg_upgrade option called "--include-logical-replication-slots".
This allows nodes with logical replication slots to be upgraded. The commit can
be divided into two parts: one for pg_dump and another for pg_upgrade.

For pg_dump this commit includes a new option called "--logical-replication-slots-only".
This option can be used to dump logical replication slots. When this option is
specified, the slot_name, plugin, and two_phase parameters are extracted from
pg_replication_slots. An SQL file is then generated which executes
pg_create_logical_replication_slot() with the extracted parameters.

For pg_upgrade, when '--include-logical-replication-slots' is specified, it executes
pg_dump with the new "--logical-replication-slots-only" option and restores from the
dump. Note that we cannot dump replication slots at the same time as the schema
dump because we need to separate the timing of restoring replication slots and
other objects. Replication slots, in  particular, should not be restored before
executing the pg_resetwal command because it will remove WALs that are required
by the slots.

The significant advantage of this commit is that it makes it easy to continue
logical replication even after upgrading the publisher node. Previously, pg_upgrade
allowed copying publications to a new node. With this new commit, adjusting the
connection string to the new publisher will cause the apply worker on the subscriber
to connect to the new publisher automatically. This enables seamless continuation
of logical replication, even after an upgrade.

Author: Hayato Kuroda
Reviewed-by: Peter Smith, Julien Rouhaud, Vignesh C
---
 doc/src/sgml/ref/pgupgrade.sgml               |  11 ++
 src/bin/pg_dump/pg_backup.h                   |   1 +
 src/bin/pg_dump/pg_dump.c                     | 155 ++++++++++++++++++
 src/bin/pg_dump/pg_dump.h                     |  17 +-
 src/bin/pg_dump/pg_dump_sort.c                |  11 +-
 src/bin/pg_upgrade/Makefile                   |   3 +
 src/bin/pg_upgrade/check.c                    |  67 ++++++++
 src/bin/pg_upgrade/dump.c                     |  24 +++
 src/bin/pg_upgrade/info.c                     | 112 ++++++++++++-
 src/bin/pg_upgrade/meson.build                |   1 +
 src/bin/pg_upgrade/option.c                   |   7 +
 src/bin/pg_upgrade/pg_upgrade.c               |  61 +++++++
 src/bin/pg_upgrade/pg_upgrade.h               |  21 +++
 .../t/003_logical_replication_slots.pl        | 115 +++++++++++++
 src/tools/pgindent/typedefs.list              |   3 +
 15 files changed, 605 insertions(+), 4 deletions(-)
 create mode 100644 src/bin/pg_upgrade/t/003_logical_replication_slots.pl

diff --git a/doc/src/sgml/ref/pgupgrade.sgml b/doc/src/sgml/ref/pgupgrade.sgml
index 7816b4c685..94e90ff506 100644
--- a/doc/src/sgml/ref/pgupgrade.sgml
+++ b/doc/src/sgml/ref/pgupgrade.sgml
@@ -240,6 +240,17 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--include-logical-replication-slots</option></term>
+      <listitem>
+       <para>
+        Upgrade logical replication slots. Only permanent replication slots
+        are included. Note that pg_upgrade does not check the installation of
+        plugins.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
       <term><option>-?</option></term>
       <term><option>--help</option></term>
diff --git a/src/bin/pg_dump/pg_backup.h b/src/bin/pg_dump/pg_backup.h
index aba780ef4b..0a4e931f9b 100644
--- a/src/bin/pg_dump/pg_backup.h
+++ b/src/bin/pg_dump/pg_backup.h
@@ -187,6 +187,7 @@ typedef struct _dumpOptions
 	int			use_setsessauth;
 	int			enable_row_security;
 	int			load_via_partition_root;
+	int			logical_slots_only;
 
 	/* default, if no "inclusion" switches appear, is to dump everything */
 	bool		include_everything;
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index 41a51ec5cd..0ff503a736 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -328,6 +328,9 @@ static void setupDumpWorker(Archive *AH);
 static TableInfo *getRootTableInfo(const TableInfo *tbinfo);
 static bool forcePartitionRootLoad(const TableInfo *tbinfo);
 
+static void getLogicalReplicationSlots(Archive *fout);
+static void dumpLogicalReplicationSlot(Archive *fout,
+									   const LogicalReplicationSlotInfo *slotinfo);
 
 int
 main(int argc, char **argv)
@@ -431,6 +434,7 @@ main(int argc, char **argv)
 		{"table-and-children", required_argument, NULL, 12},
 		{"exclude-table-and-children", required_argument, NULL, 13},
 		{"exclude-table-data-and-children", required_argument, NULL, 14},
+		{"logical-replication-slots-only", no_argument, NULL, 15},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -657,6 +661,10 @@ main(int argc, char **argv)
 										  optarg);
 				break;
 
+			case 15:			/* dump only replication slot(s) */
+				dopt.logical_slots_only = true;
+				break;
+
 			default:
 				/* getopt_long already emitted a complaint */
 				pg_log_error_hint("Try \"%s --help\" for more information.", progname);
@@ -714,6 +722,21 @@ main(int argc, char **argv)
 	if (dopt.do_nothing && dopt.dump_inserts == 0)
 		pg_fatal("option --on-conflict-do-nothing requires option --inserts, --rows-per-insert, or --column-inserts");
 
+	if (dopt.logical_slots_only)
+	{
+		if (!dopt.binary_upgrade)
+			pg_fatal("options --logical-replication-slots-only requires option --binary-upgrade");
+
+		if (dopt.dataOnly)
+			pg_fatal("options --logical-replication-slots-only and -a/--data-only cannot be used together");
+
+		if (dopt.schemaOnly)
+			pg_fatal("options --logical-replication-slots-only and -s/--schema-only cannot be used together");
+
+		if (dopt.outputClean)
+			pg_fatal("options --logical-replication-slots-only and -c/--clean cannot be used together");
+	}
+
 	/* Identify archive format to emit */
 	archiveFormat = parseArchiveFormat(format, &archiveMode);
 
@@ -876,6 +899,16 @@ main(int argc, char **argv)
 			pg_fatal("no matching extensions were found");
 	}
 
+	/*
+	 * If dump logical-replication-slots-only was requested, dump only them
+	 * and skip everything else.
+	 */
+	if (dopt.logical_slots_only)
+	{
+		getLogicalReplicationSlots(fout);
+		goto dump;
+	}
+
 	/*
 	 * Dumping LOs is the default for dumps where an inclusion switch is not
 	 * used (an "include everything" dump).  -B can be used to exclude LOs
@@ -936,6 +969,8 @@ main(int argc, char **argv)
 	if (!dopt.no_security_labels)
 		collectSecLabels(fout);
 
+dump:
+
 	/* Lastly, create dummy objects to represent the section boundaries */
 	boundaryObjs = createBoundaryObjects();
 
@@ -1109,6 +1144,10 @@ help(const char *progname)
 			 "                               servers matching PATTERN\n"));
 	printf(_("  --inserts                    dump data as INSERT commands, rather than COPY\n"));
 	printf(_("  --load-via-partition-root    load partitions via the root table\n"));
+	/*
+	 * The option --logical-replication-slots-only is used only by pg_upgrade
+	 * and should not be called by users, which is why it is not listed.
+	 */
 	printf(_("  --no-comments                do not dump comments\n"));
 	printf(_("  --no-publications            do not dump publications\n"));
 	printf(_("  --no-security-labels         do not dump security label assignments\n"));
@@ -10252,6 +10291,10 @@ dumpDumpableObject(Archive *fout, DumpableObject *dobj)
 		case DO_SUBSCRIPTION:
 			dumpSubscription(fout, (const SubscriptionInfo *) dobj);
 			break;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			dumpLogicalReplicationSlot(fout,
+									   (const LogicalReplicationSlotInfo *) dobj);
+			break;
 		case DO_PRE_DATA_BOUNDARY:
 		case DO_POST_DATA_BOUNDARY:
 			/* never dumped, nothing to do */
@@ -18227,6 +18270,7 @@ addBoundaryDependencies(DumpableObject **dobjs, int numObjs,
 			case DO_PUBLICATION_REL:
 			case DO_PUBLICATION_TABLE_IN_SCHEMA:
 			case DO_SUBSCRIPTION:
+			case DO_LOGICAL_REPLICATION_SLOT:
 				/* Post-data objects: must come after the post-data boundary */
 				addObjectDependency(dobj, postDataBound->dumpId);
 				break;
@@ -18488,3 +18532,114 @@ appendReloptionsArrayAH(PQExpBuffer buffer, const char *reloptions,
 	if (!res)
 		pg_log_warning("could not parse %s array", "reloptions");
 }
+
+/*
+ * getLogicalReplicationSlots
+ *	  get information about replication slots
+ */
+static void
+getLogicalReplicationSlots(Archive *fout)
+{
+	PGresult   *res;
+	LogicalReplicationSlotInfo *slotinfo;
+	PQExpBuffer query;
+
+	int			i_slotname;
+	int			i_plugin;
+	int			i_twophase;
+	int			i,
+				ntups;
+
+	/* Check whether we should dump or not */
+	if (fout->remoteVersion < 160000)
+		return;
+
+	Assert(fout->dopt->logical_slots_only);
+
+	query = createPQExpBuffer();
+
+	resetPQExpBuffer(query);
+
+	/*
+	 * Get replication slots.
+	 *
+	 * XXX: Which information must be extracted from old node? Currently three
+	 * attributes are extracted because they are used by
+	 * pg_create_logical_replication_slot().
+	 *
+	 * XXX: Do we have to support physical slots?
+	 */
+	appendPQExpBufferStr(query,
+						 "SELECT slot_name, plugin, two_phase "
+						 "FROM pg_catalog.pg_replication_slots "
+						 "WHERE database = current_database() AND temporary = false "
+						 "AND wal_status IN ('reserved', 'extended');");
+
+	res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
+
+	ntups = PQntuples(res);
+
+	i_slotname = PQfnumber(res, "slot_name");
+	i_plugin = PQfnumber(res, "plugin");
+	i_twophase = PQfnumber(res, "two_phase");
+
+	slotinfo = pg_malloc(ntups * sizeof(LogicalReplicationSlotInfo));
+
+	for (i = 0; i < ntups; i++)
+	{
+		slotinfo[i].dobj.objType = DO_LOGICAL_REPLICATION_SLOT;
+
+		slotinfo[i].dobj.catId.tableoid = InvalidOid;
+		slotinfo[i].dobj.catId.oid = InvalidOid;
+		AssignDumpId(&slotinfo[i].dobj);
+
+		slotinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_slotname));
+
+		slotinfo[i].plugin = pg_strdup(PQgetvalue(res, i, i_plugin));
+		slotinfo[i].twophase = (strcmp(PQgetvalue(res, i, i_twophase), "t") == 0);
+
+		/*
+		 * Note: Currently we do not have any options to include/exclude slots
+		 * in dumping, so all the slots must be selected.
+		 */
+		slotinfo[i].dobj.dump = DUMP_COMPONENT_DEFINITION;
+	}
+	PQclear(res);
+
+	destroyPQExpBuffer(query);
+}
+
+/*
+ * dumpLogicalReplicationSlot
+ *	  dump creation functions for the given logical replication slots
+ */
+static void
+dumpLogicalReplicationSlot(Archive *fout,
+						   const LogicalReplicationSlotInfo *slotinfo)
+{
+	Assert(fout->dopt->logical_slots_only);
+
+	if (slotinfo->dobj.dump & DUMP_COMPONENT_DEFINITION)
+	{
+		PQExpBuffer query = createPQExpBuffer();
+
+		/*
+		 * XXX: For simplification, pg_create_logical_replication_slot() is
+		 * used. Is it sufficient?
+		 */
+		appendPQExpBuffer(query, "SELECT pg_catalog.pg_create_logical_replication_slot(");
+		appendStringLiteralAH(query, slotinfo->dobj.name, fout);
+		appendPQExpBuffer(query, ", ");
+		appendStringLiteralAH(query, slotinfo->plugin, fout);
+		appendPQExpBuffer(query, ", false, %s);",
+						  slotinfo->twophase ? "true" : "false");
+
+		ArchiveEntry(fout, slotinfo->dobj.catId, slotinfo->dobj.dumpId,
+					 ARCHIVE_OPTS(.tag = slotinfo->dobj.name,
+								  .description = "REPLICATION SLOT",
+								  .section = SECTION_POST_DATA,
+								  .createStmt = query->data));
+
+		destroyPQExpBuffer(query);
+	}
+}
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index ed6ce41ad7..de081c35ae 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -82,7 +82,8 @@ typedef enum
 	DO_PUBLICATION,
 	DO_PUBLICATION_REL,
 	DO_PUBLICATION_TABLE_IN_SCHEMA,
-	DO_SUBSCRIPTION
+	DO_SUBSCRIPTION,
+	DO_LOGICAL_REPLICATION_SLOT
 } DumpableObjectType;
 
 /*
@@ -666,6 +667,20 @@ typedef struct _SubscriptionInfo
 	char	   *subpasswordrequired;
 } SubscriptionInfo;
 
+/*
+ * The LogicalReplicationSlotInfo struct is used to represent replication
+ * slots.
+ *
+ * XXX: add more attributes if needed
+ */
+typedef struct _LogicalReplicationSlotInfo
+{
+	DumpableObject dobj;
+	char	   *plugin;
+	char	   *slottype;
+	bool		twophase;
+} LogicalReplicationSlotInfo;
+
 /*
  *	common utility functions
  */
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index 745578d855..4e12e46dc5 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -93,6 +93,7 @@ enum dbObjectTypePriorities
 	PRIO_PUBLICATION_REL,
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,
 	PRIO_SUBSCRIPTION,
+	PRIO_LOGICAL_REPLICATION_SLOT,
 	PRIO_DEFAULT_ACL,			/* done in ACL pass */
 	PRIO_EVENT_TRIGGER,			/* must be next to last! */
 	PRIO_REFRESH_MATVIEW		/* must be last! */
@@ -146,10 +147,11 @@ static const int dbObjectTypePriority[] =
 	PRIO_PUBLICATION,			/* DO_PUBLICATION */
 	PRIO_PUBLICATION_REL,		/* DO_PUBLICATION_REL */
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,	/* DO_PUBLICATION_TABLE_IN_SCHEMA */
-	PRIO_SUBSCRIPTION			/* DO_SUBSCRIPTION */
+	PRIO_SUBSCRIPTION,			/* DO_SUBSCRIPTION */
+	PRIO_LOGICAL_REPLICATION_SLOT	/* DO_LOGICAL_REPLICATION_SLOT */
 };
 
-StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_SUBSCRIPTION + 1),
+StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_LOGICAL_REPLICATION_SLOT + 1),
 				 "array length mismatch");
 
 static DumpId preDataBoundId;
@@ -1498,6 +1500,11 @@ describeDumpableObject(DumpableObject *obj, char *buf, int bufsize)
 					 "SUBSCRIPTION (ID %d OID %u)",
 					 obj->dumpId, obj->catId.oid);
 			return;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			snprintf(buf, bufsize,
+					 "LOGICAL REPLICATION SLOT (ID %d NAME %s)",
+					 obj->dumpId, obj->name);
+			return;
 		case DO_PRE_DATA_BOUNDARY:
 			snprintf(buf, bufsize,
 					 "PRE-DATA BOUNDARY  (ID %d)",
diff --git a/src/bin/pg_upgrade/Makefile b/src/bin/pg_upgrade/Makefile
index 5834513add..815d1a7ca1 100644
--- a/src/bin/pg_upgrade/Makefile
+++ b/src/bin/pg_upgrade/Makefile
@@ -3,6 +3,9 @@
 PGFILEDESC = "pg_upgrade - an in-place binary upgrade utility"
 PGAPPICON = win32
 
+# required for 003_logical_replication_slots.pl
+EXTRA_INSTALL=contrib/test_decoding
+
 subdir = src/bin/pg_upgrade
 top_builddir = ../../..
 include $(top_builddir)/src/Makefile.global
diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index fea159689e..f9cba8548e 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -30,6 +30,7 @@ static void check_for_jsonb_9_4_usage(ClusterInfo *cluster);
 static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
+static void check_for_parameter_settings(ClusterInfo *new_cluster);
 
 
 /*
@@ -89,6 +90,10 @@ check_and_dump_old_cluster(bool live_check)
 	/* Extract a list of databases and tables from the old cluster */
 	get_db_and_rel_infos(&old_cluster);
 
+	/* Additionally, extract a list of logical replication slots if required */
+	if (user_opts.include_logical_slots)
+		get_logical_slot_infos(&old_cluster);
+
 	init_tablespaces();
 
 	get_loadable_libraries();
@@ -189,6 +194,17 @@ check_new_cluster(void)
 {
 	get_db_and_rel_infos(&new_cluster);
 
+	/*
+	 * Do additional works if --include-logical-replication-slots is required.
+	 * These must be done before check_new_cluster_is_empty() because the
+	 * slot_arr attribute of the new_cluster will be checked in the function.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		get_logical_slot_infos(&new_cluster);
+		check_for_parameter_settings(&new_cluster);
+	}
+
 	check_new_cluster_is_empty();
 
 	check_loadable_libraries();
@@ -364,6 +380,22 @@ check_new_cluster_is_empty(void)
 						 rel_arr->rels[relnum].nspname,
 						 rel_arr->rels[relnum].relname);
 		}
+
+		/*
+		 * If --include-logical-replication-slots is required, check the
+		 * existence of slots
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			LogicalSlotInfoArr *slot_arr = &new_cluster.dbarr.dbs[dbnum].slot_arr;
+
+			/* if nslots > 0, report just first entry and exit */
+			if (slot_arr->nslots)
+				pg_fatal("New cluster database \"%s\" is not empty: found logical replication slot \"%s\"",
+						 new_cluster.dbarr.dbs[dbnum].db_name,
+						 slot_arr->slots[0].slotname);
+		}
+
 	}
 }
 
@@ -1402,3 +1434,38 @@ check_for_user_defined_encoding_conversions(ClusterInfo *cluster)
 	else
 		check_ok();
 }
+
+/*
+ * Verify parameter settings for creating logical replication slots
+ */
+static void
+check_for_parameter_settings(ClusterInfo *new_cluster)
+{
+	PGresult   *res;
+	PGconn	   *conn = connectToServer(new_cluster, "template1");
+	int			max_replication_slots;
+	char	   *wal_level;
+
+	prep_status("Checking for logical replication slots");
+
+	res = executeQueryOrDie(conn, "SHOW max_replication_slots;");
+	max_replication_slots = atoi(PQgetvalue(res, 0, 0));
+
+	if (max_replication_slots == 0)
+		pg_fatal("max_replication_slots must be greater than 0");
+
+	PQclear(res);
+
+	res = executeQueryOrDie(conn, "SHOW wal_level;");
+	wal_level = PQgetvalue(res, 0, 0);
+
+	if (strcmp(wal_level, "logical") != 0)
+		pg_fatal("wal_level must be \"logical\", but is set to \"%s\"",
+				 wal_level);
+
+	PQclear(res);
+
+	PQfinish(conn);
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/dump.c b/src/bin/pg_upgrade/dump.c
index 6c8c82dca8..e6b90864f5 100644
--- a/src/bin/pg_upgrade/dump.c
+++ b/src/bin/pg_upgrade/dump.c
@@ -59,6 +59,30 @@ generate_old_dump(void)
 						   log_opts.dumpdir,
 						   sql_file_name, escaped_connstr.data);
 
+		/*
+		 * Dump logical replication slots if needed.
+		 *
+		 * XXX We cannot dump replication slots at the same time as the schema
+		 * dump because we need to separate the timing of restoring
+		 * replication slots and other objects. Replication slots, in
+		 * particular, should not be restored before executing the pg_resetwal
+		 * command because it will remove WALs that are required by the slots.
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			char		slots_file_name[MAXPGPATH];
+
+			snprintf(slots_file_name, sizeof(slots_file_name),
+					 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+			parallel_exec_prog(log_file_name, NULL,
+							   "\"%s/pg_dump\" %s --logical-replication-slots-only "
+							   "--quote-all-identifiers --binary-upgrade %s "
+							   "--file=\"%s/%s\" %s",
+							   new_cluster.bindir, cluster_conn_opts(&old_cluster),
+							   log_opts.verbose ? "--verbose" : "",
+							   log_opts.dumpdir,
+							   slots_file_name, escaped_connstr.data);
+		}
 		termPQExpBuffer(&escaped_connstr);
 	}
 
diff --git a/src/bin/pg_upgrade/info.c b/src/bin/pg_upgrade/info.c
index 85ed15ae4a..79b4738cb8 100644
--- a/src/bin/pg_upgrade/info.c
+++ b/src/bin/pg_upgrade/info.c
@@ -23,10 +23,11 @@ static void free_db_and_rel_infos(DbInfoArr *db_arr);
 static void get_template0_info(ClusterInfo *cluster);
 static void get_db_infos(ClusterInfo *cluster);
 static void get_rel_infos(ClusterInfo *cluster, DbInfo *dbinfo);
+static void get_logical_slot_infos_per_db(ClusterInfo *cluster, DbInfo *dbinfo);
 static void free_rel_infos(RelInfoArr *rel_arr);
 static void print_db_infos(DbInfoArr *db_arr);
 static void print_rel_infos(RelInfoArr *rel_arr);
-
+static void print_slot_infos(LogicalSlotInfoArr *slot_arr);
 
 /*
  * gen_db_file_maps()
@@ -600,6 +601,95 @@ get_rel_infos(ClusterInfo *cluster, DbInfo *dbinfo)
 	dbinfo->rel_arr.nrels = num_rels;
 }
 
+/*
+ * get_logical_slot_infos()
+ *
+ * Higher level routine to generate LogicalSlotInfoArr for all databases.
+ */
+void
+get_logical_slot_infos(ClusterInfo *cluster)
+{
+	int			dbnum;
+
+	if (cluster == &old_cluster)
+		pg_log(PG_VERBOSE, "\nsource databases:");
+	else
+		pg_log(PG_VERBOSE, "\ntarget databases:");
+
+	for (dbnum = 0; dbnum < cluster->dbarr.ndbs; dbnum++)
+	{
+		DbInfo *pDbInfo = &cluster->dbarr.dbs[dbnum];
+
+		get_logical_slot_infos_per_db(cluster, pDbInfo);
+
+		if (log_opts.verbose)
+		{
+			pg_log(PG_VERBOSE, "Database: %s", pDbInfo->db_name);
+			print_slot_infos(&pDbInfo->slot_arr);
+		}
+	}
+}
+
+/*
+ * get_logical_slot_infos_per_db()
+ *
+ * gets the LogicalSlotInfos for all the logical replication slots of the database
+ * referred to by "dbinfo".
+ */
+static void
+get_logical_slot_infos_per_db(ClusterInfo *cluster, DbInfo *dbinfo)
+{
+	PGconn	   *conn = connectToServer(cluster,
+									   dbinfo->db_name);
+	PGresult   *res;
+	LogicalSlotInfo *slotinfos = NULL;
+
+	int			ntups;
+	int			num_slots = 0;
+
+	char		query[QUERY_ALLOC];
+
+	query[0] = '\0';			/* initialize query string to empty */
+
+	snprintf(query, sizeof(query),
+			 "SELECT slot_name, plugin, two_phase "
+			 "FROM pg_catalog.pg_replication_slots "
+			 "WHERE database = current_database() AND temporary = false "
+			 "AND wal_status IN ('reserved', 'extended');");
+
+	res = executeQueryOrDie(conn, "%s", query);
+
+	ntups = PQntuples(res);
+
+	if (ntups)
+	{
+		int			slotnum;
+		int			i_slotname;
+		int			i_plugin;
+		int			i_twophase;
+
+		slotinfos = (LogicalSlotInfo *) pg_malloc(sizeof(LogicalSlotInfo) * ntups);
+
+		i_slotname = PQfnumber(res, "slot_name");
+		i_plugin = PQfnumber(res, "plugin");
+		i_twophase = PQfnumber(res, "two_phase");
+
+		for (slotnum = 0; slotnum < ntups; slotnum++)
+		{
+			LogicalSlotInfo *curr = &slotinfos[num_slots++];
+
+			curr->slotname = pg_strdup(PQgetvalue(res, slotnum, i_slotname));
+			curr->plugin = pg_strdup(PQgetvalue(res, slotnum, i_plugin));
+			curr->two_phase = (strcmp(PQgetvalue(res, slotnum, i_twophase), "t") == 0);
+		}
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	dbinfo->slot_arr.slots = slotinfos;
+	dbinfo->slot_arr.nslots = num_slots;
+}
 
 static void
 free_db_and_rel_infos(DbInfoArr *db_arr)
@@ -610,6 +700,14 @@ free_db_and_rel_infos(DbInfoArr *db_arr)
 	{
 		free_rel_infos(&db_arr->dbs[dbnum].rel_arr);
 		pg_free(db_arr->dbs[dbnum].db_name);
+
+		/*
+		 * db_arr has an additional attribute, LogicalSlotInfoArr slot_arr, but
+		 * there is no need to free it. It has a valid member only when the
+		 * cluster had logical replication slots in the previous call. However,
+		 * in this case, a FATAL error is thrown, and we cannot reach this
+		 * point.
+		 */
 	}
 	pg_free(db_arr->dbs);
 	db_arr->dbs = NULL;
@@ -660,3 +758,15 @@ print_rel_infos(RelInfoArr *rel_arr)
 			   rel_arr->rels[relnum].reloid,
 			   rel_arr->rels[relnum].tablespace);
 }
+
+static void
+print_slot_infos(LogicalSlotInfoArr *slot_arr)
+{
+	int			slotnum;
+
+	for (slotnum = 0; slotnum < slot_arr->nslots; slotnum++)
+		pg_log(PG_VERBOSE, "slotname: \"%s\", plugin: \"%s\", two_phase: %d",
+			   slot_arr->slots[slotnum].slotname,
+			   slot_arr->slots[slotnum].plugin,
+			   slot_arr->slots[slotnum].two_phase);
+}
diff --git a/src/bin/pg_upgrade/meson.build b/src/bin/pg_upgrade/meson.build
index 12a97f84e2..228f29b688 100644
--- a/src/bin/pg_upgrade/meson.build
+++ b/src/bin/pg_upgrade/meson.build
@@ -42,6 +42,7 @@ tests += {
     'tests': [
       't/001_basic.pl',
       't/002_pg_upgrade.pl',
+      't/003_logical_replication_slots.pl',
     ],
     'test_kwargs': {'priority': 40}, # pg_upgrade tests are slow
   },
diff --git a/src/bin/pg_upgrade/option.c b/src/bin/pg_upgrade/option.c
index 640361009e..df66a5ffe6 100644
--- a/src/bin/pg_upgrade/option.c
+++ b/src/bin/pg_upgrade/option.c
@@ -57,6 +57,7 @@ parseCommandLine(int argc, char *argv[])
 		{"verbose", no_argument, NULL, 'v'},
 		{"clone", no_argument, NULL, 1},
 		{"copy", no_argument, NULL, 2},
+		{"include-logical-replication-slots", no_argument, NULL, 3},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -199,6 +200,10 @@ parseCommandLine(int argc, char *argv[])
 				user_opts.transfer_mode = TRANSFER_MODE_COPY;
 				break;
 
+			case 3:
+				user_opts.include_logical_slots = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						os_info.progname);
@@ -289,6 +294,8 @@ usage(void)
 	printf(_("  -V, --version                 display version information, then exit\n"));
 	printf(_("  --clone                       clone instead of copying files to new cluster\n"));
 	printf(_("  --copy                        copy files to new cluster (default)\n"));
+	printf(_("  --include-logical-replication-slots\n"
+			 "                                upgrade logical replication slots\n"));
 	printf(_("  -?, --help                    show this help, then exit\n"));
 	printf(_("\n"
 			 "Before running pg_upgrade you must:\n"
diff --git a/src/bin/pg_upgrade/pg_upgrade.c b/src/bin/pg_upgrade/pg_upgrade.c
index 75bab0a04c..373a9ef490 100644
--- a/src/bin/pg_upgrade/pg_upgrade.c
+++ b/src/bin/pg_upgrade/pg_upgrade.c
@@ -59,6 +59,7 @@ static void copy_xact_xlog_xid(void);
 static void set_frozenxids(bool minmxid_only);
 static void make_outputdirs(char *pgdata);
 static void setup(char *argv0, bool *live_check);
+static void create_logical_replication_slots(void);
 
 ClusterInfo old_cluster,
 			new_cluster;
@@ -188,6 +189,19 @@ main(int argc, char **argv)
 			  new_cluster.pgdata);
 	check_ok();
 
+	/*
+	 * Create logical replication slots if requested.
+	 *
+	 * Note: This must be done after doing pg_resetwal command because the
+	 * command will remove required WALs.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		start_postmaster(&new_cluster, true);
+		create_logical_replication_slots();
+		stop_postmaster(false);
+	}
+
 	if (user_opts.do_sync)
 	{
 		prep_status("Sync data directory to disk");
@@ -860,3 +874,50 @@ set_frozenxids(bool minmxid_only)
 
 	check_ok();
 }
+
+/*
+ * create_logical_replication_slots()
+ *
+ * Similar to create_new_objects() but only restores logical replication slots.
+ */
+static void
+create_logical_replication_slots(void)
+{
+	int			dbnum;
+
+	prep_status_progress("Restoring logical replication slots in the new cluster");
+
+	for (dbnum = 0; dbnum < old_cluster.dbarr.ndbs; dbnum++)
+	{
+		char		slots_file_name[MAXPGPATH],
+					log_file_name[MAXPGPATH];
+		DbInfo	   *old_db = &old_cluster.dbarr.dbs[dbnum];
+
+		pg_log(PG_STATUS, "%s", old_db->db_name);
+
+		snprintf(slots_file_name, sizeof(slots_file_name),
+				 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+		snprintf(log_file_name, sizeof(log_file_name),
+				 DB_DUMP_LOG_FILE_MASK, old_db->db_oid);
+
+		parallel_exec_prog(log_file_name,
+						   NULL,
+						   "\"%s/psql\" %s --echo-queries --set ON_ERROR_STOP=on "
+						   "--no-psqlrc --dbname %s -f \"%s/%s\"",
+						   new_cluster.bindir,
+						   cluster_conn_opts(&new_cluster),
+						   old_db->db_name,
+						   log_opts.dumpdir,
+						   slots_file_name);
+	}
+
+	/* reap all children */
+	while (reap_child(true) == true)
+		;
+
+	end_progress_output();
+	check_ok();
+
+	/* update new_cluster info again */
+	get_logical_slot_infos(&new_cluster);
+}
diff --git a/src/bin/pg_upgrade/pg_upgrade.h b/src/bin/pg_upgrade/pg_upgrade.h
index 3eea0139c7..7adbb50807 100644
--- a/src/bin/pg_upgrade/pg_upgrade.h
+++ b/src/bin/pg_upgrade/pg_upgrade.h
@@ -29,6 +29,7 @@
 /* contains both global db information and CREATE DATABASE commands */
 #define GLOBALS_DUMP_FILE	"pg_upgrade_dump_globals.sql"
 #define DB_DUMP_FILE_MASK	"pg_upgrade_dump_%u.custom"
+#define DB_DUMP_LOGICAL_SLOTS_FILE_MASK	"pg_upgrade_dump_%u_logical_slots.sql"
 
 /*
  * Base directories that include all the files generated internally, from the
@@ -150,6 +151,22 @@ typedef struct
 	int			nrels;
 } RelInfoArr;
 
+/*
+ * Structure to store logical replication slot information
+ */
+typedef struct
+{
+	char	   *slotname;		/* slot name */
+	char	   *plugin;			/* plugin */
+	bool		two_phase;		/* Can the slot decode 2PC? */
+} LogicalSlotInfo;
+
+typedef struct
+{
+	LogicalSlotInfo *slots;
+	int			nslots;
+} LogicalSlotInfoArr;
+
 /*
  * The following structure represents a relation mapping.
  */
@@ -176,6 +193,7 @@ typedef struct
 	char		db_tablespace[MAXPGPATH];	/* database default tablespace
 											 * path */
 	RelInfoArr	rel_arr;		/* array of all user relinfos */
+	LogicalSlotInfoArr slot_arr;	/* array of all LogicalSlotInfo */
 } DbInfo;
 
 /*
@@ -304,6 +322,8 @@ typedef struct
 	transferMode transfer_mode; /* copy files or link them? */
 	int			jobs;			/* number of processes/threads to use */
 	char	   *socketdir;		/* directory to use for Unix sockets */
+	bool		include_logical_slots;	/* true -> dump and restore logical
+										 * replication slots */
 } UserOpts;
 
 typedef struct
@@ -400,6 +420,7 @@ FileNameMap *gen_db_file_maps(DbInfo *old_db,
 							  DbInfo *new_db, int *nmaps, const char *old_pgdata,
 							  const char *new_pgdata);
 void		get_db_and_rel_infos(ClusterInfo *cluster);
+void		get_logical_slot_infos(ClusterInfo *cluster);
 
 /* option.c */
 
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
new file mode 100644
index 0000000000..525a7704cf
--- /dev/null
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -0,0 +1,115 @@
+# Copyright (c) 2023, PostgreSQL Global Development Group
+
+# Tests for upgrading replication slots
+
+use strict;
+use warnings;
+
+use File::Path qw(rmtree);
+
+use PostgreSQL::Test::Cluster;
+use PostgreSQL::Test::Utils;
+use Test::More;
+
+# Can be changed to test the other modes
+my $mode = $ENV{PG_TEST_PG_UPGRADE_MODE} || '--copy';
+
+# Initialize old node
+my $old_node = PostgreSQL::Test::Cluster->new('old_node');
+$old_node->init(allows_streaming => 'logical');
+$old_node->start;
+
+# Initialize new node
+my $new_node = PostgreSQL::Test::Cluster->new('new_node');
+$new_node->init(allows_streaming => 1);
+
+my $bindir = $new_node->config_data('--bindir');
+
+$old_node->stop;
+
+# Cause a failure at the start of pg_upgrade because wal_level is replica
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with wrong wal_level');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+# Preparations for the subsequent test. The case max_replication_slots is set
+# to 0 is prohibited.
+$new_node->append_conf('postgresql.conf', "wal_level = 'logical'");
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 0");
+
+# Cause a failure at the start of pg_upgrade because max_replication_slots is 0
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with wrong max_replication_slots');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+# Preparations for the subsequent test. max_replication_slots is set to
+# non-zero value to succeed the pg_upgrade
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 10");
+
+# Create a slot on old node, and generate WALs
+$old_node->start;
+$old_node->safe_psql(
+	'postgres', qq[
+	SELECT pg_create_logical_replication_slot('test_slot', 'test_decoding', false, true);
+	CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a;
+]);
+
+my $result = $old_node->safe_psql('postgres',
+	"SELECT count (*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)"
+);
+is($result, qq(12), 'ensure WALs are not consumed yet');
+$old_node->stop;
+
+# Actual run, pg_upgrade_output.d is removed at the end
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots'
+	],
+	'run of pg_upgrade of old node');
+ok( !-d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+$new_node->start;
+$result = $new_node->safe_psql('postgres',
+	"SELECT slot_name, two_phase FROM pg_replication_slots");
+is($result, qq(test_slot|t), 'check the slot exists on new node');
+
+done_testing();
diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list
index b4058b88c3..5944cb34ea 100644
--- a/src/tools/pgindent/typedefs.list
+++ b/src/tools/pgindent/typedefs.list
@@ -1479,6 +1479,7 @@ LogicalRepBeginData
 LogicalRepCommitData
 LogicalRepCommitPreparedTxnData
 LogicalRepCtxStruct
+LogicalReplicationSlotInfo
 LogicalRepMode
 LogicalRepMsgType
 LogicalRepPartMapEntry
@@ -1492,6 +1493,8 @@ LogicalRepTupleData
 LogicalRepTyp
 LogicalRepWorker
 LogicalRewriteMappingData
+LogicalSlotInfo
+LogicalSlotInfoArr
 LogicalTape
 LogicalTapeSet
 LsnReadQueue
-- 
2.27.0

From 582ae94c099a3d33fe51f0fb32eff89c5b273ed0 Mon Sep 17 00:00:00 2001
From: Julien Rouhaud <julien.rouhaud@free.fr>
Date: Fri, 14 Apr 2023 13:49:09 +0800
Subject: [PATCH v13 2/4] Always persist to disk logical slots during a
 shutdown checkpoint.

It's entirely possible for a logical slot to have a confirmed_flush_lsn higher
than the last value saved on disk while not being marked as dirty.  It's
currently not a problem to lose that value during a clean shutdown / restart
cycle, but a later patch adding support for pg_upgrade of publications and
logical slots will rely on that value being properly persisted to disk.

Author: Julien Rouhaud
Reviewed-by: FIXME
Discussion: FIXME
---
 src/backend/access/transam/xlog.c |  2 +-
 src/backend/replication/slot.c    | 26 ++++++++++++++++----------
 src/include/replication/slot.h    |  2 +-
 3 files changed, 18 insertions(+), 12 deletions(-)

diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index bc5a8e0569..78b4528f2c 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7011,7 +7011,7 @@ static void
 CheckPointGuts(XLogRecPtr checkPointRedo, int flags)
 {
 	CheckPointRelationMap();
-	CheckPointReplicationSlots();
+	CheckPointReplicationSlots(flags & CHECKPOINT_IS_SHUTDOWN);
 	CheckPointSnapBuild();
 	CheckPointLogicalRewriteHeap();
 	CheckPointReplicationOrigin();
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index 8021aaa0a8..aeea6ffd1f 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -109,7 +109,8 @@ static void ReplicationSlotDropPtr(ReplicationSlot *slot);
 /* internal persistency functions */
 static void RestoreSlotFromDisk(const char *name);
 static void CreateSlotOnDisk(ReplicationSlot *slot);
-static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel);
+static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+						   bool is_shutdown);
 
 /*
  * Report shared-memory space needed by ReplicationSlotsShmemInit.
@@ -783,7 +784,7 @@ ReplicationSlotSave(void)
 	Assert(MyReplicationSlot != NULL);
 
 	sprintf(path, "pg_replslot/%s", NameStr(MyReplicationSlot->data.name));
-	SaveSlotToPath(MyReplicationSlot, path, ERROR);
+	SaveSlotToPath(MyReplicationSlot, path, ERROR, false);
 }
 
 /*
@@ -1565,11 +1566,10 @@ restart:
 /*
  * Flush all replication slots to disk.
  *
- * This needn't actually be part of a checkpoint, but it's a convenient
- * location.
+ * is_shutdown is true in case of a shutdown checkpoint.
  */
 void
-CheckPointReplicationSlots(void)
+CheckPointReplicationSlots(bool is_shutdown)
 {
 	int			i;
 
@@ -1594,7 +1594,7 @@ CheckPointReplicationSlots(void)
 
 		/* save the slot to disk, locking is handled in SaveSlotToPath() */
 		sprintf(path, "pg_replslot/%s", NameStr(s->data.name));
-		SaveSlotToPath(s, path, LOG);
+		SaveSlotToPath(s, path, LOG, is_shutdown);
 	}
 	LWLockRelease(ReplicationSlotAllocationLock);
 }
@@ -1700,7 +1700,7 @@ CreateSlotOnDisk(ReplicationSlot *slot)
 
 	/* Write the actual state file. */
 	slot->dirty = true;			/* signal that we really need to write */
-	SaveSlotToPath(slot, tmppath, ERROR);
+	SaveSlotToPath(slot, tmppath, ERROR, false);
 
 	/* Rename the directory into place. */
 	if (rename(tmppath, path) != 0)
@@ -1726,7 +1726,8 @@ CreateSlotOnDisk(ReplicationSlot *slot)
  * Shared functionality between saving and creating a replication slot.
  */
 static void
-SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
+SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+			   bool is_shutdown)
 {
 	char		tmppath[MAXPGPATH];
 	char		path[MAXPGPATH];
@@ -1740,8 +1741,13 @@ SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
 	slot->just_dirtied = false;
 	SpinLockRelease(&slot->mutex);
 
-	/* and don't do anything if there's nothing to write */
-	if (!was_dirty)
+	/*
+	 * and don't do anything if there's nothing to write, unless it's this is
+	 * called for a logical slot during a shutdown checkpoint, as we want to
+	 * persist the confirmed_flush_lsn in that case, even if that's the only
+	 * modification.
+	 */
+	if (!was_dirty && !is_shutdown && !SlotIsLogical(slot))
 		return;
 
 	LWLockAcquire(&slot->io_in_progress_lock, LW_EXCLUSIVE);
diff --git a/src/include/replication/slot.h b/src/include/replication/slot.h
index a8a89dc784..7ca37c9f70 100644
--- a/src/include/replication/slot.h
+++ b/src/include/replication/slot.h
@@ -241,7 +241,7 @@ extern void ReplicationSlotNameForTablesync(Oid suboid, Oid relid, char *syncslo
 extern void ReplicationSlotDropAtPubNode(WalReceiverConn *wrconn, char *slotname, bool missing_ok);
 
 extern void StartupReplicationSlots(void);
-extern void CheckPointReplicationSlots(void);
+extern void CheckPointReplicationSlots(bool is_shutdown);
 
 extern void CheckSlotRequirements(void);
 extern void CheckSlotPermissions(void);
-- 
2.27.0

From db7658c30223a2fe131d76c40f8b70011a650760 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Fri, 14 Apr 2023 08:59:03 +0000
Subject: [PATCH v13 3/4] pg_upgrade: Add check function for
 --include-logical-replication-slots option

---
 src/bin/pg_upgrade/check.c                    | 80 ++++++++++++++++++-
 .../t/003_logical_replication_slots.pl        | 30 ++++++-
 2 files changed, 108 insertions(+), 2 deletions(-)

diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index f9cba8548e..61ca36a853 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -9,7 +9,10 @@
 
 #include "postgres_fe.h"
 
+#include "access/xlogrecord.h"
+#include "access/xlog_internal.h"
 #include "catalog/pg_authid_d.h"
+#include "catalog/pg_control.h"
 #include "catalog/pg_collation.h"
 #include "fe_utils/string_utils.h"
 #include "mb/pg_wchar.h"
@@ -31,7 +34,7 @@ static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
 static void check_for_parameter_settings(ClusterInfo *new_cluster);
-
+static void check_for_confirmed_flush_lsn(ClusterInfo *cluster);
 
 /*
  * fix_path_separator
@@ -108,6 +111,8 @@ check_and_dump_old_cluster(bool live_check)
 	check_for_composite_data_type_usage(&old_cluster);
 	check_for_reg_data_type_usage(&old_cluster);
 	check_for_isn_and_int8_passing_mismatch(&old_cluster);
+	if (user_opts.include_logical_slots)
+		check_for_confirmed_flush_lsn(&old_cluster);
 
 	/*
 	 * PG 16 increased the size of the 'aclitem' type, which breaks the on-disk
@@ -1446,6 +1451,10 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 	int			max_replication_slots;
 	char	   *wal_level;
 
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(new_cluster->major_version < 1600))
+		return;
+
 	prep_status("Checking for logical replication slots");
 
 	res = executeQueryOrDie(conn, "SHOW max_replication_slots;");
@@ -1469,3 +1478,72 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 
 	check_ok();
 }
+
+/*
+ * Verify that all logical replication slots consumed all WALs, except a
+ * CHECKPOINT_SHUTDOWN record.
+ */
+static void
+check_for_confirmed_flush_lsn(ClusterInfo *cluster)
+{
+	int			i,
+				ntups,
+				i_slotname;
+	bool		is_error = false;
+	PGresult   *res;
+	DbInfo	   *active_db = &cluster->dbarr.dbs[0];
+	PGconn	   *conn = connectToServer(cluster, active_db->db_name);
+
+	Assert(user_opts.include_logical_slots);
+
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(cluster->major_version < 1600))
+		return;
+
+	prep_status("Checking for logical replication slots");
+
+	/*
+	 * Check that all logical replication slots have reached the current WAL
+	 * position, except for the CHECKPOINT_SHUTDOWN record. Even if all WALs
+	 * are consumed before shutting down the node, the checkpointer generates
+	 * a CHECKPOINT_SHUTDOWN record at shutdown, which cannot be consumed by
+	 * any slots. Therefore, we must allow for a difference between
+	 * pg_current_wal_insert_lsn() and confirmed_flush_lsn.
+	 */
+#define SHUTDOWN_RECORD_SIZE  (SizeOfXLogRecord + \
+							   SizeOfXLogRecordDataHeaderShort + \
+							   sizeof(CheckPoint))
+
+	res = executeQueryOrDie(conn,
+							"SELECT slot_name FROM pg_catalog.pg_replication_slots "
+							"WHERE (pg_catalog.pg_current_wal_insert_lsn() - confirmed_flush_lsn) > %d "
+							"AND temporary = false AND wal_status IN ('reserved', 'extended');",
+							(int) (SizeOfXLogLongPHD + SHUTDOWN_RECORD_SIZE));
+
+#undef SHUTDOWN_RECORD_SIZE
+
+	ntups = PQntuples(res);
+	i_slotname = PQfnumber(res, "slot_name");
+
+	for (i = 0; i < ntups; i++)
+	{
+		char	   *slotname;
+
+		is_error = true;
+
+		slotname = PQgetvalue(res, i, i_slotname);
+
+		pg_log(PG_WARNING,
+			   "\nWARNING: logical replication slot \"%s\" has not consumed WALs yet",
+			   slotname);
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	if (is_error)
+		pg_fatal("--include-logical-replication-slots requires that all "
+				 "logical replication slots consumed all the WALs");
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
index 525a7704cf..21fefca084 100644
--- a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -85,11 +85,39 @@ $old_node->safe_psql(
 ]);
 
 my $result = $old_node->safe_psql('postgres',
-	"SELECT count (*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)"
+	"SELECT count(*) FROM pg_logical_slot_peek_changes('test_slot', NULL, NULL)"
 );
+
 is($result, qq(12), 'ensure WALs are not consumed yet');
 $old_node->stop;
 
+# Cause a failure at the start of pg_upgrade because test_slot does not
+# finish consuming all the WALs
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with idle replication slots');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+$old_node->start;
+$old_node->safe_psql('postgres',
+	"SELECT count (*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)"
+);
+$old_node->stop;
+
 # Actual run, pg_upgrade_output.d is removed at the end
 command_ok(
 	[
-- 
2.27.0

From 6c86780aadf28d67e2571095ffd9ef2b3e3370c5 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Mon, 24 Apr 2023 11:03:28 +0000
Subject: [PATCH v13 4/4] Change the method used to check logical replication
 slots during the live check

When a live check is requested, there is a possibility of additional changes
occurring, which may cause the current WAL position to exceed the confirmed_flush_lsn
of the slot. As a result, we check the confirmed_flush_lsn of each logical slot
instead. This is sufficient as all the WAL records will be sent during thepublisher's
shutdown.
---
 src/bin/pg_upgrade/check.c                    | 68 ++++++++++++++++++-
 .../t/003_logical_replication_slots.pl        | 66 +++++++++++++++++-
 2 files changed, 130 insertions(+), 4 deletions(-)

diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index 61ca36a853..528576b00e 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -35,6 +35,7 @@ static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
 static void check_for_parameter_settings(ClusterInfo *new_cluster);
 static void check_for_confirmed_flush_lsn(ClusterInfo *cluster);
+static void check_are_logical_slots_active(ClusterInfo *cluster);
 
 /*
  * fix_path_separator
@@ -112,7 +113,19 @@ check_and_dump_old_cluster(bool live_check)
 	check_for_reg_data_type_usage(&old_cluster);
 	check_for_isn_and_int8_passing_mismatch(&old_cluster);
 	if (user_opts.include_logical_slots)
-		check_for_confirmed_flush_lsn(&old_cluster);
+	{
+		/*
+		 * The method used to check logical replication slots is dependent on
+		 * the value of the live_check parameter. This change was implemented
+		 * because, during a live check, it is possible for additional changes
+		 * to occur at the old node, which could cause the current WAL position
+		 * to exceed the confirmed_flush_lsn of the slot.
+		 */
+		if (live_check)
+			check_are_logical_slots_active(&old_cluster);
+		else
+			check_for_confirmed_flush_lsn(&old_cluster);
+	}
 
 	/*
 	 * PG 16 increased the size of the 'aclitem' type, which breaks the on-disk
@@ -1479,6 +1492,59 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 	check_ok();
 }
 
+/*
+ * Verify that all logical replication slots are active
+ */
+static void
+check_are_logical_slots_active(ClusterInfo *cluster)
+{
+	int			i,
+				ntups,
+				i_slotname;
+	bool		is_error = false;
+	PGresult   *res;
+	DbInfo	   *active_db = &cluster->dbarr.dbs[0];
+	PGconn	   *conn = connectToServer(cluster, active_db->db_name);
+
+	Assert(user_opts.include_logical_slots);
+
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(cluster->major_version < 1600))
+		return;
+
+	prep_status("Checking for logical replication slots");
+
+	res = executeQueryOrDie(conn,
+							"SELECT slot_name FROM pg_catalog.pg_replication_slots "
+							"WHERE active IS FALSE "
+							"AND temporary = false AND wal_status IN ('reserved', 'extended');");
+
+	ntups = PQntuples(res);
+	i_slotname = PQfnumber(res, "slot_name");
+
+	for (i = 0; i < ntups; i++)
+	{
+		char	   *slotname;
+
+		is_error = true;
+
+		slotname = PQgetvalue(res, i, i_slotname);
+
+		pg_log(PG_WARNING,
+			   "\nWARNING: logical replication slot \"%s\" is not active",
+			   slotname);
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	if (is_error)
+		pg_fatal("--include-logical-replication-slots with --check requires that "
+				 "all logical replication slots are active");
+
+	check_ok();
+}
+
 /*
  * Verify that all logical replication slots consumed all WALs, except a
  * CHECKPOINT_SHUTDOWN record.
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
index 21fefca084..d7fd864bd7 100644
--- a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -19,6 +19,10 @@ my $old_node = PostgreSQL::Test::Cluster->new('old_node');
 $old_node->init(allows_streaming => 'logical');
 $old_node->start;
 
+# Initialize subscriber, which will be used only for --check
+my $subscriber = PostgreSQL::Test::Cluster->new('subscriber');
+$subscriber->init(allows_streaming => 'logical');
+
 # Initialize new node
 my $new_node = PostgreSQL::Test::Cluster->new('new_node');
 $new_node->init(allows_streaming => 1);
@@ -76,15 +80,71 @@ rmtree($new_node->data_dir . "/pg_upgrade_output.d");
 # non-zero value to succeed the pg_upgrade
 $new_node->append_conf('postgresql.conf', "max_replication_slots = 10");
 
-# Create a slot on old node, and generate WALs
+# Setup logical replication
 $old_node->start;
+$old_node->safe_psql('postgres',
+	"CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a");
+$old_node->safe_psql('postgres', "CREATE PUBLICATION pub FOR ALL TABLES");
+my $old_connstr = $old_node->connstr . ' dbname=postgres';
+
+$subscriber->start;
+$subscriber->safe_psql('postgres', "CREATE TABLE tbl (a int)");
+$subscriber->safe_psql('postgres',
+	"CREATE SUBSCRIPTION sub CONNECTION '$old_connstr' PUBLICATION pub WITH (copy_data = true)"
+);
+
+# Wait for initial table sync to finish
+$subscriber->wait_for_subscription_sync($old_node, 'sub');
+
+my $result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(10), 'check initial rows on subscriber');
+
+# Start a background session and open a transaction (not committed yet)
+my $bsession = $old_node->background_psql('postgres');
+$bsession->query_safe(
+	q{
+BEGIN;
+INSERT INTO tbl VALUES (generate_series(11, 20))
+});
+
+$result = $old_node->safe_psql('postgres',
+	"SELECT count(*) FROM pg_replication_slots WHERE pg_current_wal_insert_lsn() > confirmed_flush_lsn"
+);
+is($result, qq(1),
+	'check the current WAL position exceeds confirmed_flush_lsn');
+
+# Run pg_upgrade --check. In the command the status of each logical slots will
+# be checked and then this will be succeeded.
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+		'--check'
+	],
+	'run of pg_upgrade of old node');
+ok( !-d $old_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+# Cleanup
+$bsession->query_safe("ABORT");
+$bsession->quit;
+$subscriber->safe_psql('postgres', "DROP SUBSCRIPTION sub");
+
+# Create a slot on old node, and generate WALs
 $old_node->safe_psql(
 	'postgres', qq[
 	SELECT pg_create_logical_replication_slot('test_slot', 'test_decoding', false, true);
-	CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a;
+	INSERT INTO tbl VALUES (generate_series(11, 20));
 ]);
 
-my $result = $old_node->safe_psql('postgres',
+$result = $old_node->safe_psql('postgres',
 	"SELECT count(*) FROM pg_logical_slot_peek_changes('test_slot', NULL, NULL)"
 );
 
-- 
2.27.0

From bd94104de8a846ce8762b79fbdb4829691c57475 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Mon, 24 Apr 2023 11:03:28 +0000
Subject: [PATCH v14 4/4] Change the method used to check logical replication
 slots during the live check

When a live check is requested, there is a possibility of additional changes
occurring, which may cause the current WAL position to exceed the confirmed_flush_lsn
of the slot. As a result, we check the confirmed_flush_lsn of each logical slot
instead. This is sufficient as all the WAL records will be sent during thepublisher's
shutdown.
---
 src/bin/pg_upgrade/check.c                    | 68 ++++++++++++++++++-
 .../t/003_logical_replication_slots.pl        | 66 +++++++++++++++++-
 2 files changed, 130 insertions(+), 4 deletions(-)

diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index 61ca36a853..528576b00e 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -35,6 +35,7 @@ static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
 static void check_for_parameter_settings(ClusterInfo *new_cluster);
 static void check_for_confirmed_flush_lsn(ClusterInfo *cluster);
+static void check_are_logical_slots_active(ClusterInfo *cluster);
 
 /*
  * fix_path_separator
@@ -112,7 +113,19 @@ check_and_dump_old_cluster(bool live_check)
 	check_for_reg_data_type_usage(&old_cluster);
 	check_for_isn_and_int8_passing_mismatch(&old_cluster);
 	if (user_opts.include_logical_slots)
-		check_for_confirmed_flush_lsn(&old_cluster);
+	{
+		/*
+		 * The method used to check logical replication slots is dependent on
+		 * the value of the live_check parameter. This change was implemented
+		 * because, during a live check, it is possible for additional changes
+		 * to occur at the old node, which could cause the current WAL position
+		 * to exceed the confirmed_flush_lsn of the slot.
+		 */
+		if (live_check)
+			check_are_logical_slots_active(&old_cluster);
+		else
+			check_for_confirmed_flush_lsn(&old_cluster);
+	}
 
 	/*
 	 * PG 16 increased the size of the 'aclitem' type, which breaks the on-disk
@@ -1479,6 +1492,59 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 	check_ok();
 }
 
+/*
+ * Verify that all logical replication slots are active
+ */
+static void
+check_are_logical_slots_active(ClusterInfo *cluster)
+{
+	int			i,
+				ntups,
+				i_slotname;
+	bool		is_error = false;
+	PGresult   *res;
+	DbInfo	   *active_db = &cluster->dbarr.dbs[0];
+	PGconn	   *conn = connectToServer(cluster, active_db->db_name);
+
+	Assert(user_opts.include_logical_slots);
+
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(cluster->major_version < 1600))
+		return;
+
+	prep_status("Checking for logical replication slots");
+
+	res = executeQueryOrDie(conn,
+							"SELECT slot_name FROM pg_catalog.pg_replication_slots "
+							"WHERE active IS FALSE "
+							"AND temporary = false AND wal_status IN ('reserved', 'extended');");
+
+	ntups = PQntuples(res);
+	i_slotname = PQfnumber(res, "slot_name");
+
+	for (i = 0; i < ntups; i++)
+	{
+		char	   *slotname;
+
+		is_error = true;
+
+		slotname = PQgetvalue(res, i, i_slotname);
+
+		pg_log(PG_WARNING,
+			   "\nWARNING: logical replication slot \"%s\" is not active",
+			   slotname);
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	if (is_error)
+		pg_fatal("--include-logical-replication-slots with --check requires that "
+				 "all logical replication slots are active");
+
+	check_ok();
+}
+
 /*
  * Verify that all logical replication slots consumed all WALs, except a
  * CHECKPOINT_SHUTDOWN record.
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
index 21fefca084..d7fd864bd7 100644
--- a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -19,6 +19,10 @@ my $old_node = PostgreSQL::Test::Cluster->new('old_node');
 $old_node->init(allows_streaming => 'logical');
 $old_node->start;
 
+# Initialize subscriber, which will be used only for --check
+my $subscriber = PostgreSQL::Test::Cluster->new('subscriber');
+$subscriber->init(allows_streaming => 'logical');
+
 # Initialize new node
 my $new_node = PostgreSQL::Test::Cluster->new('new_node');
 $new_node->init(allows_streaming => 1);
@@ -76,15 +80,71 @@ rmtree($new_node->data_dir . "/pg_upgrade_output.d");
 # non-zero value to succeed the pg_upgrade
 $new_node->append_conf('postgresql.conf', "max_replication_slots = 10");
 
-# Create a slot on old node, and generate WALs
+# Setup logical replication
 $old_node->start;
+$old_node->safe_psql('postgres',
+	"CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a");
+$old_node->safe_psql('postgres', "CREATE PUBLICATION pub FOR ALL TABLES");
+my $old_connstr = $old_node->connstr . ' dbname=postgres';
+
+$subscriber->start;
+$subscriber->safe_psql('postgres', "CREATE TABLE tbl (a int)");
+$subscriber->safe_psql('postgres',
+	"CREATE SUBSCRIPTION sub CONNECTION '$old_connstr' PUBLICATION pub WITH (copy_data = true)"
+);
+
+# Wait for initial table sync to finish
+$subscriber->wait_for_subscription_sync($old_node, 'sub');
+
+my $result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(10), 'check initial rows on subscriber');
+
+# Start a background session and open a transaction (not committed yet)
+my $bsession = $old_node->background_psql('postgres');
+$bsession->query_safe(
+	q{
+BEGIN;
+INSERT INTO tbl VALUES (generate_series(11, 20))
+});
+
+$result = $old_node->safe_psql('postgres',
+	"SELECT count(*) FROM pg_replication_slots WHERE pg_current_wal_insert_lsn() > confirmed_flush_lsn"
+);
+is($result, qq(1),
+	'check the current WAL position exceeds confirmed_flush_lsn');
+
+# Run pg_upgrade --check. In the command the status of each logical slots will
+# be checked and then this will be succeeded.
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+		'--check'
+	],
+	'run of pg_upgrade of old node');
+ok( !-d $old_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+# Cleanup
+$bsession->query_safe("ABORT");
+$bsession->quit;
+$subscriber->safe_psql('postgres', "DROP SUBSCRIPTION sub");
+
+# Create a slot on old node, and generate WALs
 $old_node->safe_psql(
 	'postgres', qq[
 	SELECT pg_create_logical_replication_slot('test_slot', 'test_decoding', false, true);
-	CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a;
+	INSERT INTO tbl VALUES (generate_series(11, 20));
 ]);
 
-my $result = $old_node->safe_psql('postgres',
+$result = $old_node->safe_psql('postgres',
 	"SELECT count(*) FROM pg_logical_slot_peek_changes('test_slot', NULL, NULL)"
 );
 
-- 
2.27.0

From 01fe267cfdf81c68d01905c1cb556f9897e76017 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Tue, 4 Apr 2023 05:49:34 +0000
Subject: [PATCH v14 1/4] pg_upgrade: Add --include-logical-replication-slots
 option

This commit introduces a new pg_upgrade option called "--include-logical-replication-slots".
This allows nodes with logical replication slots to be upgraded. The commit can
be divided into two parts: one for pg_dump and another for pg_upgrade.

For pg_dump this commit includes a new option called "--logical-replication-slots-only".
This option can be used to dump logical replication slots. When this option is
specified, the slot_name, plugin, and two_phase parameters are extracted from
pg_replication_slots. An SQL file is then generated which executes
pg_create_logical_replication_slot() with the extracted parameters.

For pg_upgrade, when '--include-logical-replication-slots' is specified, it executes
pg_dump with the new "--logical-replication-slots-only" option and restores from the
dump. Note that we cannot dump replication slots at the same time as the schema
dump because we need to separate the timing of restoring replication slots and
other objects. Replication slots, in  particular, should not be restored before
executing the pg_resetwal command because it will remove WALs that are required
by the slots.

The significant advantage of this commit is that it makes it easy to continue
logical replication even after upgrading the publisher node. Previously, pg_upgrade
allowed copying publications to a new node. With this new commit, adjusting the
connection string to the new publisher will cause the apply worker on the subscriber
to connect to the new publisher automatically. This enables seamless continuation
of logical replication, even after an upgrade.

Author: Hayato Kuroda
Reviewed-by: Peter Smith, Julien Rouhaud, Vignesh C
---
 doc/src/sgml/ref/pgupgrade.sgml               |  11 ++
 src/bin/pg_dump/pg_backup.h                   |   1 +
 src/bin/pg_dump/pg_dump.c                     | 156 ++++++++++++++++++
 src/bin/pg_dump/pg_dump.h                     |  17 +-
 src/bin/pg_dump/pg_dump_sort.c                |  11 +-
 src/bin/pg_upgrade/Makefile                   |   3 +
 src/bin/pg_upgrade/check.c                    |  67 ++++++++
 src/bin/pg_upgrade/dump.c                     |  24 +++
 src/bin/pg_upgrade/info.c                     | 111 ++++++++++++-
 src/bin/pg_upgrade/meson.build                |   1 +
 src/bin/pg_upgrade/option.c                   |   7 +
 src/bin/pg_upgrade/pg_upgrade.c               |  61 +++++++
 src/bin/pg_upgrade/pg_upgrade.h               |  21 +++
 .../t/003_logical_replication_slots.pl        | 115 +++++++++++++
 src/tools/pgindent/typedefs.list              |   3 +
 15 files changed, 605 insertions(+), 4 deletions(-)
 create mode 100644 src/bin/pg_upgrade/t/003_logical_replication_slots.pl

diff --git a/doc/src/sgml/ref/pgupgrade.sgml b/doc/src/sgml/ref/pgupgrade.sgml
index 7816b4c685..94e90ff506 100644
--- a/doc/src/sgml/ref/pgupgrade.sgml
+++ b/doc/src/sgml/ref/pgupgrade.sgml
@@ -240,6 +240,17 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--include-logical-replication-slots</option></term>
+      <listitem>
+       <para>
+        Upgrade logical replication slots. Only permanent replication slots
+        are included. Note that pg_upgrade does not check the installation of
+        plugins.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
       <term><option>-?</option></term>
       <term><option>--help</option></term>
diff --git a/src/bin/pg_dump/pg_backup.h b/src/bin/pg_dump/pg_backup.h
index aba780ef4b..0a4e931f9b 100644
--- a/src/bin/pg_dump/pg_backup.h
+++ b/src/bin/pg_dump/pg_backup.h
@@ -187,6 +187,7 @@ typedef struct _dumpOptions
 	int			use_setsessauth;
 	int			enable_row_security;
 	int			load_via_partition_root;
+	int			logical_slots_only;
 
 	/* default, if no "inclusion" switches appear, is to dump everything */
 	bool		include_everything;
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index 41a51ec5cd..880ef3bd1d 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -328,6 +328,9 @@ static void setupDumpWorker(Archive *AH);
 static TableInfo *getRootTableInfo(const TableInfo *tbinfo);
 static bool forcePartitionRootLoad(const TableInfo *tbinfo);
 
+static void getLogicalReplicationSlots(Archive *fout);
+static void dumpLogicalReplicationSlot(Archive *fout,
+									   const LogicalReplicationSlotInfo *slotinfo);
 
 int
 main(int argc, char **argv)
@@ -431,6 +434,7 @@ main(int argc, char **argv)
 		{"table-and-children", required_argument, NULL, 12},
 		{"exclude-table-and-children", required_argument, NULL, 13},
 		{"exclude-table-data-and-children", required_argument, NULL, 14},
+		{"logical-replication-slots-only", no_argument, NULL, 15},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -657,6 +661,10 @@ main(int argc, char **argv)
 										  optarg);
 				break;
 
+			case 15:			/* dump only replication slot(s) */
+				dopt.logical_slots_only = true;
+				break;
+
 			default:
 				/* getopt_long already emitted a complaint */
 				pg_log_error_hint("Try \"%s --help\" for more information.", progname);
@@ -714,6 +722,21 @@ main(int argc, char **argv)
 	if (dopt.do_nothing && dopt.dump_inserts == 0)
 		pg_fatal("option --on-conflict-do-nothing requires option --inserts, --rows-per-insert, or --column-inserts");
 
+	if (dopt.logical_slots_only)
+	{
+		if (!dopt.binary_upgrade)
+			pg_fatal("options --logical-replication-slots-only requires option --binary-upgrade");
+
+		if (dopt.dataOnly)
+			pg_fatal("options --logical-replication-slots-only and -a/--data-only cannot be used together");
+
+		if (dopt.schemaOnly)
+			pg_fatal("options --logical-replication-slots-only and -s/--schema-only cannot be used together");
+
+		if (dopt.outputClean)
+			pg_fatal("options --logical-replication-slots-only and -c/--clean cannot be used together");
+	}
+
 	/* Identify archive format to emit */
 	archiveFormat = parseArchiveFormat(format, &archiveMode);
 
@@ -876,6 +899,16 @@ main(int argc, char **argv)
 			pg_fatal("no matching extensions were found");
 	}
 
+	/*
+	 * If dump logical-replication-slots-only was requested, dump only them
+	 * and skip everything else.
+	 */
+	if (dopt.logical_slots_only)
+	{
+		getLogicalReplicationSlots(fout);
+		goto dump;
+	}
+
 	/*
 	 * Dumping LOs is the default for dumps where an inclusion switch is not
 	 * used (an "include everything" dump).  -B can be used to exclude LOs
@@ -936,6 +969,8 @@ main(int argc, char **argv)
 	if (!dopt.no_security_labels)
 		collectSecLabels(fout);
 
+dump:
+
 	/* Lastly, create dummy objects to represent the section boundaries */
 	boundaryObjs = createBoundaryObjects();
 
@@ -1109,6 +1144,11 @@ help(const char *progname)
 			 "                               servers matching PATTERN\n"));
 	printf(_("  --inserts                    dump data as INSERT commands, rather than COPY\n"));
 	printf(_("  --load-via-partition-root    load partitions via the root table\n"));
+
+	/*
+	 * The option --logical-replication-slots-only is used only by pg_upgrade
+	 * and should not be called by users, which is why it is not listed.
+	 */
 	printf(_("  --no-comments                do not dump comments\n"));
 	printf(_("  --no-publications            do not dump publications\n"));
 	printf(_("  --no-security-labels         do not dump security label assignments\n"));
@@ -10252,6 +10292,10 @@ dumpDumpableObject(Archive *fout, DumpableObject *dobj)
 		case DO_SUBSCRIPTION:
 			dumpSubscription(fout, (const SubscriptionInfo *) dobj);
 			break;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			dumpLogicalReplicationSlot(fout,
+									   (const LogicalReplicationSlotInfo *) dobj);
+			break;
 		case DO_PRE_DATA_BOUNDARY:
 		case DO_POST_DATA_BOUNDARY:
 			/* never dumped, nothing to do */
@@ -18227,6 +18271,7 @@ addBoundaryDependencies(DumpableObject **dobjs, int numObjs,
 			case DO_PUBLICATION_REL:
 			case DO_PUBLICATION_TABLE_IN_SCHEMA:
 			case DO_SUBSCRIPTION:
+			case DO_LOGICAL_REPLICATION_SLOT:
 				/* Post-data objects: must come after the post-data boundary */
 				addObjectDependency(dobj, postDataBound->dumpId);
 				break;
@@ -18488,3 +18533,114 @@ appendReloptionsArrayAH(PQExpBuffer buffer, const char *reloptions,
 	if (!res)
 		pg_log_warning("could not parse %s array", "reloptions");
 }
+
+/*
+ * getLogicalReplicationSlots
+ *	  get information about replication slots
+ */
+static void
+getLogicalReplicationSlots(Archive *fout)
+{
+	PGresult   *res;
+	LogicalReplicationSlotInfo *slotinfo;
+	PQExpBuffer query;
+
+	int			i_slotname;
+	int			i_plugin;
+	int			i_twophase;
+	int			i,
+				ntups;
+
+	/* Check whether we should dump or not */
+	if (fout->remoteVersion < 160000)
+		return;
+
+	Assert(fout->dopt->logical_slots_only);
+
+	query = createPQExpBuffer();
+
+	resetPQExpBuffer(query);
+
+	/*
+	 * Get replication slots.
+	 *
+	 * XXX: Which information must be extracted from old node? Currently three
+	 * attributes are extracted because they are used by
+	 * pg_create_logical_replication_slot().
+	 *
+	 * XXX: Do we have to support physical slots?
+	 */
+	appendPQExpBufferStr(query,
+						 "SELECT slot_name, plugin, two_phase "
+						 "FROM pg_catalog.pg_replication_slots "
+						 "WHERE database = current_database() AND temporary = false "
+						 "AND wal_status IN ('reserved', 'extended');");
+
+	res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
+
+	ntups = PQntuples(res);
+
+	i_slotname = PQfnumber(res, "slot_name");
+	i_plugin = PQfnumber(res, "plugin");
+	i_twophase = PQfnumber(res, "two_phase");
+
+	slotinfo = pg_malloc(ntups * sizeof(LogicalReplicationSlotInfo));
+
+	for (i = 0; i < ntups; i++)
+	{
+		slotinfo[i].dobj.objType = DO_LOGICAL_REPLICATION_SLOT;
+
+		slotinfo[i].dobj.catId.tableoid = InvalidOid;
+		slotinfo[i].dobj.catId.oid = InvalidOid;
+		AssignDumpId(&slotinfo[i].dobj);
+
+		slotinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_slotname));
+
+		slotinfo[i].plugin = pg_strdup(PQgetvalue(res, i, i_plugin));
+		slotinfo[i].twophase = (strcmp(PQgetvalue(res, i, i_twophase), "t") == 0);
+
+		/*
+		 * Note: Currently we do not have any options to include/exclude slots
+		 * in dumping, so all the slots must be selected.
+		 */
+		slotinfo[i].dobj.dump = DUMP_COMPONENT_DEFINITION;
+	}
+	PQclear(res);
+
+	destroyPQExpBuffer(query);
+}
+
+/*
+ * dumpLogicalReplicationSlot
+ *	  dump creation functions for the given logical replication slots
+ */
+static void
+dumpLogicalReplicationSlot(Archive *fout,
+						   const LogicalReplicationSlotInfo *slotinfo)
+{
+	Assert(fout->dopt->logical_slots_only);
+
+	if (slotinfo->dobj.dump & DUMP_COMPONENT_DEFINITION)
+	{
+		PQExpBuffer query = createPQExpBuffer();
+
+		/*
+		 * XXX: For simplification, pg_create_logical_replication_slot() is
+		 * used. Is it sufficient?
+		 */
+		appendPQExpBuffer(query, "SELECT pg_catalog.pg_create_logical_replication_slot(");
+		appendStringLiteralAH(query, slotinfo->dobj.name, fout);
+		appendPQExpBuffer(query, ", ");
+		appendStringLiteralAH(query, slotinfo->plugin, fout);
+		appendPQExpBuffer(query, ", false, %s);",
+						  slotinfo->twophase ? "true" : "false");
+
+		ArchiveEntry(fout, slotinfo->dobj.catId, slotinfo->dobj.dumpId,
+					 ARCHIVE_OPTS(.tag = slotinfo->dobj.name,
+								  .description = "REPLICATION SLOT",
+								  .section = SECTION_POST_DATA,
+								  .createStmt = query->data));
+
+		destroyPQExpBuffer(query);
+	}
+}
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index ed6ce41ad7..de081c35ae 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -82,7 +82,8 @@ typedef enum
 	DO_PUBLICATION,
 	DO_PUBLICATION_REL,
 	DO_PUBLICATION_TABLE_IN_SCHEMA,
-	DO_SUBSCRIPTION
+	DO_SUBSCRIPTION,
+	DO_LOGICAL_REPLICATION_SLOT
 } DumpableObjectType;
 
 /*
@@ -666,6 +667,20 @@ typedef struct _SubscriptionInfo
 	char	   *subpasswordrequired;
 } SubscriptionInfo;
 
+/*
+ * The LogicalReplicationSlotInfo struct is used to represent replication
+ * slots.
+ *
+ * XXX: add more attributes if needed
+ */
+typedef struct _LogicalReplicationSlotInfo
+{
+	DumpableObject dobj;
+	char	   *plugin;
+	char	   *slottype;
+	bool		twophase;
+} LogicalReplicationSlotInfo;
+
 /*
  *	common utility functions
  */
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index 745578d855..4e12e46dc5 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -93,6 +93,7 @@ enum dbObjectTypePriorities
 	PRIO_PUBLICATION_REL,
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,
 	PRIO_SUBSCRIPTION,
+	PRIO_LOGICAL_REPLICATION_SLOT,
 	PRIO_DEFAULT_ACL,			/* done in ACL pass */
 	PRIO_EVENT_TRIGGER,			/* must be next to last! */
 	PRIO_REFRESH_MATVIEW		/* must be last! */
@@ -146,10 +147,11 @@ static const int dbObjectTypePriority[] =
 	PRIO_PUBLICATION,			/* DO_PUBLICATION */
 	PRIO_PUBLICATION_REL,		/* DO_PUBLICATION_REL */
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,	/* DO_PUBLICATION_TABLE_IN_SCHEMA */
-	PRIO_SUBSCRIPTION			/* DO_SUBSCRIPTION */
+	PRIO_SUBSCRIPTION,			/* DO_SUBSCRIPTION */
+	PRIO_LOGICAL_REPLICATION_SLOT	/* DO_LOGICAL_REPLICATION_SLOT */
 };
 
-StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_SUBSCRIPTION + 1),
+StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_LOGICAL_REPLICATION_SLOT + 1),
 				 "array length mismatch");
 
 static DumpId preDataBoundId;
@@ -1498,6 +1500,11 @@ describeDumpableObject(DumpableObject *obj, char *buf, int bufsize)
 					 "SUBSCRIPTION (ID %d OID %u)",
 					 obj->dumpId, obj->catId.oid);
 			return;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			snprintf(buf, bufsize,
+					 "LOGICAL REPLICATION SLOT (ID %d NAME %s)",
+					 obj->dumpId, obj->name);
+			return;
 		case DO_PRE_DATA_BOUNDARY:
 			snprintf(buf, bufsize,
 					 "PRE-DATA BOUNDARY  (ID %d)",
diff --git a/src/bin/pg_upgrade/Makefile b/src/bin/pg_upgrade/Makefile
index 5834513add..815d1a7ca1 100644
--- a/src/bin/pg_upgrade/Makefile
+++ b/src/bin/pg_upgrade/Makefile
@@ -3,6 +3,9 @@
 PGFILEDESC = "pg_upgrade - an in-place binary upgrade utility"
 PGAPPICON = win32
 
+# required for 003_logical_replication_slots.pl
+EXTRA_INSTALL=contrib/test_decoding
+
 subdir = src/bin/pg_upgrade
 top_builddir = ../../..
 include $(top_builddir)/src/Makefile.global
diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index fea159689e..f9cba8548e 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -30,6 +30,7 @@ static void check_for_jsonb_9_4_usage(ClusterInfo *cluster);
 static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
+static void check_for_parameter_settings(ClusterInfo *new_cluster);
 
 
 /*
@@ -89,6 +90,10 @@ check_and_dump_old_cluster(bool live_check)
 	/* Extract a list of databases and tables from the old cluster */
 	get_db_and_rel_infos(&old_cluster);
 
+	/* Additionally, extract a list of logical replication slots if required */
+	if (user_opts.include_logical_slots)
+		get_logical_slot_infos(&old_cluster);
+
 	init_tablespaces();
 
 	get_loadable_libraries();
@@ -189,6 +194,17 @@ check_new_cluster(void)
 {
 	get_db_and_rel_infos(&new_cluster);
 
+	/*
+	 * Do additional works if --include-logical-replication-slots is required.
+	 * These must be done before check_new_cluster_is_empty() because the
+	 * slot_arr attribute of the new_cluster will be checked in the function.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		get_logical_slot_infos(&new_cluster);
+		check_for_parameter_settings(&new_cluster);
+	}
+
 	check_new_cluster_is_empty();
 
 	check_loadable_libraries();
@@ -364,6 +380,22 @@ check_new_cluster_is_empty(void)
 						 rel_arr->rels[relnum].nspname,
 						 rel_arr->rels[relnum].relname);
 		}
+
+		/*
+		 * If --include-logical-replication-slots is required, check the
+		 * existence of slots
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			LogicalSlotInfoArr *slot_arr = &new_cluster.dbarr.dbs[dbnum].slot_arr;
+
+			/* if nslots > 0, report just first entry and exit */
+			if (slot_arr->nslots)
+				pg_fatal("New cluster database \"%s\" is not empty: found logical replication slot \"%s\"",
+						 new_cluster.dbarr.dbs[dbnum].db_name,
+						 slot_arr->slots[0].slotname);
+		}
+
 	}
 }
 
@@ -1402,3 +1434,38 @@ check_for_user_defined_encoding_conversions(ClusterInfo *cluster)
 	else
 		check_ok();
 }
+
+/*
+ * Verify parameter settings for creating logical replication slots
+ */
+static void
+check_for_parameter_settings(ClusterInfo *new_cluster)
+{
+	PGresult   *res;
+	PGconn	   *conn = connectToServer(new_cluster, "template1");
+	int			max_replication_slots;
+	char	   *wal_level;
+
+	prep_status("Checking for logical replication slots");
+
+	res = executeQueryOrDie(conn, "SHOW max_replication_slots;");
+	max_replication_slots = atoi(PQgetvalue(res, 0, 0));
+
+	if (max_replication_slots == 0)
+		pg_fatal("max_replication_slots must be greater than 0");
+
+	PQclear(res);
+
+	res = executeQueryOrDie(conn, "SHOW wal_level;");
+	wal_level = PQgetvalue(res, 0, 0);
+
+	if (strcmp(wal_level, "logical") != 0)
+		pg_fatal("wal_level must be \"logical\", but is set to \"%s\"",
+				 wal_level);
+
+	PQclear(res);
+
+	PQfinish(conn);
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/dump.c b/src/bin/pg_upgrade/dump.c
index 6c8c82dca8..e6b90864f5 100644
--- a/src/bin/pg_upgrade/dump.c
+++ b/src/bin/pg_upgrade/dump.c
@@ -59,6 +59,30 @@ generate_old_dump(void)
 						   log_opts.dumpdir,
 						   sql_file_name, escaped_connstr.data);
 
+		/*
+		 * Dump logical replication slots if needed.
+		 *
+		 * XXX We cannot dump replication slots at the same time as the schema
+		 * dump because we need to separate the timing of restoring
+		 * replication slots and other objects. Replication slots, in
+		 * particular, should not be restored before executing the pg_resetwal
+		 * command because it will remove WALs that are required by the slots.
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			char		slots_file_name[MAXPGPATH];
+
+			snprintf(slots_file_name, sizeof(slots_file_name),
+					 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+			parallel_exec_prog(log_file_name, NULL,
+							   "\"%s/pg_dump\" %s --logical-replication-slots-only "
+							   "--quote-all-identifiers --binary-upgrade %s "
+							   "--file=\"%s/%s\" %s",
+							   new_cluster.bindir, cluster_conn_opts(&old_cluster),
+							   log_opts.verbose ? "--verbose" : "",
+							   log_opts.dumpdir,
+							   slots_file_name, escaped_connstr.data);
+		}
 		termPQExpBuffer(&escaped_connstr);
 	}
 
diff --git a/src/bin/pg_upgrade/info.c b/src/bin/pg_upgrade/info.c
index 85ed15ae4a..139c4b3cde 100644
--- a/src/bin/pg_upgrade/info.c
+++ b/src/bin/pg_upgrade/info.c
@@ -23,10 +23,11 @@ static void free_db_and_rel_infos(DbInfoArr *db_arr);
 static void get_template0_info(ClusterInfo *cluster);
 static void get_db_infos(ClusterInfo *cluster);
 static void get_rel_infos(ClusterInfo *cluster, DbInfo *dbinfo);
+static void get_logical_slot_infos_per_db(ClusterInfo *cluster, DbInfo *dbinfo);
 static void free_rel_infos(RelInfoArr *rel_arr);
 static void print_db_infos(DbInfoArr *db_arr);
 static void print_rel_infos(RelInfoArr *rel_arr);
-
+static void print_slot_infos(LogicalSlotInfoArr *slot_arr);
 
 /*
  * gen_db_file_maps()
@@ -600,6 +601,94 @@ get_rel_infos(ClusterInfo *cluster, DbInfo *dbinfo)
 	dbinfo->rel_arr.nrels = num_rels;
 }
 
+/*
+ * get_logical_slot_infos()
+ *
+ * Higher level routine to generate LogicalSlotInfoArr for all databases.
+ */
+void
+get_logical_slot_infos(ClusterInfo *cluster)
+{
+	int			dbnum;
+
+	if (cluster == &old_cluster)
+		pg_log(PG_VERBOSE, "\nsource databases:");
+	else
+		pg_log(PG_VERBOSE, "\ntarget databases:");
+
+	for (dbnum = 0; dbnum < cluster->dbarr.ndbs; dbnum++)
+	{
+		DbInfo	   *pDbInfo = &cluster->dbarr.dbs[dbnum];
+
+		get_logical_slot_infos_per_db(cluster, pDbInfo);
+
+		if (log_opts.verbose)
+		{
+			pg_log(PG_VERBOSE, "Database: \"%s\"", pDbInfo->db_name);
+			print_slot_infos(&pDbInfo->slot_arr);
+		}
+	}
+}
+
+/*
+ * get_logical_slot_infos_per_db()
+ *
+ * gets the LogicalSlotInfos for all the logical replication slots of the database
+ * referred to by "dbinfo".
+ */
+static void
+get_logical_slot_infos_per_db(ClusterInfo *cluster, DbInfo *dbinfo)
+{
+	PGconn	   *conn = connectToServer(cluster,
+									   dbinfo->db_name);
+	PGresult   *res;
+	LogicalSlotInfo *slotinfos = NULL;
+
+	int			num_slots;
+
+	char		query[QUERY_ALLOC];
+
+	query[0] = '\0';			/* initialize query string to empty */
+
+	snprintf(query, sizeof(query),
+			 "SELECT slot_name, plugin, two_phase "
+			 "FROM pg_catalog.pg_replication_slots "
+			 "WHERE database = current_database() AND temporary = false "
+			 "AND wal_status IN ('reserved', 'extended');");
+
+	res = executeQueryOrDie(conn, "%s", query);
+
+	num_slots = PQntuples(res);
+
+	if (num_slots)
+	{
+		int			slotnum;
+		int			i_slotname;
+		int			i_plugin;
+		int			i_twophase;
+
+		slotinfos = (LogicalSlotInfo *) pg_malloc(sizeof(LogicalSlotInfo) * num_slots);
+
+		i_slotname = PQfnumber(res, "slot_name");
+		i_plugin = PQfnumber(res, "plugin");
+		i_twophase = PQfnumber(res, "two_phase");
+
+		for (slotnum = 0; slotnum < num_slots; slotnum++)
+		{
+			LogicalSlotInfo *curr = &slotinfos[slotnum];
+
+			curr->slotname = pg_strdup(PQgetvalue(res, slotnum, i_slotname));
+			curr->plugin = pg_strdup(PQgetvalue(res, slotnum, i_plugin));
+			curr->two_phase = (strcmp(PQgetvalue(res, slotnum, i_twophase), "t") == 0);
+		}
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	dbinfo->slot_arr.slots = slotinfos;
+	dbinfo->slot_arr.nslots = num_slots;
+}
 
 static void
 free_db_and_rel_infos(DbInfoArr *db_arr)
@@ -610,6 +699,14 @@ free_db_and_rel_infos(DbInfoArr *db_arr)
 	{
 		free_rel_infos(&db_arr->dbs[dbnum].rel_arr);
 		pg_free(db_arr->dbs[dbnum].db_name);
+
+		/*
+		 * db_arr has an additional attribute, LogicalSlotInfoArr slot_arr,
+		 * but there is no need to free it. It has a valid member only when
+		 * the cluster had logical replication slots in the previous call.
+		 * However, in this case, a FATAL error is thrown, and we cannot reach
+		 * this point.
+		 */
 	}
 	pg_free(db_arr->dbs);
 	db_arr->dbs = NULL;
@@ -660,3 +757,15 @@ print_rel_infos(RelInfoArr *rel_arr)
 			   rel_arr->rels[relnum].reloid,
 			   rel_arr->rels[relnum].tablespace);
 }
+
+static void
+print_slot_infos(LogicalSlotInfoArr *slot_arr)
+{
+	int			slotnum;
+
+	for (slotnum = 0; slotnum < slot_arr->nslots; slotnum++)
+		pg_log(PG_VERBOSE, "slotname: \"%s\", plugin: \"%s\", two_phase: %d",
+			   slot_arr->slots[slotnum].slotname,
+			   slot_arr->slots[slotnum].plugin,
+			   slot_arr->slots[slotnum].two_phase);
+}
diff --git a/src/bin/pg_upgrade/meson.build b/src/bin/pg_upgrade/meson.build
index 12a97f84e2..228f29b688 100644
--- a/src/bin/pg_upgrade/meson.build
+++ b/src/bin/pg_upgrade/meson.build
@@ -42,6 +42,7 @@ tests += {
     'tests': [
       't/001_basic.pl',
       't/002_pg_upgrade.pl',
+      't/003_logical_replication_slots.pl',
     ],
     'test_kwargs': {'priority': 40}, # pg_upgrade tests are slow
   },
diff --git a/src/bin/pg_upgrade/option.c b/src/bin/pg_upgrade/option.c
index 640361009e..df66a5ffe6 100644
--- a/src/bin/pg_upgrade/option.c
+++ b/src/bin/pg_upgrade/option.c
@@ -57,6 +57,7 @@ parseCommandLine(int argc, char *argv[])
 		{"verbose", no_argument, NULL, 'v'},
 		{"clone", no_argument, NULL, 1},
 		{"copy", no_argument, NULL, 2},
+		{"include-logical-replication-slots", no_argument, NULL, 3},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -199,6 +200,10 @@ parseCommandLine(int argc, char *argv[])
 				user_opts.transfer_mode = TRANSFER_MODE_COPY;
 				break;
 
+			case 3:
+				user_opts.include_logical_slots = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						os_info.progname);
@@ -289,6 +294,8 @@ usage(void)
 	printf(_("  -V, --version                 display version information, then exit\n"));
 	printf(_("  --clone                       clone instead of copying files to new cluster\n"));
 	printf(_("  --copy                        copy files to new cluster (default)\n"));
+	printf(_("  --include-logical-replication-slots\n"
+			 "                                upgrade logical replication slots\n"));
 	printf(_("  -?, --help                    show this help, then exit\n"));
 	printf(_("\n"
 			 "Before running pg_upgrade you must:\n"
diff --git a/src/bin/pg_upgrade/pg_upgrade.c b/src/bin/pg_upgrade/pg_upgrade.c
index 75bab0a04c..373a9ef490 100644
--- a/src/bin/pg_upgrade/pg_upgrade.c
+++ b/src/bin/pg_upgrade/pg_upgrade.c
@@ -59,6 +59,7 @@ static void copy_xact_xlog_xid(void);
 static void set_frozenxids(bool minmxid_only);
 static void make_outputdirs(char *pgdata);
 static void setup(char *argv0, bool *live_check);
+static void create_logical_replication_slots(void);
 
 ClusterInfo old_cluster,
 			new_cluster;
@@ -188,6 +189,19 @@ main(int argc, char **argv)
 			  new_cluster.pgdata);
 	check_ok();
 
+	/*
+	 * Create logical replication slots if requested.
+	 *
+	 * Note: This must be done after doing pg_resetwal command because the
+	 * command will remove required WALs.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		start_postmaster(&new_cluster, true);
+		create_logical_replication_slots();
+		stop_postmaster(false);
+	}
+
 	if (user_opts.do_sync)
 	{
 		prep_status("Sync data directory to disk");
@@ -860,3 +874,50 @@ set_frozenxids(bool minmxid_only)
 
 	check_ok();
 }
+
+/*
+ * create_logical_replication_slots()
+ *
+ * Similar to create_new_objects() but only restores logical replication slots.
+ */
+static void
+create_logical_replication_slots(void)
+{
+	int			dbnum;
+
+	prep_status_progress("Restoring logical replication slots in the new cluster");
+
+	for (dbnum = 0; dbnum < old_cluster.dbarr.ndbs; dbnum++)
+	{
+		char		slots_file_name[MAXPGPATH],
+					log_file_name[MAXPGPATH];
+		DbInfo	   *old_db = &old_cluster.dbarr.dbs[dbnum];
+
+		pg_log(PG_STATUS, "%s", old_db->db_name);
+
+		snprintf(slots_file_name, sizeof(slots_file_name),
+				 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+		snprintf(log_file_name, sizeof(log_file_name),
+				 DB_DUMP_LOG_FILE_MASK, old_db->db_oid);
+
+		parallel_exec_prog(log_file_name,
+						   NULL,
+						   "\"%s/psql\" %s --echo-queries --set ON_ERROR_STOP=on "
+						   "--no-psqlrc --dbname %s -f \"%s/%s\"",
+						   new_cluster.bindir,
+						   cluster_conn_opts(&new_cluster),
+						   old_db->db_name,
+						   log_opts.dumpdir,
+						   slots_file_name);
+	}
+
+	/* reap all children */
+	while (reap_child(true) == true)
+		;
+
+	end_progress_output();
+	check_ok();
+
+	/* update new_cluster info again */
+	get_logical_slot_infos(&new_cluster);
+}
diff --git a/src/bin/pg_upgrade/pg_upgrade.h b/src/bin/pg_upgrade/pg_upgrade.h
index 3eea0139c7..7adbb50807 100644
--- a/src/bin/pg_upgrade/pg_upgrade.h
+++ b/src/bin/pg_upgrade/pg_upgrade.h
@@ -29,6 +29,7 @@
 /* contains both global db information and CREATE DATABASE commands */
 #define GLOBALS_DUMP_FILE	"pg_upgrade_dump_globals.sql"
 #define DB_DUMP_FILE_MASK	"pg_upgrade_dump_%u.custom"
+#define DB_DUMP_LOGICAL_SLOTS_FILE_MASK	"pg_upgrade_dump_%u_logical_slots.sql"
 
 /*
  * Base directories that include all the files generated internally, from the
@@ -150,6 +151,22 @@ typedef struct
 	int			nrels;
 } RelInfoArr;
 
+/*
+ * Structure to store logical replication slot information
+ */
+typedef struct
+{
+	char	   *slotname;		/* slot name */
+	char	   *plugin;			/* plugin */
+	bool		two_phase;		/* Can the slot decode 2PC? */
+} LogicalSlotInfo;
+
+typedef struct
+{
+	LogicalSlotInfo *slots;
+	int			nslots;
+} LogicalSlotInfoArr;
+
 /*
  * The following structure represents a relation mapping.
  */
@@ -176,6 +193,7 @@ typedef struct
 	char		db_tablespace[MAXPGPATH];	/* database default tablespace
 											 * path */
 	RelInfoArr	rel_arr;		/* array of all user relinfos */
+	LogicalSlotInfoArr slot_arr;	/* array of all LogicalSlotInfo */
 } DbInfo;
 
 /*
@@ -304,6 +322,8 @@ typedef struct
 	transferMode transfer_mode; /* copy files or link them? */
 	int			jobs;			/* number of processes/threads to use */
 	char	   *socketdir;		/* directory to use for Unix sockets */
+	bool		include_logical_slots;	/* true -> dump and restore logical
+										 * replication slots */
 } UserOpts;
 
 typedef struct
@@ -400,6 +420,7 @@ FileNameMap *gen_db_file_maps(DbInfo *old_db,
 							  DbInfo *new_db, int *nmaps, const char *old_pgdata,
 							  const char *new_pgdata);
 void		get_db_and_rel_infos(ClusterInfo *cluster);
+void		get_logical_slot_infos(ClusterInfo *cluster);
 
 /* option.c */
 
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
new file mode 100644
index 0000000000..525a7704cf
--- /dev/null
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -0,0 +1,115 @@
+# Copyright (c) 2023, PostgreSQL Global Development Group
+
+# Tests for upgrading replication slots
+
+use strict;
+use warnings;
+
+use File::Path qw(rmtree);
+
+use PostgreSQL::Test::Cluster;
+use PostgreSQL::Test::Utils;
+use Test::More;
+
+# Can be changed to test the other modes
+my $mode = $ENV{PG_TEST_PG_UPGRADE_MODE} || '--copy';
+
+# Initialize old node
+my $old_node = PostgreSQL::Test::Cluster->new('old_node');
+$old_node->init(allows_streaming => 'logical');
+$old_node->start;
+
+# Initialize new node
+my $new_node = PostgreSQL::Test::Cluster->new('new_node');
+$new_node->init(allows_streaming => 1);
+
+my $bindir = $new_node->config_data('--bindir');
+
+$old_node->stop;
+
+# Cause a failure at the start of pg_upgrade because wal_level is replica
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with wrong wal_level');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+# Preparations for the subsequent test. The case max_replication_slots is set
+# to 0 is prohibited.
+$new_node->append_conf('postgresql.conf', "wal_level = 'logical'");
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 0");
+
+# Cause a failure at the start of pg_upgrade because max_replication_slots is 0
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with wrong max_replication_slots');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+# Preparations for the subsequent test. max_replication_slots is set to
+# non-zero value to succeed the pg_upgrade
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 10");
+
+# Create a slot on old node, and generate WALs
+$old_node->start;
+$old_node->safe_psql(
+	'postgres', qq[
+	SELECT pg_create_logical_replication_slot('test_slot', 'test_decoding', false, true);
+	CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a;
+]);
+
+my $result = $old_node->safe_psql('postgres',
+	"SELECT count (*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)"
+);
+is($result, qq(12), 'ensure WALs are not consumed yet');
+$old_node->stop;
+
+# Actual run, pg_upgrade_output.d is removed at the end
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots'
+	],
+	'run of pg_upgrade of old node');
+ok( !-d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+$new_node->start;
+$result = $new_node->safe_psql('postgres',
+	"SELECT slot_name, two_phase FROM pg_replication_slots");
+is($result, qq(test_slot|t), 'check the slot exists on new node');
+
+done_testing();
diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list
index b4058b88c3..5944cb34ea 100644
--- a/src/tools/pgindent/typedefs.list
+++ b/src/tools/pgindent/typedefs.list
@@ -1479,6 +1479,7 @@ LogicalRepBeginData
 LogicalRepCommitData
 LogicalRepCommitPreparedTxnData
 LogicalRepCtxStruct
+LogicalReplicationSlotInfo
 LogicalRepMode
 LogicalRepMsgType
 LogicalRepPartMapEntry
@@ -1492,6 +1493,8 @@ LogicalRepTupleData
 LogicalRepTyp
 LogicalRepWorker
 LogicalRewriteMappingData
+LogicalSlotInfo
+LogicalSlotInfoArr
 LogicalTape
 LogicalTapeSet
 LsnReadQueue
-- 
2.27.0

From add3123cdab69dae07b2bd81645ee98c75acabd2 Mon Sep 17 00:00:00 2001
From: Julien Rouhaud <julien.rouhaud@free.fr>
Date: Fri, 14 Apr 2023 13:49:09 +0800
Subject: [PATCH v14 2/4] Always persist to disk logical slots during a
 shutdown checkpoint.

It's entirely possible for a logical slot to have a confirmed_flush_lsn higher
than the last value saved on disk while not being marked as dirty.  It's
currently not a problem to lose that value during a clean shutdown / restart
cycle, but a later patch adding support for pg_upgrade of publications and
logical slots will rely on that value being properly persisted to disk.

Author: Julien Rouhaud
Reviewed-by: FIXME
Discussion: FIXME
---
 src/backend/access/transam/xlog.c |  2 +-
 src/backend/replication/slot.c    | 26 ++++++++++++++++----------
 src/include/replication/slot.h    |  2 +-
 3 files changed, 18 insertions(+), 12 deletions(-)

diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index bc5a8e0569..78b4528f2c 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7011,7 +7011,7 @@ static void
 CheckPointGuts(XLogRecPtr checkPointRedo, int flags)
 {
 	CheckPointRelationMap();
-	CheckPointReplicationSlots();
+	CheckPointReplicationSlots(flags & CHECKPOINT_IS_SHUTDOWN);
 	CheckPointSnapBuild();
 	CheckPointLogicalRewriteHeap();
 	CheckPointReplicationOrigin();
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index 8021aaa0a8..aeea6ffd1f 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -109,7 +109,8 @@ static void ReplicationSlotDropPtr(ReplicationSlot *slot);
 /* internal persistency functions */
 static void RestoreSlotFromDisk(const char *name);
 static void CreateSlotOnDisk(ReplicationSlot *slot);
-static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel);
+static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+						   bool is_shutdown);
 
 /*
  * Report shared-memory space needed by ReplicationSlotsShmemInit.
@@ -783,7 +784,7 @@ ReplicationSlotSave(void)
 	Assert(MyReplicationSlot != NULL);
 
 	sprintf(path, "pg_replslot/%s", NameStr(MyReplicationSlot->data.name));
-	SaveSlotToPath(MyReplicationSlot, path, ERROR);
+	SaveSlotToPath(MyReplicationSlot, path, ERROR, false);
 }
 
 /*
@@ -1565,11 +1566,10 @@ restart:
 /*
  * Flush all replication slots to disk.
  *
- * This needn't actually be part of a checkpoint, but it's a convenient
- * location.
+ * is_shutdown is true in case of a shutdown checkpoint.
  */
 void
-CheckPointReplicationSlots(void)
+CheckPointReplicationSlots(bool is_shutdown)
 {
 	int			i;
 
@@ -1594,7 +1594,7 @@ CheckPointReplicationSlots(void)
 
 		/* save the slot to disk, locking is handled in SaveSlotToPath() */
 		sprintf(path, "pg_replslot/%s", NameStr(s->data.name));
-		SaveSlotToPath(s, path, LOG);
+		SaveSlotToPath(s, path, LOG, is_shutdown);
 	}
 	LWLockRelease(ReplicationSlotAllocationLock);
 }
@@ -1700,7 +1700,7 @@ CreateSlotOnDisk(ReplicationSlot *slot)
 
 	/* Write the actual state file. */
 	slot->dirty = true;			/* signal that we really need to write */
-	SaveSlotToPath(slot, tmppath, ERROR);
+	SaveSlotToPath(slot, tmppath, ERROR, false);
 
 	/* Rename the directory into place. */
 	if (rename(tmppath, path) != 0)
@@ -1726,7 +1726,8 @@ CreateSlotOnDisk(ReplicationSlot *slot)
  * Shared functionality between saving and creating a replication slot.
  */
 static void
-SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
+SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+			   bool is_shutdown)
 {
 	char		tmppath[MAXPGPATH];
 	char		path[MAXPGPATH];
@@ -1740,8 +1741,13 @@ SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
 	slot->just_dirtied = false;
 	SpinLockRelease(&slot->mutex);
 
-	/* and don't do anything if there's nothing to write */
-	if (!was_dirty)
+	/*
+	 * and don't do anything if there's nothing to write, unless it's this is
+	 * called for a logical slot during a shutdown checkpoint, as we want to
+	 * persist the confirmed_flush_lsn in that case, even if that's the only
+	 * modification.
+	 */
+	if (!was_dirty && !is_shutdown && !SlotIsLogical(slot))
 		return;
 
 	LWLockAcquire(&slot->io_in_progress_lock, LW_EXCLUSIVE);
diff --git a/src/include/replication/slot.h b/src/include/replication/slot.h
index a8a89dc784..7ca37c9f70 100644
--- a/src/include/replication/slot.h
+++ b/src/include/replication/slot.h
@@ -241,7 +241,7 @@ extern void ReplicationSlotNameForTablesync(Oid suboid, Oid relid, char *syncslo
 extern void ReplicationSlotDropAtPubNode(WalReceiverConn *wrconn, char *slotname, bool missing_ok);
 
 extern void StartupReplicationSlots(void);
-extern void CheckPointReplicationSlots(void);
+extern void CheckPointReplicationSlots(bool is_shutdown);
 
 extern void CheckSlotRequirements(void);
 extern void CheckSlotPermissions(void);
-- 
2.27.0

From cd3e33315d823b1202fda003622dbaac1b8ad9fa Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Fri, 14 Apr 2023 08:59:03 +0000
Subject: [PATCH v14 3/4] pg_upgrade: Add check function for
 --include-logical-replication-slots option

---
 src/bin/pg_upgrade/check.c                    | 80 ++++++++++++++++++-
 .../t/003_logical_replication_slots.pl        | 30 ++++++-
 2 files changed, 108 insertions(+), 2 deletions(-)

diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index f9cba8548e..61ca36a853 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -9,7 +9,10 @@
 
 #include "postgres_fe.h"
 
+#include "access/xlogrecord.h"
+#include "access/xlog_internal.h"
 #include "catalog/pg_authid_d.h"
+#include "catalog/pg_control.h"
 #include "catalog/pg_collation.h"
 #include "fe_utils/string_utils.h"
 #include "mb/pg_wchar.h"
@@ -31,7 +34,7 @@ static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
 static void check_for_parameter_settings(ClusterInfo *new_cluster);
-
+static void check_for_confirmed_flush_lsn(ClusterInfo *cluster);
 
 /*
  * fix_path_separator
@@ -108,6 +111,8 @@ check_and_dump_old_cluster(bool live_check)
 	check_for_composite_data_type_usage(&old_cluster);
 	check_for_reg_data_type_usage(&old_cluster);
 	check_for_isn_and_int8_passing_mismatch(&old_cluster);
+	if (user_opts.include_logical_slots)
+		check_for_confirmed_flush_lsn(&old_cluster);
 
 	/*
 	 * PG 16 increased the size of the 'aclitem' type, which breaks the on-disk
@@ -1446,6 +1451,10 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 	int			max_replication_slots;
 	char	   *wal_level;
 
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(new_cluster->major_version < 1600))
+		return;
+
 	prep_status("Checking for logical replication slots");
 
 	res = executeQueryOrDie(conn, "SHOW max_replication_slots;");
@@ -1469,3 +1478,72 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 
 	check_ok();
 }
+
+/*
+ * Verify that all logical replication slots consumed all WALs, except a
+ * CHECKPOINT_SHUTDOWN record.
+ */
+static void
+check_for_confirmed_flush_lsn(ClusterInfo *cluster)
+{
+	int			i,
+				ntups,
+				i_slotname;
+	bool		is_error = false;
+	PGresult   *res;
+	DbInfo	   *active_db = &cluster->dbarr.dbs[0];
+	PGconn	   *conn = connectToServer(cluster, active_db->db_name);
+
+	Assert(user_opts.include_logical_slots);
+
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(cluster->major_version < 1600))
+		return;
+
+	prep_status("Checking for logical replication slots");
+
+	/*
+	 * Check that all logical replication slots have reached the current WAL
+	 * position, except for the CHECKPOINT_SHUTDOWN record. Even if all WALs
+	 * are consumed before shutting down the node, the checkpointer generates
+	 * a CHECKPOINT_SHUTDOWN record at shutdown, which cannot be consumed by
+	 * any slots. Therefore, we must allow for a difference between
+	 * pg_current_wal_insert_lsn() and confirmed_flush_lsn.
+	 */
+#define SHUTDOWN_RECORD_SIZE  (SizeOfXLogRecord + \
+							   SizeOfXLogRecordDataHeaderShort + \
+							   sizeof(CheckPoint))
+
+	res = executeQueryOrDie(conn,
+							"SELECT slot_name FROM pg_catalog.pg_replication_slots "
+							"WHERE (pg_catalog.pg_current_wal_insert_lsn() - confirmed_flush_lsn) > %d "
+							"AND temporary = false AND wal_status IN ('reserved', 'extended');",
+							(int) (SizeOfXLogLongPHD + SHUTDOWN_RECORD_SIZE));
+
+#undef SHUTDOWN_RECORD_SIZE
+
+	ntups = PQntuples(res);
+	i_slotname = PQfnumber(res, "slot_name");
+
+	for (i = 0; i < ntups; i++)
+	{
+		char	   *slotname;
+
+		is_error = true;
+
+		slotname = PQgetvalue(res, i, i_slotname);
+
+		pg_log(PG_WARNING,
+			   "\nWARNING: logical replication slot \"%s\" has not consumed WALs yet",
+			   slotname);
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	if (is_error)
+		pg_fatal("--include-logical-replication-slots requires that all "
+				 "logical replication slots consumed all the WALs");
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
index 525a7704cf..21fefca084 100644
--- a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -85,11 +85,39 @@ $old_node->safe_psql(
 ]);
 
 my $result = $old_node->safe_psql('postgres',
-	"SELECT count (*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)"
+	"SELECT count(*) FROM pg_logical_slot_peek_changes('test_slot', NULL, NULL)"
 );
+
 is($result, qq(12), 'ensure WALs are not consumed yet');
 $old_node->stop;
 
+# Cause a failure at the start of pg_upgrade because test_slot does not
+# finish consuming all the WALs
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with idle replication slots');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+$old_node->start;
+$old_node->safe_psql('postgres',
+	"SELECT count (*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)"
+);
+$old_node->stop;
+
 # Actual run, pg_upgrade_output.d is removed at the end
 command_ok(
 	[
-- 
2.27.0

From 6979c1bb7454a495fdea6ae99a99df6500e88c81 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Tue, 4 Apr 2023 05:49:34 +0000
Subject: [PATCH v15 1/4] pg_upgrade: Add --include-logical-replication-slots
 option

This commit introduces a new pg_upgrade option called "--include-logical-replication-slots".
This allows nodes with logical replication slots to be upgraded. The commit can
be divided into two parts: one for pg_dump and another for pg_upgrade.

For pg_dump this commit includes a new option called "--logical-replication-slots-only".
This option can be used to dump logical replication slots. When this option is
specified, the slot_name, plugin, and two_phase parameters are extracted from
pg_replication_slots. An SQL file is then generated which executes
pg_create_logical_replication_slot() with the extracted parameters.

For pg_upgrade, when '--include-logical-replication-slots' is specified, it executes
pg_dump with the new "--logical-replication-slots-only" option and restores from the
dump. Note that we cannot dump replication slots at the same time as the schema
dump because we need to separate the timing of restoring replication slots and
other objects. Replication slots, in  particular, should not be restored before
executing the pg_resetwal command because it will remove WALs that are required
by the slots.

The significant advantage of this commit is that it makes it easy to continue
logical replication even after upgrading the publisher node. Previously, pg_upgrade
allowed copying publications to a new node. With this new commit, adjusting the
connection string to the new publisher will cause the apply worker on the subscriber
to connect to the new publisher automatically. This enables seamless continuation
of logical replication, even after an upgrade.

Author: Hayato Kuroda
Reviewed-by: Peter Smith, Julien Rouhaud, Vignesh C, Wang Wei
---
 doc/src/sgml/ref/pgupgrade.sgml               |  11 ++
 src/bin/pg_dump/pg_backup.h                   |   1 +
 src/bin/pg_dump/pg_dump.c                     | 156 ++++++++++++++++++
 src/bin/pg_dump/pg_dump.h                     |  16 +-
 src/bin/pg_dump/pg_dump_sort.c                |  11 +-
 src/bin/pg_upgrade/Makefile                   |   3 +
 src/bin/pg_upgrade/check.c                    |  67 ++++++++
 src/bin/pg_upgrade/dump.c                     |  24 +++
 src/bin/pg_upgrade/info.c                     | 111 ++++++++++++-
 src/bin/pg_upgrade/meson.build                |   1 +
 src/bin/pg_upgrade/option.c                   |   7 +
 src/bin/pg_upgrade/pg_upgrade.c               |  61 +++++++
 src/bin/pg_upgrade/pg_upgrade.h               |  21 +++
 .../t/003_logical_replication_slots.pl        | 115 +++++++++++++
 src/tools/pgindent/typedefs.list              |   3 +
 15 files changed, 604 insertions(+), 4 deletions(-)
 create mode 100644 src/bin/pg_upgrade/t/003_logical_replication_slots.pl

diff --git a/doc/src/sgml/ref/pgupgrade.sgml b/doc/src/sgml/ref/pgupgrade.sgml
index 7816b4c685..94e90ff506 100644
--- a/doc/src/sgml/ref/pgupgrade.sgml
+++ b/doc/src/sgml/ref/pgupgrade.sgml
@@ -240,6 +240,17 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--include-logical-replication-slots</option></term>
+      <listitem>
+       <para>
+        Upgrade logical replication slots. Only permanent replication slots
+        are included. Note that pg_upgrade does not check the installation of
+        plugins.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
       <term><option>-?</option></term>
       <term><option>--help</option></term>
diff --git a/src/bin/pg_dump/pg_backup.h b/src/bin/pg_dump/pg_backup.h
index aba780ef4b..0a4e931f9b 100644
--- a/src/bin/pg_dump/pg_backup.h
+++ b/src/bin/pg_dump/pg_backup.h
@@ -187,6 +187,7 @@ typedef struct _dumpOptions
 	int			use_setsessauth;
 	int			enable_row_security;
 	int			load_via_partition_root;
+	int			logical_slots_only;
 
 	/* default, if no "inclusion" switches appear, is to dump everything */
 	bool		include_everything;
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index 3af97a6039..3c6d2462d4 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -328,6 +328,9 @@ static void setupDumpWorker(Archive *AH);
 static TableInfo *getRootTableInfo(const TableInfo *tbinfo);
 static bool forcePartitionRootLoad(const TableInfo *tbinfo);
 
+static void getLogicalReplicationSlots(Archive *fout);
+static void dumpLogicalReplicationSlot(Archive *fout,
+									   const LogicalReplicationSlotInfo *slotinfo);
 
 int
 main(int argc, char **argv)
@@ -431,6 +434,7 @@ main(int argc, char **argv)
 		{"table-and-children", required_argument, NULL, 12},
 		{"exclude-table-and-children", required_argument, NULL, 13},
 		{"exclude-table-data-and-children", required_argument, NULL, 14},
+		{"logical-replication-slots-only", no_argument, NULL, 15},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -657,6 +661,10 @@ main(int argc, char **argv)
 										  optarg);
 				break;
 
+			case 15:			/* dump only replication slot(s) */
+				dopt.logical_slots_only = true;
+				break;
+
 			default:
 				/* getopt_long already emitted a complaint */
 				pg_log_error_hint("Try \"%s --help\" for more information.", progname);
@@ -714,6 +722,21 @@ main(int argc, char **argv)
 	if (dopt.do_nothing && dopt.dump_inserts == 0)
 		pg_fatal("option --on-conflict-do-nothing requires option --inserts, --rows-per-insert, or --column-inserts");
 
+	if (dopt.logical_slots_only)
+	{
+		if (!dopt.binary_upgrade)
+			pg_fatal("options --logical-replication-slots-only requires option --binary-upgrade");
+
+		if (dopt.dataOnly)
+			pg_fatal("options --logical-replication-slots-only and -a/--data-only cannot be used together");
+
+		if (dopt.schemaOnly)
+			pg_fatal("options --logical-replication-slots-only and -s/--schema-only cannot be used together");
+
+		if (dopt.outputClean)
+			pg_fatal("options --logical-replication-slots-only and -c/--clean cannot be used together");
+	}
+
 	/* Identify archive format to emit */
 	archiveFormat = parseArchiveFormat(format, &archiveMode);
 
@@ -876,6 +899,16 @@ main(int argc, char **argv)
 			pg_fatal("no matching extensions were found");
 	}
 
+	/*
+	 * If dump logical-replication-slots-only was requested, dump only them
+	 * and skip everything else.
+	 */
+	if (dopt.logical_slots_only)
+	{
+		getLogicalReplicationSlots(fout);
+		goto dump;
+	}
+
 	/*
 	 * Dumping LOs is the default for dumps where an inclusion switch is not
 	 * used (an "include everything" dump).  -B can be used to exclude LOs
@@ -936,6 +969,8 @@ main(int argc, char **argv)
 	if (!dopt.no_security_labels)
 		collectSecLabels(fout);
 
+dump:
+
 	/* Lastly, create dummy objects to represent the section boundaries */
 	boundaryObjs = createBoundaryObjects();
 
@@ -1109,6 +1144,11 @@ help(const char *progname)
 			 "                               servers matching PATTERN\n"));
 	printf(_("  --inserts                    dump data as INSERT commands, rather than COPY\n"));
 	printf(_("  --load-via-partition-root    load partitions via the root table\n"));
+
+	/*
+	 * The option --logical-replication-slots-only is used only by pg_upgrade
+	 * and should not be called by users, which is why it is not listed.
+	 */
 	printf(_("  --no-comments                do not dump comments\n"));
 	printf(_("  --no-publications            do not dump publications\n"));
 	printf(_("  --no-security-labels         do not dump security label assignments\n"));
@@ -10235,6 +10275,10 @@ dumpDumpableObject(Archive *fout, DumpableObject *dobj)
 		case DO_SUBSCRIPTION:
 			dumpSubscription(fout, (const SubscriptionInfo *) dobj);
 			break;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			dumpLogicalReplicationSlot(fout,
+									   (const LogicalReplicationSlotInfo *) dobj);
+			break;
 		case DO_PRE_DATA_BOUNDARY:
 		case DO_POST_DATA_BOUNDARY:
 			/* never dumped, nothing to do */
@@ -18215,6 +18259,7 @@ addBoundaryDependencies(DumpableObject **dobjs, int numObjs,
 			case DO_PUBLICATION_REL:
 			case DO_PUBLICATION_TABLE_IN_SCHEMA:
 			case DO_SUBSCRIPTION:
+			case DO_LOGICAL_REPLICATION_SLOT:
 				/* Post-data objects: must come after the post-data boundary */
 				addObjectDependency(dobj, postDataBound->dumpId);
 				break;
@@ -18476,3 +18521,114 @@ appendReloptionsArrayAH(PQExpBuffer buffer, const char *reloptions,
 	if (!res)
 		pg_log_warning("could not parse %s array", "reloptions");
 }
+
+/*
+ * getLogicalReplicationSlots
+ *	  get information about replication slots
+ */
+static void
+getLogicalReplicationSlots(Archive *fout)
+{
+	PGresult   *res;
+	LogicalReplicationSlotInfo *slotinfo;
+	PQExpBuffer query;
+
+	int			i_slotname;
+	int			i_plugin;
+	int			i_twophase;
+	int			i,
+				ntups;
+
+	/* Check whether we should dump or not */
+	if (fout->remoteVersion < 160000)
+		return;
+
+	Assert(fout->dopt->logical_slots_only);
+
+	query = createPQExpBuffer();
+
+	resetPQExpBuffer(query);
+
+	/*
+	 * Get replication slots.
+	 *
+	 * XXX: Which information must be extracted from old node? Currently three
+	 * attributes are extracted because they are used by
+	 * pg_create_logical_replication_slot().
+	 *
+	 * XXX: Do we have to support physical slots?
+	 */
+	appendPQExpBufferStr(query,
+						 "SELECT slot_name, plugin, two_phase "
+						 "FROM pg_catalog.pg_replication_slots "
+						 "WHERE database = current_database() AND temporary = false "
+						 "AND wal_status IN ('reserved', 'extended');");
+
+	res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
+
+	ntups = PQntuples(res);
+
+	i_slotname = PQfnumber(res, "slot_name");
+	i_plugin = PQfnumber(res, "plugin");
+	i_twophase = PQfnumber(res, "two_phase");
+
+	slotinfo = pg_malloc(ntups * sizeof(LogicalReplicationSlotInfo));
+
+	for (i = 0; i < ntups; i++)
+	{
+		slotinfo[i].dobj.objType = DO_LOGICAL_REPLICATION_SLOT;
+
+		slotinfo[i].dobj.catId.tableoid = InvalidOid;
+		slotinfo[i].dobj.catId.oid = InvalidOid;
+		AssignDumpId(&slotinfo[i].dobj);
+
+		slotinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_slotname));
+
+		slotinfo[i].plugin = pg_strdup(PQgetvalue(res, i, i_plugin));
+		slotinfo[i].twophase = (strcmp(PQgetvalue(res, i, i_twophase), "t") == 0);
+
+		/*
+		 * Note: Currently we do not have any options to include/exclude slots
+		 * in dumping, so all the slots must be selected.
+		 */
+		slotinfo[i].dobj.dump = DUMP_COMPONENT_DEFINITION;
+	}
+	PQclear(res);
+
+	destroyPQExpBuffer(query);
+}
+
+/*
+ * dumpLogicalReplicationSlot
+ *	  dump creation functions for the given logical replication slots
+ */
+static void
+dumpLogicalReplicationSlot(Archive *fout,
+						   const LogicalReplicationSlotInfo *slotinfo)
+{
+	Assert(fout->dopt->logical_slots_only);
+
+	if (slotinfo->dobj.dump & DUMP_COMPONENT_DEFINITION)
+	{
+		PQExpBuffer query = createPQExpBuffer();
+
+		/*
+		 * XXX: For simplification, pg_create_logical_replication_slot() is
+		 * used. Is it sufficient?
+		 */
+		appendPQExpBuffer(query, "SELECT pg_catalog.pg_create_logical_replication_slot(");
+		appendStringLiteralAH(query, slotinfo->dobj.name, fout);
+		appendPQExpBuffer(query, ", ");
+		appendStringLiteralAH(query, slotinfo->plugin, fout);
+		appendPQExpBuffer(query, ", false, %s);",
+						  slotinfo->twophase ? "true" : "false");
+
+		ArchiveEntry(fout, slotinfo->dobj.catId, slotinfo->dobj.dumpId,
+					 ARCHIVE_OPTS(.tag = slotinfo->dobj.name,
+								  .description = "REPLICATION SLOT",
+								  .section = SECTION_POST_DATA,
+								  .createStmt = query->data));
+
+		destroyPQExpBuffer(query);
+	}
+}
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index ed6ce41ad7..0582cd84b1 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -82,7 +82,8 @@ typedef enum
 	DO_PUBLICATION,
 	DO_PUBLICATION_REL,
 	DO_PUBLICATION_TABLE_IN_SCHEMA,
-	DO_SUBSCRIPTION
+	DO_SUBSCRIPTION,
+	DO_LOGICAL_REPLICATION_SLOT
 } DumpableObjectType;
 
 /*
@@ -666,6 +667,19 @@ typedef struct _SubscriptionInfo
 	char	   *subpasswordrequired;
 } SubscriptionInfo;
 
+/*
+ * The LogicalReplicationSlotInfo struct is used to represent replication
+ * slots.
+ *
+ * XXX: add more attributes if needed
+ */
+typedef struct _LogicalReplicationSlotInfo
+{
+	DumpableObject dobj;
+	char	   *plugin;
+	bool		twophase;
+} LogicalReplicationSlotInfo;
+
 /*
  *	common utility functions
  */
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index 745578d855..4e12e46dc5 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -93,6 +93,7 @@ enum dbObjectTypePriorities
 	PRIO_PUBLICATION_REL,
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,
 	PRIO_SUBSCRIPTION,
+	PRIO_LOGICAL_REPLICATION_SLOT,
 	PRIO_DEFAULT_ACL,			/* done in ACL pass */
 	PRIO_EVENT_TRIGGER,			/* must be next to last! */
 	PRIO_REFRESH_MATVIEW		/* must be last! */
@@ -146,10 +147,11 @@ static const int dbObjectTypePriority[] =
 	PRIO_PUBLICATION,			/* DO_PUBLICATION */
 	PRIO_PUBLICATION_REL,		/* DO_PUBLICATION_REL */
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,	/* DO_PUBLICATION_TABLE_IN_SCHEMA */
-	PRIO_SUBSCRIPTION			/* DO_SUBSCRIPTION */
+	PRIO_SUBSCRIPTION,			/* DO_SUBSCRIPTION */
+	PRIO_LOGICAL_REPLICATION_SLOT	/* DO_LOGICAL_REPLICATION_SLOT */
 };
 
-StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_SUBSCRIPTION + 1),
+StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_LOGICAL_REPLICATION_SLOT + 1),
 				 "array length mismatch");
 
 static DumpId preDataBoundId;
@@ -1498,6 +1500,11 @@ describeDumpableObject(DumpableObject *obj, char *buf, int bufsize)
 					 "SUBSCRIPTION (ID %d OID %u)",
 					 obj->dumpId, obj->catId.oid);
 			return;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			snprintf(buf, bufsize,
+					 "LOGICAL REPLICATION SLOT (ID %d NAME %s)",
+					 obj->dumpId, obj->name);
+			return;
 		case DO_PRE_DATA_BOUNDARY:
 			snprintf(buf, bufsize,
 					 "PRE-DATA BOUNDARY  (ID %d)",
diff --git a/src/bin/pg_upgrade/Makefile b/src/bin/pg_upgrade/Makefile
index 5834513add..815d1a7ca1 100644
--- a/src/bin/pg_upgrade/Makefile
+++ b/src/bin/pg_upgrade/Makefile
@@ -3,6 +3,9 @@
 PGFILEDESC = "pg_upgrade - an in-place binary upgrade utility"
 PGAPPICON = win32
 
+# required for 003_logical_replication_slots.pl
+EXTRA_INSTALL=contrib/test_decoding
+
 subdir = src/bin/pg_upgrade
 top_builddir = ../../..
 include $(top_builddir)/src/Makefile.global
diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index 64024e3b9e..f08fcc7705 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -30,6 +30,7 @@ static void check_for_jsonb_9_4_usage(ClusterInfo *cluster);
 static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
+static void check_for_parameter_settings(ClusterInfo *new_cluster);
 
 
 /*
@@ -89,6 +90,10 @@ check_and_dump_old_cluster(bool live_check)
 	/* Extract a list of databases and tables from the old cluster */
 	get_db_and_rel_infos(&old_cluster);
 
+	/* Additionally, extract a list of logical replication slots if required */
+	if (user_opts.include_logical_slots)
+		get_logical_slot_infos(&old_cluster);
+
 	init_tablespaces();
 
 	get_loadable_libraries();
@@ -189,6 +194,17 @@ check_new_cluster(void)
 {
 	get_db_and_rel_infos(&new_cluster);
 
+	/*
+	 * Do additional works if --include-logical-replication-slots is required.
+	 * These must be done before check_new_cluster_is_empty() because the
+	 * slot_arr attribute of the new_cluster will be checked in the function.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		get_logical_slot_infos(&new_cluster);
+		check_for_parameter_settings(&new_cluster);
+	}
+
 	check_new_cluster_is_empty();
 
 	check_loadable_libraries();
@@ -364,6 +380,22 @@ check_new_cluster_is_empty(void)
 						 rel_arr->rels[relnum].nspname,
 						 rel_arr->rels[relnum].relname);
 		}
+
+		/*
+		 * If --include-logical-replication-slots is required, check the
+		 * existence of slots
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			LogicalSlotInfoArr *slot_arr = &new_cluster.dbarr.dbs[dbnum].slot_arr;
+
+			/* if nslots > 0, report just first entry and exit */
+			if (slot_arr->nslots)
+				pg_fatal("New cluster database \"%s\" is not empty: found logical replication slot \"%s\"",
+						 new_cluster.dbarr.dbs[dbnum].db_name,
+						 slot_arr->slots[0].slotname);
+		}
+
 	}
 }
 
@@ -1402,3 +1434,38 @@ check_for_user_defined_encoding_conversions(ClusterInfo *cluster)
 	else
 		check_ok();
 }
+
+/*
+ * Verify parameter settings for creating logical replication slots
+ */
+static void
+check_for_parameter_settings(ClusterInfo *new_cluster)
+{
+	PGresult   *res;
+	PGconn	   *conn = connectToServer(new_cluster, "template1");
+	int			max_replication_slots;
+	char	   *wal_level;
+
+	prep_status("Checking for logical replication slots");
+
+	res = executeQueryOrDie(conn, "SHOW max_replication_slots;");
+	max_replication_slots = atoi(PQgetvalue(res, 0, 0));
+
+	if (max_replication_slots == 0)
+		pg_fatal("max_replication_slots must be greater than 0");
+
+	PQclear(res);
+
+	res = executeQueryOrDie(conn, "SHOW wal_level;");
+	wal_level = PQgetvalue(res, 0, 0);
+
+	if (strcmp(wal_level, "logical") != 0)
+		pg_fatal("wal_level must be \"logical\", but is set to \"%s\"",
+				 wal_level);
+
+	PQclear(res);
+
+	PQfinish(conn);
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/dump.c b/src/bin/pg_upgrade/dump.c
index 6c8c82dca8..e6b90864f5 100644
--- a/src/bin/pg_upgrade/dump.c
+++ b/src/bin/pg_upgrade/dump.c
@@ -59,6 +59,30 @@ generate_old_dump(void)
 						   log_opts.dumpdir,
 						   sql_file_name, escaped_connstr.data);
 
+		/*
+		 * Dump logical replication slots if needed.
+		 *
+		 * XXX We cannot dump replication slots at the same time as the schema
+		 * dump because we need to separate the timing of restoring
+		 * replication slots and other objects. Replication slots, in
+		 * particular, should not be restored before executing the pg_resetwal
+		 * command because it will remove WALs that are required by the slots.
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			char		slots_file_name[MAXPGPATH];
+
+			snprintf(slots_file_name, sizeof(slots_file_name),
+					 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+			parallel_exec_prog(log_file_name, NULL,
+							   "\"%s/pg_dump\" %s --logical-replication-slots-only "
+							   "--quote-all-identifiers --binary-upgrade %s "
+							   "--file=\"%s/%s\" %s",
+							   new_cluster.bindir, cluster_conn_opts(&old_cluster),
+							   log_opts.verbose ? "--verbose" : "",
+							   log_opts.dumpdir,
+							   slots_file_name, escaped_connstr.data);
+		}
 		termPQExpBuffer(&escaped_connstr);
 	}
 
diff --git a/src/bin/pg_upgrade/info.c b/src/bin/pg_upgrade/info.c
index a9988abfe1..a605056bbe 100644
--- a/src/bin/pg_upgrade/info.c
+++ b/src/bin/pg_upgrade/info.c
@@ -23,10 +23,11 @@ static void free_db_and_rel_infos(DbInfoArr *db_arr);
 static void get_template0_info(ClusterInfo *cluster);
 static void get_db_infos(ClusterInfo *cluster);
 static void get_rel_infos(ClusterInfo *cluster, DbInfo *dbinfo);
+static void get_logical_slot_infos_per_db(ClusterInfo *cluster, DbInfo *dbinfo);
 static void free_rel_infos(RelInfoArr *rel_arr);
 static void print_db_infos(DbInfoArr *db_arr);
 static void print_rel_infos(RelInfoArr *rel_arr);
-
+static void print_slot_infos(LogicalSlotInfoArr *slot_arr);
 
 /*
  * gen_db_file_maps()
@@ -600,6 +601,94 @@ get_rel_infos(ClusterInfo *cluster, DbInfo *dbinfo)
 	dbinfo->rel_arr.nrels = num_rels;
 }
 
+/*
+ * get_logical_slot_infos()
+ *
+ * Higher level routine to generate LogicalSlotInfoArr for all databases.
+ */
+void
+get_logical_slot_infos(ClusterInfo *cluster)
+{
+	int			dbnum;
+
+	if (cluster == &old_cluster)
+		pg_log(PG_VERBOSE, "\nsource databases:");
+	else
+		pg_log(PG_VERBOSE, "\ntarget databases:");
+
+	for (dbnum = 0; dbnum < cluster->dbarr.ndbs; dbnum++)
+	{
+		DbInfo	   *pDbInfo = &cluster->dbarr.dbs[dbnum];
+
+		get_logical_slot_infos_per_db(cluster, pDbInfo);
+
+		if (log_opts.verbose)
+		{
+			pg_log(PG_VERBOSE, "Database: \"%s\"", pDbInfo->db_name);
+			print_slot_infos(&pDbInfo->slot_arr);
+		}
+	}
+}
+
+/*
+ * get_logical_slot_infos_per_db()
+ *
+ * gets the LogicalSlotInfos for all the logical replication slots of the database
+ * referred to by "dbinfo".
+ */
+static void
+get_logical_slot_infos_per_db(ClusterInfo *cluster, DbInfo *dbinfo)
+{
+	PGconn	   *conn = connectToServer(cluster,
+									   dbinfo->db_name);
+	PGresult   *res;
+	LogicalSlotInfo *slotinfos = NULL;
+
+	int			num_slots;
+
+	char		query[QUERY_ALLOC];
+
+	query[0] = '\0';			/* initialize query string to empty */
+
+	snprintf(query, sizeof(query),
+			 "SELECT slot_name, plugin, two_phase "
+			 "FROM pg_catalog.pg_replication_slots "
+			 "WHERE database = current_database() AND temporary = false "
+			 "AND wal_status IN ('reserved', 'extended');");
+
+	res = executeQueryOrDie(conn, "%s", query);
+
+	num_slots = PQntuples(res);
+
+	if (num_slots)
+	{
+		int			slotnum;
+		int			i_slotname;
+		int			i_plugin;
+		int			i_twophase;
+
+		slotinfos = (LogicalSlotInfo *) pg_malloc(sizeof(LogicalSlotInfo) * num_slots);
+
+		i_slotname = PQfnumber(res, "slot_name");
+		i_plugin = PQfnumber(res, "plugin");
+		i_twophase = PQfnumber(res, "two_phase");
+
+		for (slotnum = 0; slotnum < num_slots; slotnum++)
+		{
+			LogicalSlotInfo *curr = &slotinfos[slotnum];
+
+			curr->slotname = pg_strdup(PQgetvalue(res, slotnum, i_slotname));
+			curr->plugin = pg_strdup(PQgetvalue(res, slotnum, i_plugin));
+			curr->two_phase = (strcmp(PQgetvalue(res, slotnum, i_twophase), "t") == 0);
+		}
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	dbinfo->slot_arr.slots = slotinfos;
+	dbinfo->slot_arr.nslots = num_slots;
+}
 
 static void
 free_db_and_rel_infos(DbInfoArr *db_arr)
@@ -610,6 +699,14 @@ free_db_and_rel_infos(DbInfoArr *db_arr)
 	{
 		free_rel_infos(&db_arr->dbs[dbnum].rel_arr);
 		pg_free(db_arr->dbs[dbnum].db_name);
+
+		/*
+		 * db_arr has an additional attribute, LogicalSlotInfoArr slot_arr,
+		 * but there is no need to free it. It has a valid member only when
+		 * the cluster had logical replication slots in the previous call.
+		 * However, in this case, a FATAL error is thrown, and we cannot reach
+		 * this point.
+		 */
 	}
 	pg_free(db_arr->dbs);
 	db_arr->dbs = NULL;
@@ -660,3 +757,15 @@ print_rel_infos(RelInfoArr *rel_arr)
 			   rel_arr->rels[relnum].reloid,
 			   rel_arr->rels[relnum].tablespace);
 }
+
+static void
+print_slot_infos(LogicalSlotInfoArr *slot_arr)
+{
+	int			slotnum;
+
+	for (slotnum = 0; slotnum < slot_arr->nslots; slotnum++)
+		pg_log(PG_VERBOSE, "slotname: \"%s\", plugin: \"%s\", two_phase: %d",
+			   slot_arr->slots[slotnum].slotname,
+			   slot_arr->slots[slotnum].plugin,
+			   slot_arr->slots[slotnum].two_phase);
+}
diff --git a/src/bin/pg_upgrade/meson.build b/src/bin/pg_upgrade/meson.build
index 12a97f84e2..228f29b688 100644
--- a/src/bin/pg_upgrade/meson.build
+++ b/src/bin/pg_upgrade/meson.build
@@ -42,6 +42,7 @@ tests += {
     'tests': [
       't/001_basic.pl',
       't/002_pg_upgrade.pl',
+      't/003_logical_replication_slots.pl',
     ],
     'test_kwargs': {'priority': 40}, # pg_upgrade tests are slow
   },
diff --git a/src/bin/pg_upgrade/option.c b/src/bin/pg_upgrade/option.c
index 640361009e..df66a5ffe6 100644
--- a/src/bin/pg_upgrade/option.c
+++ b/src/bin/pg_upgrade/option.c
@@ -57,6 +57,7 @@ parseCommandLine(int argc, char *argv[])
 		{"verbose", no_argument, NULL, 'v'},
 		{"clone", no_argument, NULL, 1},
 		{"copy", no_argument, NULL, 2},
+		{"include-logical-replication-slots", no_argument, NULL, 3},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -199,6 +200,10 @@ parseCommandLine(int argc, char *argv[])
 				user_opts.transfer_mode = TRANSFER_MODE_COPY;
 				break;
 
+			case 3:
+				user_opts.include_logical_slots = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						os_info.progname);
@@ -289,6 +294,8 @@ usage(void)
 	printf(_("  -V, --version                 display version information, then exit\n"));
 	printf(_("  --clone                       clone instead of copying files to new cluster\n"));
 	printf(_("  --copy                        copy files to new cluster (default)\n"));
+	printf(_("  --include-logical-replication-slots\n"
+			 "                                upgrade logical replication slots\n"));
 	printf(_("  -?, --help                    show this help, then exit\n"));
 	printf(_("\n"
 			 "Before running pg_upgrade you must:\n"
diff --git a/src/bin/pg_upgrade/pg_upgrade.c b/src/bin/pg_upgrade/pg_upgrade.c
index 4562dafcff..49e16234ed 100644
--- a/src/bin/pg_upgrade/pg_upgrade.c
+++ b/src/bin/pg_upgrade/pg_upgrade.c
@@ -59,6 +59,7 @@ static void copy_xact_xlog_xid(void);
 static void set_frozenxids(bool minmxid_only);
 static void make_outputdirs(char *pgdata);
 static void setup(char *argv0, bool *live_check);
+static void create_logical_replication_slots(void);
 
 ClusterInfo old_cluster,
 			new_cluster;
@@ -188,6 +189,19 @@ main(int argc, char **argv)
 			  new_cluster.pgdata);
 	check_ok();
 
+	/*
+	 * Create logical replication slots if requested.
+	 *
+	 * Note: This must be done after doing pg_resetwal command because the
+	 * command will remove required WALs.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		start_postmaster(&new_cluster, true);
+		create_logical_replication_slots();
+		stop_postmaster(false);
+	}
+
 	if (user_opts.do_sync)
 	{
 		prep_status("Sync data directory to disk");
@@ -860,3 +874,50 @@ set_frozenxids(bool minmxid_only)
 
 	check_ok();
 }
+
+/*
+ * create_logical_replication_slots()
+ *
+ * Similar to create_new_objects() but only restores logical replication slots.
+ */
+static void
+create_logical_replication_slots(void)
+{
+	int			dbnum;
+
+	prep_status_progress("Restoring logical replication slots in the new cluster");
+
+	for (dbnum = 0; dbnum < old_cluster.dbarr.ndbs; dbnum++)
+	{
+		char		slots_file_name[MAXPGPATH],
+					log_file_name[MAXPGPATH];
+		DbInfo	   *old_db = &old_cluster.dbarr.dbs[dbnum];
+
+		pg_log(PG_STATUS, "%s", old_db->db_name);
+
+		snprintf(slots_file_name, sizeof(slots_file_name),
+				 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+		snprintf(log_file_name, sizeof(log_file_name),
+				 DB_DUMP_LOG_FILE_MASK, old_db->db_oid);
+
+		parallel_exec_prog(log_file_name,
+						   NULL,
+						   "\"%s/psql\" %s --echo-queries --set ON_ERROR_STOP=on "
+						   "--no-psqlrc --dbname %s -f \"%s/%s\"",
+						   new_cluster.bindir,
+						   cluster_conn_opts(&new_cluster),
+						   old_db->db_name,
+						   log_opts.dumpdir,
+						   slots_file_name);
+	}
+
+	/* reap all children */
+	while (reap_child(true) == true)
+		;
+
+	end_progress_output();
+	check_ok();
+
+	/* update new_cluster info again */
+	get_logical_slot_infos(&new_cluster);
+}
diff --git a/src/bin/pg_upgrade/pg_upgrade.h b/src/bin/pg_upgrade/pg_upgrade.h
index 3eea0139c7..7adbb50807 100644
--- a/src/bin/pg_upgrade/pg_upgrade.h
+++ b/src/bin/pg_upgrade/pg_upgrade.h
@@ -29,6 +29,7 @@
 /* contains both global db information and CREATE DATABASE commands */
 #define GLOBALS_DUMP_FILE	"pg_upgrade_dump_globals.sql"
 #define DB_DUMP_FILE_MASK	"pg_upgrade_dump_%u.custom"
+#define DB_DUMP_LOGICAL_SLOTS_FILE_MASK	"pg_upgrade_dump_%u_logical_slots.sql"
 
 /*
  * Base directories that include all the files generated internally, from the
@@ -150,6 +151,22 @@ typedef struct
 	int			nrels;
 } RelInfoArr;
 
+/*
+ * Structure to store logical replication slot information
+ */
+typedef struct
+{
+	char	   *slotname;		/* slot name */
+	char	   *plugin;			/* plugin */
+	bool		two_phase;		/* Can the slot decode 2PC? */
+} LogicalSlotInfo;
+
+typedef struct
+{
+	LogicalSlotInfo *slots;
+	int			nslots;
+} LogicalSlotInfoArr;
+
 /*
  * The following structure represents a relation mapping.
  */
@@ -176,6 +193,7 @@ typedef struct
 	char		db_tablespace[MAXPGPATH];	/* database default tablespace
 											 * path */
 	RelInfoArr	rel_arr;		/* array of all user relinfos */
+	LogicalSlotInfoArr slot_arr;	/* array of all LogicalSlotInfo */
 } DbInfo;
 
 /*
@@ -304,6 +322,8 @@ typedef struct
 	transferMode transfer_mode; /* copy files or link them? */
 	int			jobs;			/* number of processes/threads to use */
 	char	   *socketdir;		/* directory to use for Unix sockets */
+	bool		include_logical_slots;	/* true -> dump and restore logical
+										 * replication slots */
 } UserOpts;
 
 typedef struct
@@ -400,6 +420,7 @@ FileNameMap *gen_db_file_maps(DbInfo *old_db,
 							  DbInfo *new_db, int *nmaps, const char *old_pgdata,
 							  const char *new_pgdata);
 void		get_db_and_rel_infos(ClusterInfo *cluster);
+void		get_logical_slot_infos(ClusterInfo *cluster);
 
 /* option.c */
 
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
new file mode 100644
index 0000000000..525a7704cf
--- /dev/null
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -0,0 +1,115 @@
+# Copyright (c) 2023, PostgreSQL Global Development Group
+
+# Tests for upgrading replication slots
+
+use strict;
+use warnings;
+
+use File::Path qw(rmtree);
+
+use PostgreSQL::Test::Cluster;
+use PostgreSQL::Test::Utils;
+use Test::More;
+
+# Can be changed to test the other modes
+my $mode = $ENV{PG_TEST_PG_UPGRADE_MODE} || '--copy';
+
+# Initialize old node
+my $old_node = PostgreSQL::Test::Cluster->new('old_node');
+$old_node->init(allows_streaming => 'logical');
+$old_node->start;
+
+# Initialize new node
+my $new_node = PostgreSQL::Test::Cluster->new('new_node');
+$new_node->init(allows_streaming => 1);
+
+my $bindir = $new_node->config_data('--bindir');
+
+$old_node->stop;
+
+# Cause a failure at the start of pg_upgrade because wal_level is replica
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with wrong wal_level');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+# Preparations for the subsequent test. The case max_replication_slots is set
+# to 0 is prohibited.
+$new_node->append_conf('postgresql.conf', "wal_level = 'logical'");
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 0");
+
+# Cause a failure at the start of pg_upgrade because max_replication_slots is 0
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with wrong max_replication_slots');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+# Preparations for the subsequent test. max_replication_slots is set to
+# non-zero value to succeed the pg_upgrade
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 10");
+
+# Create a slot on old node, and generate WALs
+$old_node->start;
+$old_node->safe_psql(
+	'postgres', qq[
+	SELECT pg_create_logical_replication_slot('test_slot', 'test_decoding', false, true);
+	CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a;
+]);
+
+my $result = $old_node->safe_psql('postgres',
+	"SELECT count (*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)"
+);
+is($result, qq(12), 'ensure WALs are not consumed yet');
+$old_node->stop;
+
+# Actual run, pg_upgrade_output.d is removed at the end
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots'
+	],
+	'run of pg_upgrade of old node');
+ok( !-d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+$new_node->start;
+$result = $new_node->safe_psql('postgres',
+	"SELECT slot_name, two_phase FROM pg_replication_slots");
+is($result, qq(test_slot|t), 'check the slot exists on new node');
+
+done_testing();
diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list
index 260854747b..965219524d 100644
--- a/src/tools/pgindent/typedefs.list
+++ b/src/tools/pgindent/typedefs.list
@@ -1487,6 +1487,7 @@ LogicalRepBeginData
 LogicalRepCommitData
 LogicalRepCommitPreparedTxnData
 LogicalRepCtxStruct
+LogicalReplicationSlotInfo
 LogicalRepMsgType
 LogicalRepPartMapEntry
 LogicalRepPreparedTxnData
@@ -1499,6 +1500,8 @@ LogicalRepTupleData
 LogicalRepTyp
 LogicalRepWorker
 LogicalRewriteMappingData
+LogicalSlotInfo
+LogicalSlotInfoArr
 LogicalTape
 LogicalTapeSet
 LsnReadQueue
-- 
2.27.0

From 874ceca95db1e9d0818ac39c2169bacf9a3fe975 Mon Sep 17 00:00:00 2001
From: Julien Rouhaud <julien.rouhaud@free.fr>
Date: Fri, 14 Apr 2023 13:49:09 +0800
Subject: [PATCH v15 2/4] Always persist to disk logical slots during a
 shutdown checkpoint.

It's entirely possible for a logical slot to have a confirmed_flush_lsn higher
than the last value saved on disk while not being marked as dirty.  It's
currently not a problem to lose that value during a clean shutdown / restart
cycle, but a later patch adding support for pg_upgrade of publications and
logical slots will rely on that value being properly persisted to disk.

Author: Julien Rouhaud
Reviewed-by: Wang Wei
Discussion: FIXME
---
 src/backend/access/transam/xlog.c |  2 +-
 src/backend/replication/slot.c    | 26 ++++++++++++++++----------
 src/include/replication/slot.h    |  2 +-
 3 files changed, 18 insertions(+), 12 deletions(-)

diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index b2430f617c..384245c157 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7011,7 +7011,7 @@ static void
 CheckPointGuts(XLogRecPtr checkPointRedo, int flags)
 {
 	CheckPointRelationMap();
-	CheckPointReplicationSlots();
+	CheckPointReplicationSlots(flags & CHECKPOINT_IS_SHUTDOWN);
 	CheckPointSnapBuild();
 	CheckPointLogicalRewriteHeap();
 	CheckPointReplicationOrigin();
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index 8021aaa0a8..526323e87b 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -109,7 +109,8 @@ static void ReplicationSlotDropPtr(ReplicationSlot *slot);
 /* internal persistency functions */
 static void RestoreSlotFromDisk(const char *name);
 static void CreateSlotOnDisk(ReplicationSlot *slot);
-static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel);
+static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+						   bool is_shutdown);
 
 /*
  * Report shared-memory space needed by ReplicationSlotsShmemInit.
@@ -783,7 +784,7 @@ ReplicationSlotSave(void)
 	Assert(MyReplicationSlot != NULL);
 
 	sprintf(path, "pg_replslot/%s", NameStr(MyReplicationSlot->data.name));
-	SaveSlotToPath(MyReplicationSlot, path, ERROR);
+	SaveSlotToPath(MyReplicationSlot, path, ERROR, false);
 }
 
 /*
@@ -1565,11 +1566,10 @@ restart:
 /*
  * Flush all replication slots to disk.
  *
- * This needn't actually be part of a checkpoint, but it's a convenient
- * location.
+ * is_shutdown is true in case of a shutdown checkpoint.
  */
 void
-CheckPointReplicationSlots(void)
+CheckPointReplicationSlots(bool is_shutdown)
 {
 	int			i;
 
@@ -1594,7 +1594,7 @@ CheckPointReplicationSlots(void)
 
 		/* save the slot to disk, locking is handled in SaveSlotToPath() */
 		sprintf(path, "pg_replslot/%s", NameStr(s->data.name));
-		SaveSlotToPath(s, path, LOG);
+		SaveSlotToPath(s, path, LOG, is_shutdown);
 	}
 	LWLockRelease(ReplicationSlotAllocationLock);
 }
@@ -1700,7 +1700,7 @@ CreateSlotOnDisk(ReplicationSlot *slot)
 
 	/* Write the actual state file. */
 	slot->dirty = true;			/* signal that we really need to write */
-	SaveSlotToPath(slot, tmppath, ERROR);
+	SaveSlotToPath(slot, tmppath, ERROR, false);
 
 	/* Rename the directory into place. */
 	if (rename(tmppath, path) != 0)
@@ -1726,7 +1726,8 @@ CreateSlotOnDisk(ReplicationSlot *slot)
  * Shared functionality between saving and creating a replication slot.
  */
 static void
-SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
+SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+			   bool is_shutdown)
 {
 	char		tmppath[MAXPGPATH];
 	char		path[MAXPGPATH];
@@ -1740,8 +1741,13 @@ SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
 	slot->just_dirtied = false;
 	SpinLockRelease(&slot->mutex);
 
-	/* and don't do anything if there's nothing to write */
-	if (!was_dirty)
+	/*
+	 * and don't do anything if there's nothing to write, unless it's this is
+	 * called for a logical slot during a shutdown checkpoint, as we want to
+	 * persist the confirmed_flush_lsn in that case, even if that's the only
+	 * modification.
+	 */
+	if (!was_dirty && (SlotIsPhysical(slot) || !is_shutdown))
 		return;
 
 	LWLockAcquire(&slot->io_in_progress_lock, LW_EXCLUSIVE);
diff --git a/src/include/replication/slot.h b/src/include/replication/slot.h
index a8a89dc784..7ca37c9f70 100644
--- a/src/include/replication/slot.h
+++ b/src/include/replication/slot.h
@@ -241,7 +241,7 @@ extern void ReplicationSlotNameForTablesync(Oid suboid, Oid relid, char *syncslo
 extern void ReplicationSlotDropAtPubNode(WalReceiverConn *wrconn, char *slotname, bool missing_ok);
 
 extern void StartupReplicationSlots(void);
-extern void CheckPointReplicationSlots(void);
+extern void CheckPointReplicationSlots(bool is_shutdown);
 
 extern void CheckSlotRequirements(void);
 extern void CheckSlotPermissions(void);
-- 
2.27.0

From 08e2bd69b941c9bedfe9f5a927d7339a15ad3e12 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Fri, 14 Apr 2023 08:59:03 +0000
Subject: [PATCH v15 3/4] pg_upgrade: Add check function for
 --include-logical-replication-slots option

Author: Hayato Kuroda
Reviewed-by: Wang Wei
---
 src/bin/pg_upgrade/check.c                    | 80 ++++++++++++++++++-
 .../t/003_logical_replication_slots.pl        | 30 ++++++-
 2 files changed, 108 insertions(+), 2 deletions(-)

diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index f08fcc7705..391144570f 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -9,7 +9,10 @@
 
 #include "postgres_fe.h"
 
+#include "access/xlogrecord.h"
+#include "access/xlog_internal.h"
 #include "catalog/pg_authid_d.h"
+#include "catalog/pg_control.h"
 #include "catalog/pg_collation.h"
 #include "fe_utils/string_utils.h"
 #include "mb/pg_wchar.h"
@@ -31,7 +34,7 @@ static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
 static void check_for_parameter_settings(ClusterInfo *new_cluster);
-
+static void check_for_confirmed_flush_lsn(ClusterInfo *cluster);
 
 /*
  * fix_path_separator
@@ -108,6 +111,8 @@ check_and_dump_old_cluster(bool live_check)
 	check_for_composite_data_type_usage(&old_cluster);
 	check_for_reg_data_type_usage(&old_cluster);
 	check_for_isn_and_int8_passing_mismatch(&old_cluster);
+	if (user_opts.include_logical_slots)
+		check_for_confirmed_flush_lsn(&old_cluster);
 
 	/*
 	 * PG 16 increased the size of the 'aclitem' type, which breaks the
@@ -1446,6 +1451,10 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 	int			max_replication_slots;
 	char	   *wal_level;
 
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(new_cluster->major_version) <= 1500)
+		return;
+
 	prep_status("Checking for logical replication slots");
 
 	res = executeQueryOrDie(conn, "SHOW max_replication_slots;");
@@ -1469,3 +1478,72 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 
 	check_ok();
 }
+
+/*
+ * Verify that all logical replication slots consumed all WALs, except a
+ * CHECKPOINT_SHUTDOWN record.
+ */
+static void
+check_for_confirmed_flush_lsn(ClusterInfo *cluster)
+{
+	int			i,
+				ntups,
+				i_slotname;
+	bool		is_error = false;
+	PGresult   *res;
+	DbInfo	   *active_db = &cluster->dbarr.dbs[0];
+	PGconn	   *conn = connectToServer(cluster, active_db->db_name);
+
+	Assert(user_opts.include_logical_slots);
+
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(cluster->major_version) <= 1500)
+		return;
+
+	prep_status("Checking for logical replication slots");
+
+	/*
+	 * Check that all logical replication slots have reached the current WAL
+	 * position, except for the CHECKPOINT_SHUTDOWN record. Even if all WALs
+	 * are consumed before shutting down the node, the checkpointer generates
+	 * a CHECKPOINT_SHUTDOWN record at shutdown, which cannot be consumed by
+	 * any slots. Therefore, we must allow for a difference between
+	 * pg_current_wal_insert_lsn() and confirmed_flush_lsn.
+	 */
+#define SHUTDOWN_RECORD_SIZE  (SizeOfXLogRecord + \
+							   SizeOfXLogRecordDataHeaderShort + \
+							   sizeof(CheckPoint))
+
+	res = executeQueryOrDie(conn,
+							"SELECT slot_name FROM pg_catalog.pg_replication_slots "
+							"WHERE (pg_catalog.pg_current_wal_insert_lsn() - confirmed_flush_lsn) > %d "
+							"AND temporary = false AND wal_status IN ('reserved', 'extended');",
+							(int) (SizeOfXLogLongPHD + SHUTDOWN_RECORD_SIZE));
+
+#undef SHUTDOWN_RECORD_SIZE
+
+	ntups = PQntuples(res);
+	i_slotname = PQfnumber(res, "slot_name");
+
+	for (i = 0; i < ntups; i++)
+	{
+		char	   *slotname;
+
+		is_error = true;
+
+		slotname = PQgetvalue(res, i, i_slotname);
+
+		pg_log(PG_WARNING,
+			   "\nWARNING: logical replication slot \"%s\" has not consumed WALs yet",
+			   slotname);
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	if (is_error)
+		pg_fatal("--include-logical-replication-slots requires that all "
+				 "logical replication slots consumed all the WALs");
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
index 525a7704cf..21fefca084 100644
--- a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -85,11 +85,39 @@ $old_node->safe_psql(
 ]);
 
 my $result = $old_node->safe_psql('postgres',
-	"SELECT count (*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)"
+	"SELECT count(*) FROM pg_logical_slot_peek_changes('test_slot', NULL, NULL)"
 );
+
 is($result, qq(12), 'ensure WALs are not consumed yet');
 $old_node->stop;
 
+# Cause a failure at the start of pg_upgrade because test_slot does not
+# finish consuming all the WALs
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with idle replication slots');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+$old_node->start;
+$old_node->safe_psql('postgres',
+	"SELECT count (*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)"
+);
+$old_node->stop;
+
 # Actual run, pg_upgrade_output.d is removed at the end
 command_ok(
 	[
-- 
2.27.0

From 0258212ea524a5e0250939407f705121e0597a10 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Mon, 24 Apr 2023 11:03:28 +0000
Subject: [PATCH v15 4/4] Change the method used to check logical replication
 slots during the live check

When a live check is requested, there is a possibility of additional changes
occurring, which may cause the current WAL position to exceed the confirmed_flush_lsn
of the slot. As a result, we check the confirmed_flush_lsn of each logical slot
instead. This is sufficient as all the WAL records will be sent during thepublisher's
shutdown.

Author: Hayato Kuroda
Reviewed-by: Wang Wei
---
 src/bin/pg_upgrade/check.c                    | 68 ++++++++++++++++++-
 .../t/003_logical_replication_slots.pl        | 66 +++++++++++++++++-
 2 files changed, 130 insertions(+), 4 deletions(-)

diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index 391144570f..b3da2e9193 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -35,6 +35,7 @@ static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
 static void check_for_parameter_settings(ClusterInfo *new_cluster);
 static void check_for_confirmed_flush_lsn(ClusterInfo *cluster);
+static void check_are_logical_slots_active(ClusterInfo *cluster);
 
 /*
  * fix_path_separator
@@ -112,7 +113,19 @@ check_and_dump_old_cluster(bool live_check)
 	check_for_reg_data_type_usage(&old_cluster);
 	check_for_isn_and_int8_passing_mismatch(&old_cluster);
 	if (user_opts.include_logical_slots)
-		check_for_confirmed_flush_lsn(&old_cluster);
+	{
+		/*
+		 * The method used to check logical replication slots is dependent on
+		 * the value of the live_check parameter. This change was implemented
+		 * because, during a live check, it is possible for additional changes
+		 * to occur at the old node, which could cause the current WAL position
+		 * to exceed the confirmed_flush_lsn of the slot.
+		 */
+		if (live_check)
+			check_are_logical_slots_active(&old_cluster);
+		else
+			check_for_confirmed_flush_lsn(&old_cluster);
+	}
 
 	/*
 	 * PG 16 increased the size of the 'aclitem' type, which breaks the
@@ -1479,6 +1492,59 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 	check_ok();
 }
 
+/*
+ * Verify that all logical replication slots are active
+ */
+static void
+check_are_logical_slots_active(ClusterInfo *cluster)
+{
+	int			i,
+				ntups,
+				i_slotname;
+	bool		is_error = false;
+	PGresult   *res;
+	DbInfo	   *active_db = &cluster->dbarr.dbs[0];
+	PGconn	   *conn = connectToServer(cluster, active_db->db_name);
+
+	Assert(user_opts.include_logical_slots);
+
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(cluster->major_version) <= 1500)
+		return;
+
+	prep_status("Checking for logical replication slots");
+
+	res = executeQueryOrDie(conn,
+							"SELECT slot_name FROM pg_catalog.pg_replication_slots "
+							"WHERE active IS FALSE "
+							"AND temporary = false AND wal_status IN ('reserved', 'extended');");
+
+	ntups = PQntuples(res);
+	i_slotname = PQfnumber(res, "slot_name");
+
+	for (i = 0; i < ntups; i++)
+	{
+		char	   *slotname;
+
+		is_error = true;
+
+		slotname = PQgetvalue(res, i, i_slotname);
+
+		pg_log(PG_WARNING,
+			   "\nWARNING: logical replication slot \"%s\" is not active",
+			   slotname);
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	if (is_error)
+		pg_fatal("--include-logical-replication-slots with --check requires that "
+				 "all logical replication slots are active");
+
+	check_ok();
+}
+
 /*
  * Verify that all logical replication slots consumed all WALs, except a
  * CHECKPOINT_SHUTDOWN record.
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
index 21fefca084..d7fd864bd7 100644
--- a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -19,6 +19,10 @@ my $old_node = PostgreSQL::Test::Cluster->new('old_node');
 $old_node->init(allows_streaming => 'logical');
 $old_node->start;
 
+# Initialize subscriber, which will be used only for --check
+my $subscriber = PostgreSQL::Test::Cluster->new('subscriber');
+$subscriber->init(allows_streaming => 'logical');
+
 # Initialize new node
 my $new_node = PostgreSQL::Test::Cluster->new('new_node');
 $new_node->init(allows_streaming => 1);
@@ -76,15 +80,71 @@ rmtree($new_node->data_dir . "/pg_upgrade_output.d");
 # non-zero value to succeed the pg_upgrade
 $new_node->append_conf('postgresql.conf', "max_replication_slots = 10");
 
-# Create a slot on old node, and generate WALs
+# Setup logical replication
 $old_node->start;
+$old_node->safe_psql('postgres',
+	"CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a");
+$old_node->safe_psql('postgres', "CREATE PUBLICATION pub FOR ALL TABLES");
+my $old_connstr = $old_node->connstr . ' dbname=postgres';
+
+$subscriber->start;
+$subscriber->safe_psql('postgres', "CREATE TABLE tbl (a int)");
+$subscriber->safe_psql('postgres',
+	"CREATE SUBSCRIPTION sub CONNECTION '$old_connstr' PUBLICATION pub WITH (copy_data = true)"
+);
+
+# Wait for initial table sync to finish
+$subscriber->wait_for_subscription_sync($old_node, 'sub');
+
+my $result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(10), 'check initial rows on subscriber');
+
+# Start a background session and open a transaction (not committed yet)
+my $bsession = $old_node->background_psql('postgres');
+$bsession->query_safe(
+	q{
+BEGIN;
+INSERT INTO tbl VALUES (generate_series(11, 20))
+});
+
+$result = $old_node->safe_psql('postgres',
+	"SELECT count(*) FROM pg_replication_slots WHERE pg_current_wal_insert_lsn() > confirmed_flush_lsn"
+);
+is($result, qq(1),
+	'check the current WAL position exceeds confirmed_flush_lsn');
+
+# Run pg_upgrade --check. In the command the status of each logical slots will
+# be checked and then this will be succeeded.
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+		'--check'
+	],
+	'run of pg_upgrade of old node');
+ok( !-d $old_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+# Cleanup
+$bsession->query_safe("ABORT");
+$bsession->quit;
+$subscriber->safe_psql('postgres', "DROP SUBSCRIPTION sub");
+
+# Create a slot on old node, and generate WALs
 $old_node->safe_psql(
 	'postgres', qq[
 	SELECT pg_create_logical_replication_slot('test_slot', 'test_decoding', false, true);
-	CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a;
+	INSERT INTO tbl VALUES (generate_series(11, 20));
 ]);
 
-my $result = $old_node->safe_psql('postgres',
+$result = $old_node->safe_psql('postgres',
 	"SELECT count(*) FROM pg_logical_slot_peek_changes('test_slot', NULL, NULL)"
 );
 
-- 
2.27.0

From 9c80fc25dc9d922ef5d793457c7b84d4b5713e60 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Tue, 4 Apr 2023 05:49:34 +0000
Subject: [PATCH v16 1/4] pg_upgrade: Add --include-logical-replication-slots
 option

This commit introduces a new pg_upgrade option called "--include-logical-replication-slots".
This allows nodes with logical replication slots to be upgraded. The commit can
be divided into two parts: one for pg_dump and another for pg_upgrade.

For pg_dump this commit includes a new option called "--logical-replication-slots-only".
This option can be used to dump logical replication slots. When this option is
specified, the slot_name, plugin, and two_phase parameters are extracted from
pg_replication_slots. An SQL file is then generated which executes
pg_create_logical_replication_slot() with the extracted parameters.

For pg_upgrade, when '--include-logical-replication-slots' is specified, it executes
pg_dump with the new "--logical-replication-slots-only" option and restores from the
dump. Note that we cannot dump replication slots at the same time as the schema
dump because we need to separate the timing of restoring replication slots and
other objects. Replication slots, in  particular, should not be restored before
executing the pg_resetwal command because it will remove WALs that are required
by the slots.

The significant advantage of this commit is that it makes it easy to continue
logical replication even after upgrading the publisher node. Previously, pg_upgrade
allowed copying publications to a new node. With this new commit, adjusting the
connection string to the new publisher will cause the apply worker on the subscriber
to connect to the new publisher automatically. This enables seamless continuation
of logical replication, even after an upgrade.

Author: Hayato Kuroda
Reviewed-by: Peter Smith, Julien Rouhaud, Vignesh C, Wang Wei
---
 doc/src/sgml/ref/pgupgrade.sgml               |  11 ++
 src/bin/pg_dump/pg_backup.h                   |   1 +
 src/bin/pg_dump/pg_dump.c                     | 154 ++++++++++++++++++
 src/bin/pg_dump/pg_dump.h                     |  14 +-
 src/bin/pg_dump/pg_dump_sort.c                |  11 +-
 src/bin/pg_upgrade/Makefile                   |   3 +
 src/bin/pg_upgrade/check.c                    |  71 ++++++++
 src/bin/pg_upgrade/dump.c                     |  24 +++
 src/bin/pg_upgrade/info.c                     | 110 ++++++++++++-
 src/bin/pg_upgrade/meson.build                |   1 +
 src/bin/pg_upgrade/option.c                   |   7 +
 src/bin/pg_upgrade/pg_upgrade.c               |  61 +++++++
 src/bin/pg_upgrade/pg_upgrade.h               |  21 +++
 .../t/003_logical_replication_slots.pl        | 114 +++++++++++++
 src/tools/pgindent/typedefs.list              |   3 +
 15 files changed, 602 insertions(+), 4 deletions(-)
 create mode 100644 src/bin/pg_upgrade/t/003_logical_replication_slots.pl

diff --git a/doc/src/sgml/ref/pgupgrade.sgml b/doc/src/sgml/ref/pgupgrade.sgml
index 7816b4c685..94e90ff506 100644
--- a/doc/src/sgml/ref/pgupgrade.sgml
+++ b/doc/src/sgml/ref/pgupgrade.sgml
@@ -240,6 +240,17 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--include-logical-replication-slots</option></term>
+      <listitem>
+       <para>
+        Upgrade logical replication slots. Only permanent replication slots
+        are included. Note that pg_upgrade does not check the installation of
+        plugins.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
       <term><option>-?</option></term>
       <term><option>--help</option></term>
diff --git a/src/bin/pg_dump/pg_backup.h b/src/bin/pg_dump/pg_backup.h
index aba780ef4b..0a4e931f9b 100644
--- a/src/bin/pg_dump/pg_backup.h
+++ b/src/bin/pg_dump/pg_backup.h
@@ -187,6 +187,7 @@ typedef struct _dumpOptions
 	int			use_setsessauth;
 	int			enable_row_security;
 	int			load_via_partition_root;
+	int			logical_slots_only;
 
 	/* default, if no "inclusion" switches appear, is to dump everything */
 	bool		include_everything;
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index 5dab1ba9ea..898ce42005 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -328,6 +328,9 @@ static void setupDumpWorker(Archive *AH);
 static TableInfo *getRootTableInfo(const TableInfo *tbinfo);
 static bool forcePartitionRootLoad(const TableInfo *tbinfo);
 
+static void getLogicalReplicationSlots(Archive *fout);
+static void dumpLogicalReplicationSlot(Archive *fout,
+									   const LogicalReplicationSlotInfo *slotinfo);
 
 int
 main(int argc, char **argv)
@@ -431,6 +434,7 @@ main(int argc, char **argv)
 		{"table-and-children", required_argument, NULL, 12},
 		{"exclude-table-and-children", required_argument, NULL, 13},
 		{"exclude-table-data-and-children", required_argument, NULL, 14},
+		{"logical-replication-slots-only", no_argument, NULL, 15},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -657,6 +661,10 @@ main(int argc, char **argv)
 										  optarg);
 				break;
 
+			case 15:			/* dump only replication slot(s) */
+				dopt.logical_slots_only = true;
+				break;
+
 			default:
 				/* getopt_long already emitted a complaint */
 				pg_log_error_hint("Try \"%s --help\" for more information.", progname);
@@ -714,6 +722,21 @@ main(int argc, char **argv)
 	if (dopt.do_nothing && dopt.dump_inserts == 0)
 		pg_fatal("option --on-conflict-do-nothing requires option --inserts, --rows-per-insert, or --column-inserts");
 
+	if (dopt.logical_slots_only)
+	{
+		if (!dopt.binary_upgrade)
+			pg_fatal("options --logical-replication-slots-only requires option --binary-upgrade");
+
+		if (dopt.dataOnly)
+			pg_fatal("options --logical-replication-slots-only and -a/--data-only cannot be used together");
+
+		if (dopt.schemaOnly)
+			pg_fatal("options --logical-replication-slots-only and -s/--schema-only cannot be used together");
+
+		if (dopt.outputClean)
+			pg_fatal("options --logical-replication-slots-only and -c/--clean cannot be used together");
+	}
+
 	/* Identify archive format to emit */
 	archiveFormat = parseArchiveFormat(format, &archiveMode);
 
@@ -876,6 +899,16 @@ main(int argc, char **argv)
 			pg_fatal("no matching extensions were found");
 	}
 
+	/*
+	 * If dump logical-replication-slots-only was requested, dump only them
+	 * and skip everything else.
+	 */
+	if (dopt.logical_slots_only)
+	{
+		getLogicalReplicationSlots(fout);
+		goto dump;
+	}
+
 	/*
 	 * Dumping LOs is the default for dumps where an inclusion switch is not
 	 * used (an "include everything" dump).  -B can be used to exclude LOs
@@ -936,6 +969,8 @@ main(int argc, char **argv)
 	if (!dopt.no_security_labels)
 		collectSecLabels(fout);
 
+dump:
+
 	/* Lastly, create dummy objects to represent the section boundaries */
 	boundaryObjs = createBoundaryObjects();
 
@@ -1109,6 +1144,11 @@ help(const char *progname)
 			 "                               servers matching PATTERN\n"));
 	printf(_("  --inserts                    dump data as INSERT commands, rather than COPY\n"));
 	printf(_("  --load-via-partition-root    load partitions via the root table\n"));
+
+	/*
+	 * The option --logical-replication-slots-only is used only by pg_upgrade
+	 * and should not be called by users, which is why it is not listed.
+	 */
 	printf(_("  --no-comments                do not dump comments\n"));
 	printf(_("  --no-publications            do not dump publications\n"));
 	printf(_("  --no-security-labels         do not dump security label assignments\n"));
@@ -10237,6 +10277,10 @@ dumpDumpableObject(Archive *fout, DumpableObject *dobj)
 		case DO_SUBSCRIPTION:
 			dumpSubscription(fout, (const SubscriptionInfo *) dobj);
 			break;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			dumpLogicalReplicationSlot(fout,
+									   (const LogicalReplicationSlotInfo *) dobj);
+			break;
 		case DO_PRE_DATA_BOUNDARY:
 		case DO_POST_DATA_BOUNDARY:
 			/* never dumped, nothing to do */
@@ -18218,6 +18262,7 @@ addBoundaryDependencies(DumpableObject **dobjs, int numObjs,
 			case DO_PUBLICATION_REL:
 			case DO_PUBLICATION_TABLE_IN_SCHEMA:
 			case DO_SUBSCRIPTION:
+			case DO_LOGICAL_REPLICATION_SLOT:
 				/* Post-data objects: must come after the post-data boundary */
 				addObjectDependency(dobj, postDataBound->dumpId);
 				break;
@@ -18479,3 +18524,112 @@ appendReloptionsArrayAH(PQExpBuffer buffer, const char *reloptions,
 	if (!res)
 		pg_log_warning("could not parse %s array", "reloptions");
 }
+
+/*
+ * getLogicalReplicationSlots
+ *	  get information about replication slots
+ */
+static void
+getLogicalReplicationSlots(Archive *fout)
+{
+	PGresult   *res;
+	LogicalReplicationSlotInfo *slotinfo;
+	PQExpBuffer query;
+
+	int			i_slotname;
+	int			i_plugin;
+	int			i_twophase;
+	int			i,
+				ntups;
+
+	/* Check whether we should dump or not */
+	if (fout->remoteVersion < 160000)
+		return;
+
+	Assert(fout->dopt->logical_slots_only);
+
+	query = createPQExpBuffer();
+
+	resetPQExpBuffer(query);
+
+	/*
+	 * Get replication slots.
+	 *
+	 * XXX: Which information must be extracted from old node? Currently three
+	 * attributes are extracted because they are used by
+	 * pg_create_logical_replication_slot().
+	 */
+	appendPQExpBufferStr(query,
+						 "SELECT slot_name, plugin, two_phase "
+						 "FROM pg_catalog.pg_replication_slots "
+						 "WHERE database = current_database() AND temporary = false "
+						 "AND wal_status IN ('reserved', 'extended');");
+
+	res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
+
+	ntups = PQntuples(res);
+
+	i_slotname = PQfnumber(res, "slot_name");
+	i_plugin = PQfnumber(res, "plugin");
+	i_twophase = PQfnumber(res, "two_phase");
+
+	slotinfo = pg_malloc(ntups * sizeof(LogicalReplicationSlotInfo));
+
+	for (i = 0; i < ntups; i++)
+	{
+		slotinfo[i].dobj.objType = DO_LOGICAL_REPLICATION_SLOT;
+
+		slotinfo[i].dobj.catId.tableoid = InvalidOid;
+		slotinfo[i].dobj.catId.oid = InvalidOid;
+		AssignDumpId(&slotinfo[i].dobj);
+
+		slotinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_slotname));
+
+		slotinfo[i].plugin = pg_strdup(PQgetvalue(res, i, i_plugin));
+		slotinfo[i].twophase = (strcmp(PQgetvalue(res, i, i_twophase), "t") == 0);
+
+		/*
+		 * Note: Currently we do not have any options to include/exclude slots
+		 * in dumping, so all the slots must be selected.
+		 */
+		slotinfo[i].dobj.dump = DUMP_COMPONENT_DEFINITION;
+	}
+	PQclear(res);
+
+	destroyPQExpBuffer(query);
+}
+
+/*
+ * dumpLogicalReplicationSlot
+ *	  dump creation functions for the given logical replication slots
+ */
+static void
+dumpLogicalReplicationSlot(Archive *fout,
+						   const LogicalReplicationSlotInfo *slotinfo)
+{
+	Assert(fout->dopt->logical_slots_only);
+
+	if (slotinfo->dobj.dump & DUMP_COMPONENT_DEFINITION)
+	{
+		PQExpBuffer query = createPQExpBuffer();
+
+		/*
+		 * XXX: For simplification, pg_create_logical_replication_slot() is
+		 * used. Is it sufficient?
+		 */
+		appendPQExpBuffer(query, "SELECT pg_catalog.pg_create_logical_replication_slot(");
+		appendStringLiteralAH(query, slotinfo->dobj.name, fout);
+		appendPQExpBuffer(query, ", ");
+		appendStringLiteralAH(query, slotinfo->plugin, fout);
+		appendPQExpBuffer(query, ", false, %s);",
+						  slotinfo->twophase ? "true" : "false");
+
+		ArchiveEntry(fout, slotinfo->dobj.catId, slotinfo->dobj.dumpId,
+					 ARCHIVE_OPTS(.tag = slotinfo->dobj.name,
+								  .description = "REPLICATION SLOT",
+								  .section = SECTION_POST_DATA,
+								  .createStmt = query->data));
+
+		destroyPQExpBuffer(query);
+	}
+}
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index bc8f2ec36d..ed1866d9ab 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -82,7 +82,8 @@ typedef enum
 	DO_PUBLICATION,
 	DO_PUBLICATION_REL,
 	DO_PUBLICATION_TABLE_IN_SCHEMA,
-	DO_SUBSCRIPTION
+	DO_SUBSCRIPTION,
+	DO_LOGICAL_REPLICATION_SLOT
 } DumpableObjectType;
 
 /*
@@ -667,6 +668,17 @@ typedef struct _SubscriptionInfo
 	char	   *subpasswordrequired;
 } SubscriptionInfo;
 
+/*
+ * The LogicalReplicationSlotInfo struct is used to represent replication
+ * slots.
+ */
+typedef struct _LogicalReplicationSlotInfo
+{
+	DumpableObject dobj;
+	char	   *plugin;
+	bool		twophase;
+} LogicalReplicationSlotInfo;
+
 /*
  *	common utility functions
  */
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index 523a19c155..ae65443228 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -93,6 +93,7 @@ enum dbObjectTypePriorities
 	PRIO_PUBLICATION_REL,
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,
 	PRIO_SUBSCRIPTION,
+	PRIO_LOGICAL_REPLICATION_SLOT,
 	PRIO_DEFAULT_ACL,			/* done in ACL pass */
 	PRIO_EVENT_TRIGGER,			/* must be next to last! */
 	PRIO_REFRESH_MATVIEW		/* must be last! */
@@ -146,10 +147,11 @@ static const int dbObjectTypePriority[] =
 	PRIO_PUBLICATION,			/* DO_PUBLICATION */
 	PRIO_PUBLICATION_REL,		/* DO_PUBLICATION_REL */
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,	/* DO_PUBLICATION_TABLE_IN_SCHEMA */
-	PRIO_SUBSCRIPTION			/* DO_SUBSCRIPTION */
+	PRIO_SUBSCRIPTION,			/* DO_SUBSCRIPTION */
+	PRIO_LOGICAL_REPLICATION_SLOT	/* DO_LOGICAL_REPLICATION_SLOT */
 };
 
-StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_SUBSCRIPTION + 1),
+StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_LOGICAL_REPLICATION_SLOT + 1),
 				 "array length mismatch");
 
 static DumpId preDataBoundId;
@@ -1542,6 +1544,11 @@ describeDumpableObject(DumpableObject *obj, char *buf, int bufsize)
 					 "SUBSCRIPTION (ID %d OID %u)",
 					 obj->dumpId, obj->catId.oid);
 			return;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			snprintf(buf, bufsize,
+					 "LOGICAL REPLICATION SLOT (ID %d NAME %s)",
+					 obj->dumpId, obj->name);
+			return;
 		case DO_PRE_DATA_BOUNDARY:
 			snprintf(buf, bufsize,
 					 "PRE-DATA BOUNDARY  (ID %d)",
diff --git a/src/bin/pg_upgrade/Makefile b/src/bin/pg_upgrade/Makefile
index 5834513add..815d1a7ca1 100644
--- a/src/bin/pg_upgrade/Makefile
+++ b/src/bin/pg_upgrade/Makefile
@@ -3,6 +3,9 @@
 PGFILEDESC = "pg_upgrade - an in-place binary upgrade utility"
 PGAPPICON = win32
 
+# required for 003_logical_replication_slots.pl
+EXTRA_INSTALL=contrib/test_decoding
+
 subdir = src/bin/pg_upgrade
 top_builddir = ../../..
 include $(top_builddir)/src/Makefile.global
diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index 64024e3b9e..e16ba70a54 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -30,6 +30,7 @@ static void check_for_jsonb_9_4_usage(ClusterInfo *cluster);
 static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
+static void check_for_parameter_settings(ClusterInfo *new_cluster);
 
 
 /*
@@ -89,6 +90,10 @@ check_and_dump_old_cluster(bool live_check)
 	/* Extract a list of databases and tables from the old cluster */
 	get_db_and_rel_infos(&old_cluster);
 
+	/* Additionally, extract a list of logical replication slots if required */
+	if (user_opts.include_logical_slots)
+		get_logical_slot_infos(&old_cluster);
+
 	init_tablespaces();
 
 	get_loadable_libraries();
@@ -189,6 +194,17 @@ check_new_cluster(void)
 {
 	get_db_and_rel_infos(&new_cluster);
 
+	/*
+	 * Do additional works if --include-logical-replication-slots is required.
+	 * These must be done before check_new_cluster_is_empty() because the
+	 * slot_arr attribute of the new_cluster will be checked in the function.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		get_logical_slot_infos(&new_cluster);
+		check_for_parameter_settings(&new_cluster);
+	}
+
 	check_new_cluster_is_empty();
 
 	check_loadable_libraries();
@@ -364,6 +380,22 @@ check_new_cluster_is_empty(void)
 						 rel_arr->rels[relnum].nspname,
 						 rel_arr->rels[relnum].relname);
 		}
+
+		/*
+		 * If --include-logical-replication-slots is required, check the
+		 * existence of slots.
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			LogicalSlotInfoArr *slot_arr = &new_cluster.dbarr.dbs[dbnum].slot_arr;
+
+			/* if nslots > 0, report just first entry and exit */
+			if (slot_arr->nslots)
+				pg_fatal("New cluster database \"%s\" is not empty: found logical replication slot \"%s\"",
+						 new_cluster.dbarr.dbs[dbnum].db_name,
+						 slot_arr->slots[0].slotname);
+		}
+
 	}
 }
 
@@ -1402,3 +1434,42 @@ check_for_user_defined_encoding_conversions(ClusterInfo *cluster)
 	else
 		check_ok();
 }
+
+/*
+ * Verify parameter settings for creating logical replication slots
+ */
+static void
+check_for_parameter_settings(ClusterInfo *new_cluster)
+{
+	PGresult   *res;
+	PGconn	   *conn = connectToServer(new_cluster, "template1");
+	int			max_replication_slots;
+	char	   *wal_level;
+
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(new_cluster->major_version) <= 1500)
+		return;
+
+	prep_status("Checking parameter settings for logical replication slots");
+
+	res = executeQueryOrDie(conn, "SHOW max_replication_slots;");
+	max_replication_slots = atoi(PQgetvalue(res, 0, 0));
+
+	if (max_replication_slots == 0)
+		pg_fatal("max_replication_slots must be greater than 0");
+
+	PQclear(res);
+
+	res = executeQueryOrDie(conn, "SHOW wal_level;");
+	wal_level = PQgetvalue(res, 0, 0);
+
+	if (strcmp(wal_level, "logical") != 0)
+		pg_fatal("wal_level must be \"logical\", but is set to \"%s\"",
+				 wal_level);
+
+	PQclear(res);
+
+	PQfinish(conn);
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/dump.c b/src/bin/pg_upgrade/dump.c
index 6c8c82dca8..e6b90864f5 100644
--- a/src/bin/pg_upgrade/dump.c
+++ b/src/bin/pg_upgrade/dump.c
@@ -59,6 +59,30 @@ generate_old_dump(void)
 						   log_opts.dumpdir,
 						   sql_file_name, escaped_connstr.data);
 
+		/*
+		 * Dump logical replication slots if needed.
+		 *
+		 * XXX We cannot dump replication slots at the same time as the schema
+		 * dump because we need to separate the timing of restoring
+		 * replication slots and other objects. Replication slots, in
+		 * particular, should not be restored before executing the pg_resetwal
+		 * command because it will remove WALs that are required by the slots.
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			char		slots_file_name[MAXPGPATH];
+
+			snprintf(slots_file_name, sizeof(slots_file_name),
+					 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+			parallel_exec_prog(log_file_name, NULL,
+							   "\"%s/pg_dump\" %s --logical-replication-slots-only "
+							   "--quote-all-identifiers --binary-upgrade %s "
+							   "--file=\"%s/%s\" %s",
+							   new_cluster.bindir, cluster_conn_opts(&old_cluster),
+							   log_opts.verbose ? "--verbose" : "",
+							   log_opts.dumpdir,
+							   slots_file_name, escaped_connstr.data);
+		}
 		termPQExpBuffer(&escaped_connstr);
 	}
 
diff --git a/src/bin/pg_upgrade/info.c b/src/bin/pg_upgrade/info.c
index a9988abfe1..0b5d5d870b 100644
--- a/src/bin/pg_upgrade/info.c
+++ b/src/bin/pg_upgrade/info.c
@@ -26,7 +26,7 @@ static void get_rel_infos(ClusterInfo *cluster, DbInfo *dbinfo);
 static void free_rel_infos(RelInfoArr *rel_arr);
 static void print_db_infos(DbInfoArr *db_arr);
 static void print_rel_infos(RelInfoArr *rel_arr);
-
+static void print_slot_infos(LogicalSlotInfoArr *slot_arr);
 
 /*
  * gen_db_file_maps()
@@ -600,6 +600,94 @@ get_rel_infos(ClusterInfo *cluster, DbInfo *dbinfo)
 	dbinfo->rel_arr.nrels = num_rels;
 }
 
+/*
+ * get_logical_slot_infos_per_db()
+ *
+ * gets the LogicalSlotInfos for all the logical replication slots of the database
+ * referred to by "dbinfo".
+ */
+static void
+get_logical_slot_infos_per_db(ClusterInfo *cluster, DbInfo *dbinfo)
+{
+	PGconn	   *conn = connectToServer(cluster,
+									   dbinfo->db_name);
+	PGresult   *res;
+	LogicalSlotInfo *slotinfos = NULL;
+
+	int			num_slots;
+
+	char		query[QUERY_ALLOC];
+
+	query[0] = '\0';			/* initialize query string to empty */
+
+	snprintf(query, sizeof(query),
+			 "SELECT slot_name, plugin, two_phase "
+			 "FROM pg_catalog.pg_replication_slots "
+			 "WHERE database = current_database() AND temporary = false "
+			 "AND wal_status IN ('reserved', 'extended');");
+
+	res = executeQueryOrDie(conn, "%s", query);
+
+	num_slots = PQntuples(res);
+
+	if (num_slots)
+	{
+		int			slotnum;
+		int			i_slotname;
+		int			i_plugin;
+		int			i_twophase;
+
+		slotinfos = (LogicalSlotInfo *) pg_malloc(sizeof(LogicalSlotInfo) * num_slots);
+
+		i_slotname = PQfnumber(res, "slot_name");
+		i_plugin = PQfnumber(res, "plugin");
+		i_twophase = PQfnumber(res, "two_phase");
+
+		for (slotnum = 0; slotnum < num_slots; slotnum++)
+		{
+			LogicalSlotInfo *curr = &slotinfos[slotnum];
+
+			curr->slotname = pg_strdup(PQgetvalue(res, slotnum, i_slotname));
+			curr->plugin = pg_strdup(PQgetvalue(res, slotnum, i_plugin));
+			curr->two_phase = (strcmp(PQgetvalue(res, slotnum, i_twophase), "t") == 0);
+		}
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	dbinfo->slot_arr.slots = slotinfos;
+	dbinfo->slot_arr.nslots = num_slots;
+}
+
+/*
+ * get_logical_slot_infos()
+ *
+ * Higher level routine to generate LogicalSlotInfoArr for all databases.
+ */
+void
+get_logical_slot_infos(ClusterInfo *cluster)
+{
+	int			dbnum;
+
+	if (cluster == &old_cluster)
+		pg_log(PG_VERBOSE, "\nsource databases:");
+	else
+		pg_log(PG_VERBOSE, "\ntarget databases:");
+
+	for (dbnum = 0; dbnum < cluster->dbarr.ndbs; dbnum++)
+	{
+		DbInfo	   *pDbInfo = &cluster->dbarr.dbs[dbnum];
+
+		get_logical_slot_infos_per_db(cluster, pDbInfo);
+
+		if (log_opts.verbose)
+		{
+			pg_log(PG_VERBOSE, "Database: \"%s\"", pDbInfo->db_name);
+			print_slot_infos(&pDbInfo->slot_arr);
+		}
+	}
+}
 
 static void
 free_db_and_rel_infos(DbInfoArr *db_arr)
@@ -610,6 +698,14 @@ free_db_and_rel_infos(DbInfoArr *db_arr)
 	{
 		free_rel_infos(&db_arr->dbs[dbnum].rel_arr);
 		pg_free(db_arr->dbs[dbnum].db_name);
+
+		/*
+		 * db_arr has an additional attribute, LogicalSlotInfoArr slot_arr,
+		 * but there is no need to free it. It has a valid member only when
+		 * the cluster had logical replication slots in the previous call.
+		 * However, in this case, a FATAL error is thrown, and we cannot reach
+		 * this point.
+		 */
 	}
 	pg_free(db_arr->dbs);
 	db_arr->dbs = NULL;
@@ -660,3 +756,15 @@ print_rel_infos(RelInfoArr *rel_arr)
 			   rel_arr->rels[relnum].reloid,
 			   rel_arr->rels[relnum].tablespace);
 }
+
+static void
+print_slot_infos(LogicalSlotInfoArr *slot_arr)
+{
+	int			slotnum;
+
+	for (slotnum = 0; slotnum < slot_arr->nslots; slotnum++)
+		pg_log(PG_VERBOSE, "slotname: \"%s\", plugin: \"%s\", two_phase: %d",
+			   slot_arr->slots[slotnum].slotname,
+			   slot_arr->slots[slotnum].plugin,
+			   slot_arr->slots[slotnum].two_phase);
+}
diff --git a/src/bin/pg_upgrade/meson.build b/src/bin/pg_upgrade/meson.build
index 12a97f84e2..228f29b688 100644
--- a/src/bin/pg_upgrade/meson.build
+++ b/src/bin/pg_upgrade/meson.build
@@ -42,6 +42,7 @@ tests += {
     'tests': [
       't/001_basic.pl',
       't/002_pg_upgrade.pl',
+      't/003_logical_replication_slots.pl',
     ],
     'test_kwargs': {'priority': 40}, # pg_upgrade tests are slow
   },
diff --git a/src/bin/pg_upgrade/option.c b/src/bin/pg_upgrade/option.c
index 640361009e..df66a5ffe6 100644
--- a/src/bin/pg_upgrade/option.c
+++ b/src/bin/pg_upgrade/option.c
@@ -57,6 +57,7 @@ parseCommandLine(int argc, char *argv[])
 		{"verbose", no_argument, NULL, 'v'},
 		{"clone", no_argument, NULL, 1},
 		{"copy", no_argument, NULL, 2},
+		{"include-logical-replication-slots", no_argument, NULL, 3},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -199,6 +200,10 @@ parseCommandLine(int argc, char *argv[])
 				user_opts.transfer_mode = TRANSFER_MODE_COPY;
 				break;
 
+			case 3:
+				user_opts.include_logical_slots = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						os_info.progname);
@@ -289,6 +294,8 @@ usage(void)
 	printf(_("  -V, --version                 display version information, then exit\n"));
 	printf(_("  --clone                       clone instead of copying files to new cluster\n"));
 	printf(_("  --copy                        copy files to new cluster (default)\n"));
+	printf(_("  --include-logical-replication-slots\n"
+			 "                                upgrade logical replication slots\n"));
 	printf(_("  -?, --help                    show this help, then exit\n"));
 	printf(_("\n"
 			 "Before running pg_upgrade you must:\n"
diff --git a/src/bin/pg_upgrade/pg_upgrade.c b/src/bin/pg_upgrade/pg_upgrade.c
index 4562dafcff..49e16234ed 100644
--- a/src/bin/pg_upgrade/pg_upgrade.c
+++ b/src/bin/pg_upgrade/pg_upgrade.c
@@ -59,6 +59,7 @@ static void copy_xact_xlog_xid(void);
 static void set_frozenxids(bool minmxid_only);
 static void make_outputdirs(char *pgdata);
 static void setup(char *argv0, bool *live_check);
+static void create_logical_replication_slots(void);
 
 ClusterInfo old_cluster,
 			new_cluster;
@@ -188,6 +189,19 @@ main(int argc, char **argv)
 			  new_cluster.pgdata);
 	check_ok();
 
+	/*
+	 * Create logical replication slots if requested.
+	 *
+	 * Note: This must be done after doing pg_resetwal command because the
+	 * command will remove required WALs.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		start_postmaster(&new_cluster, true);
+		create_logical_replication_slots();
+		stop_postmaster(false);
+	}
+
 	if (user_opts.do_sync)
 	{
 		prep_status("Sync data directory to disk");
@@ -860,3 +874,50 @@ set_frozenxids(bool minmxid_only)
 
 	check_ok();
 }
+
+/*
+ * create_logical_replication_slots()
+ *
+ * Similar to create_new_objects() but only restores logical replication slots.
+ */
+static void
+create_logical_replication_slots(void)
+{
+	int			dbnum;
+
+	prep_status_progress("Restoring logical replication slots in the new cluster");
+
+	for (dbnum = 0; dbnum < old_cluster.dbarr.ndbs; dbnum++)
+	{
+		char		slots_file_name[MAXPGPATH],
+					log_file_name[MAXPGPATH];
+		DbInfo	   *old_db = &old_cluster.dbarr.dbs[dbnum];
+
+		pg_log(PG_STATUS, "%s", old_db->db_name);
+
+		snprintf(slots_file_name, sizeof(slots_file_name),
+				 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+		snprintf(log_file_name, sizeof(log_file_name),
+				 DB_DUMP_LOG_FILE_MASK, old_db->db_oid);
+
+		parallel_exec_prog(log_file_name,
+						   NULL,
+						   "\"%s/psql\" %s --echo-queries --set ON_ERROR_STOP=on "
+						   "--no-psqlrc --dbname %s -f \"%s/%s\"",
+						   new_cluster.bindir,
+						   cluster_conn_opts(&new_cluster),
+						   old_db->db_name,
+						   log_opts.dumpdir,
+						   slots_file_name);
+	}
+
+	/* reap all children */
+	while (reap_child(true) == true)
+		;
+
+	end_progress_output();
+	check_ok();
+
+	/* update new_cluster info again */
+	get_logical_slot_infos(&new_cluster);
+}
diff --git a/src/bin/pg_upgrade/pg_upgrade.h b/src/bin/pg_upgrade/pg_upgrade.h
index 3eea0139c7..7adbb50807 100644
--- a/src/bin/pg_upgrade/pg_upgrade.h
+++ b/src/bin/pg_upgrade/pg_upgrade.h
@@ -29,6 +29,7 @@
 /* contains both global db information and CREATE DATABASE commands */
 #define GLOBALS_DUMP_FILE	"pg_upgrade_dump_globals.sql"
 #define DB_DUMP_FILE_MASK	"pg_upgrade_dump_%u.custom"
+#define DB_DUMP_LOGICAL_SLOTS_FILE_MASK	"pg_upgrade_dump_%u_logical_slots.sql"
 
 /*
  * Base directories that include all the files generated internally, from the
@@ -150,6 +151,22 @@ typedef struct
 	int			nrels;
 } RelInfoArr;
 
+/*
+ * Structure to store logical replication slot information
+ */
+typedef struct
+{
+	char	   *slotname;		/* slot name */
+	char	   *plugin;			/* plugin */
+	bool		two_phase;		/* Can the slot decode 2PC? */
+} LogicalSlotInfo;
+
+typedef struct
+{
+	LogicalSlotInfo *slots;
+	int			nslots;
+} LogicalSlotInfoArr;
+
 /*
  * The following structure represents a relation mapping.
  */
@@ -176,6 +193,7 @@ typedef struct
 	char		db_tablespace[MAXPGPATH];	/* database default tablespace
 											 * path */
 	RelInfoArr	rel_arr;		/* array of all user relinfos */
+	LogicalSlotInfoArr slot_arr;	/* array of all LogicalSlotInfo */
 } DbInfo;
 
 /*
@@ -304,6 +322,8 @@ typedef struct
 	transferMode transfer_mode; /* copy files or link them? */
 	int			jobs;			/* number of processes/threads to use */
 	char	   *socketdir;		/* directory to use for Unix sockets */
+	bool		include_logical_slots;	/* true -> dump and restore logical
+										 * replication slots */
 } UserOpts;
 
 typedef struct
@@ -400,6 +420,7 @@ FileNameMap *gen_db_file_maps(DbInfo *old_db,
 							  DbInfo *new_db, int *nmaps, const char *old_pgdata,
 							  const char *new_pgdata);
 void		get_db_and_rel_infos(ClusterInfo *cluster);
+void		get_logical_slot_infos(ClusterInfo *cluster);
 
 /* option.c */
 
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
new file mode 100644
index 0000000000..b0f8efec78
--- /dev/null
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -0,0 +1,114 @@
+# Copyright (c) 2023, PostgreSQL Global Development Group
+
+# Tests for upgrading replication slots
+
+use strict;
+use warnings;
+
+use File::Path qw(rmtree);
+
+use PostgreSQL::Test::Cluster;
+use PostgreSQL::Test::Utils;
+use Test::More;
+
+# Can be changed to test the other modes
+my $mode = $ENV{PG_TEST_PG_UPGRADE_MODE} || '--copy';
+
+# Initialize old node
+my $old_node = PostgreSQL::Test::Cluster->new('old_node');
+$old_node->init(allows_streaming => 'logical');
+$old_node->start;
+
+# Initialize new node
+my $new_node = PostgreSQL::Test::Cluster->new('new_node');
+$new_node->init(allows_streaming => 1);
+
+my $bindir = $new_node->config_data('--bindir');
+
+$old_node->stop;
+
+# Cause a failure at the start of pg_upgrade because wal_level is replica
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with wrong wal_level');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+# Preparations for the subsequent test. The case max_replication_slots is set
+# to 0 is prohibited.
+$new_node->append_conf('postgresql.conf', "wal_level = 'logical'");
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 0");
+
+# Cause a failure at the start of pg_upgrade because max_replication_slots is 0
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with wrong max_replication_slots');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+# Preparations for the subsequent test. max_replication_slots is set to
+# non-zero value to succeed the pg_upgrade
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 10");
+
+# Create a slot on old node, and generate WALs
+$old_node->start;
+$old_node->safe_psql(
+	'postgres', qq[
+	SELECT pg_create_logical_replication_slot('test_slot', 'test_decoding', false, true);
+	CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a;
+]);
+
+$old_node->safe_psql('postgres',
+	"SELECT count(*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)"
+);
+$old_node->stop;
+
+# Actual run, pg_upgrade_output.d is removed at the end
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots'
+	],
+	'run of pg_upgrade of old node');
+ok( !-d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+$new_node->start;
+my $result = $new_node->safe_psql('postgres',
+	"SELECT slot_name, two_phase FROM pg_replication_slots");
+is($result, qq(test_slot|t), 'check the slot exists on new node');
+
+done_testing();
diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list
index 260854747b..ba0a933767 100644
--- a/src/tools/pgindent/typedefs.list
+++ b/src/tools/pgindent/typedefs.list
@@ -1498,7 +1498,10 @@ LogicalRepStreamAbortData
 LogicalRepTupleData
 LogicalRepTyp
 LogicalRepWorker
+LogicalReplicationSlotInfo
 LogicalRewriteMappingData
+LogicalSlotInfo
+LogicalSlotInfoArr
 LogicalTape
 LogicalTapeSet
 LsnReadQueue
-- 
2.27.0

From 2f4486024dad55ab094f9ab2d1e5812e939f3218 Mon Sep 17 00:00:00 2001
From: Julien Rouhaud <julien.rouhaud@free.fr>
Date: Fri, 14 Apr 2023 13:49:09 +0800
Subject: [PATCH v16 2/4] Always persist to disk logical slots during a
 shutdown checkpoint.

It's entirely possible for a logical slot to have a confirmed_flush_lsn higher
than the last value saved on disk while not being marked as dirty.  It's
currently not a problem to lose that value during a clean shutdown / restart
cycle, but a later patch adding support for pg_upgrade of publications and
logical slots will rely on that value being properly persisted to disk.

Author: Julien Rouhaud
Reviewed-by: Wang Wei
Discussion: FIXME
---
 src/backend/access/transam/xlog.c |  2 +-
 src/backend/replication/slot.c    | 26 ++++++++++++++++----------
 src/include/replication/slot.h    |  2 +-
 3 files changed, 18 insertions(+), 12 deletions(-)

diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index b2430f617c..384245c157 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7011,7 +7011,7 @@ static void
 CheckPointGuts(XLogRecPtr checkPointRedo, int flags)
 {
 	CheckPointRelationMap();
-	CheckPointReplicationSlots();
+	CheckPointReplicationSlots(flags & CHECKPOINT_IS_SHUTDOWN);
 	CheckPointSnapBuild();
 	CheckPointLogicalRewriteHeap();
 	CheckPointReplicationOrigin();
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index 8021aaa0a8..526323e87b 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -109,7 +109,8 @@ static void ReplicationSlotDropPtr(ReplicationSlot *slot);
 /* internal persistency functions */
 static void RestoreSlotFromDisk(const char *name);
 static void CreateSlotOnDisk(ReplicationSlot *slot);
-static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel);
+static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+						   bool is_shutdown);
 
 /*
  * Report shared-memory space needed by ReplicationSlotsShmemInit.
@@ -783,7 +784,7 @@ ReplicationSlotSave(void)
 	Assert(MyReplicationSlot != NULL);
 
 	sprintf(path, "pg_replslot/%s", NameStr(MyReplicationSlot->data.name));
-	SaveSlotToPath(MyReplicationSlot, path, ERROR);
+	SaveSlotToPath(MyReplicationSlot, path, ERROR, false);
 }
 
 /*
@@ -1565,11 +1566,10 @@ restart:
 /*
  * Flush all replication slots to disk.
  *
- * This needn't actually be part of a checkpoint, but it's a convenient
- * location.
+ * is_shutdown is true in case of a shutdown checkpoint.
  */
 void
-CheckPointReplicationSlots(void)
+CheckPointReplicationSlots(bool is_shutdown)
 {
 	int			i;
 
@@ -1594,7 +1594,7 @@ CheckPointReplicationSlots(void)
 
 		/* save the slot to disk, locking is handled in SaveSlotToPath() */
 		sprintf(path, "pg_replslot/%s", NameStr(s->data.name));
-		SaveSlotToPath(s, path, LOG);
+		SaveSlotToPath(s, path, LOG, is_shutdown);
 	}
 	LWLockRelease(ReplicationSlotAllocationLock);
 }
@@ -1700,7 +1700,7 @@ CreateSlotOnDisk(ReplicationSlot *slot)
 
 	/* Write the actual state file. */
 	slot->dirty = true;			/* signal that we really need to write */
-	SaveSlotToPath(slot, tmppath, ERROR);
+	SaveSlotToPath(slot, tmppath, ERROR, false);
 
 	/* Rename the directory into place. */
 	if (rename(tmppath, path) != 0)
@@ -1726,7 +1726,8 @@ CreateSlotOnDisk(ReplicationSlot *slot)
  * Shared functionality between saving and creating a replication slot.
  */
 static void
-SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
+SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+			   bool is_shutdown)
 {
 	char		tmppath[MAXPGPATH];
 	char		path[MAXPGPATH];
@@ -1740,8 +1741,13 @@ SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
 	slot->just_dirtied = false;
 	SpinLockRelease(&slot->mutex);
 
-	/* and don't do anything if there's nothing to write */
-	if (!was_dirty)
+	/*
+	 * and don't do anything if there's nothing to write, unless it's this is
+	 * called for a logical slot during a shutdown checkpoint, as we want to
+	 * persist the confirmed_flush_lsn in that case, even if that's the only
+	 * modification.
+	 */
+	if (!was_dirty && (SlotIsPhysical(slot) || !is_shutdown))
 		return;
 
 	LWLockAcquire(&slot->io_in_progress_lock, LW_EXCLUSIVE);
diff --git a/src/include/replication/slot.h b/src/include/replication/slot.h
index a8a89dc784..7ca37c9f70 100644
--- a/src/include/replication/slot.h
+++ b/src/include/replication/slot.h
@@ -241,7 +241,7 @@ extern void ReplicationSlotNameForTablesync(Oid suboid, Oid relid, char *syncslo
 extern void ReplicationSlotDropAtPubNode(WalReceiverConn *wrconn, char *slotname, bool missing_ok);
 
 extern void StartupReplicationSlots(void);
-extern void CheckPointReplicationSlots(void);
+extern void CheckPointReplicationSlots(bool is_shutdown);
 
 extern void CheckSlotRequirements(void);
 extern void CheckSlotPermissions(void);
-- 
2.27.0

From 77aef40053ccd2e2d7451ac9d819e11006e9e84b Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Fri, 14 Apr 2023 08:59:03 +0000
Subject: [PATCH v16 3/4] pg_upgrade: Add check function for
 --include-logical-replication-slots option

Author: Hayato Kuroda
Reviewed-by: Wang Wei, Vignesh C
---
 src/bin/pg_upgrade/check.c                    | 71 ++++++++++++++++++-
 .../t/003_logical_replication_slots.pl        | 24 +++++++
 2 files changed, 94 insertions(+), 1 deletion(-)

diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index e16ba70a54..aaa348037f 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -9,8 +9,10 @@
 
 #include "postgres_fe.h"
 
+#include "access/xlog_internal.h"
 #include "catalog/pg_authid_d.h"
 #include "catalog/pg_collation.h"
+#include "catalog/pg_control.h"
 #include "fe_utils/string_utils.h"
 #include "mb/pg_wchar.h"
 #include "pg_upgrade.h"
@@ -31,7 +33,7 @@ static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
 static void check_for_parameter_settings(ClusterInfo *new_cluster);
-
+static void check_for_confirmed_flush_lsn(ClusterInfo *cluster);
 
 /*
  * fix_path_separator
@@ -108,6 +110,8 @@ check_and_dump_old_cluster(bool live_check)
 	check_for_composite_data_type_usage(&old_cluster);
 	check_for_reg_data_type_usage(&old_cluster);
 	check_for_isn_and_int8_passing_mismatch(&old_cluster);
+	if (user_opts.include_logical_slots)
+		check_for_confirmed_flush_lsn(&old_cluster);
 
 	/*
 	 * PG 16 increased the size of the 'aclitem' type, which breaks the
@@ -1473,3 +1477,68 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 
 	check_ok();
 }
+
+/*
+ * Verify that all logical replication slots consumed all WALs, except a
+ * CHECKPOINT_SHUTDOWN record.
+ */
+static void
+check_for_confirmed_flush_lsn(ClusterInfo *cluster)
+{
+	int			i,
+				ntups,
+				i_slotname;
+	bool		is_error = false;
+	PGresult   *res;
+	DbInfo	   *active_db = &cluster->dbarr.dbs[0];
+	PGconn	   *conn = connectToServer(cluster, active_db->db_name);
+
+	Assert(user_opts.include_logical_slots);
+
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(cluster->major_version) <= 1500)
+		return;
+
+	prep_status("Checking confirmed_flush_lsn for logical replication slots");
+
+	/*
+	 * Check that all logical replication slots have reached the current WAL
+	 * position, except for the CHECKPOINT_SHUTDOWN record. Even if all WALs
+	 * are consumed before shutting down the node, the checkpointer generates
+	 * a CHECKPOINT_SHUTDOWN record at shutdown, which cannot be consumed by
+	 * any slots. Therefore, we must allow for a difference between
+	 * pg_current_wal_insert_lsn() and confirmed_flush_lsn.
+	 */
+#define SHUTDOWN_RECORD_SIZE  (SizeOfXLogRecord + \
+							   SizeOfXLogRecordDataHeaderShort + \
+							   sizeof(CheckPoint))
+
+	res = executeQueryOrDie(conn,
+							"SELECT slot_name FROM pg_catalog.pg_replication_slots "
+							"WHERE (pg_catalog.pg_current_wal_insert_lsn() - confirmed_flush_lsn) > %d "
+							"AND temporary = false AND wal_status IN ('reserved', 'extended');",
+							(int) (SizeOfXLogLongPHD + SHUTDOWN_RECORD_SIZE));
+
+#undef SHUTDOWN_RECORD_SIZE
+
+	ntups = PQntuples(res);
+	i_slotname = PQfnumber(res, "slot_name");
+
+	for (i = 0; i < ntups; i++)
+	{
+		is_error = true;
+
+		pg_log(PG_WARNING,
+			   "\nWARNING: logical replication slot \"%s\" has not consumed WALs yet",
+			   PQgetvalue(res, i, i_slotname));
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	if (is_error)
+		pg_fatal("--include-logical-replication-slots requires that all "
+				 "logical replication slots consumed all the WALs");
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
index b0f8efec78..a2bb0b5e35 100644
--- a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -84,6 +84,30 @@ $old_node->safe_psql(
 	CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a;
 ]);
 
+$old_node->stop;
+
+# Cause a failure at the start of pg_upgrade because test_slot does not
+# finish consuming all the WALs
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with idle replication slots');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+$old_node->start;
 $old_node->safe_psql('postgres',
 	"SELECT count(*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)"
 );
-- 
2.27.0

From 919b2933bf6559a6cdfda835c52442fa0e23462b Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Mon, 24 Apr 2023 11:03:28 +0000
Subject: [PATCH v16 4/4] Change the method used to check logical replication
 slots during the live check

When a live check is requested, there is a possibility of additional changes
occurring, which may cause the current WAL position to exceed the confirmed_flush_lsn
of the slot. As a result, we check the confirmed_flush_lsn of each logical slot
instead. This is sufficient as all the WAL records will be sent during the publisher's
shutdown.

Author: Hayato Kuroda
Reviewed-by: Wang Wei, Vignesh C
---
 src/bin/pg_upgrade/check.c                    | 64 ++++++++++++++++++-
 .../t/003_logical_replication_slots.pl        | 64 ++++++++++++++++++-
 2 files changed, 125 insertions(+), 3 deletions(-)

diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index aaa348037f..1e2ffdb60d 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -34,6 +34,7 @@ static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
 static void check_for_parameter_settings(ClusterInfo *new_cluster);
 static void check_for_confirmed_flush_lsn(ClusterInfo *cluster);
+static void check_are_logical_slots_active(ClusterInfo *cluster);
 
 /*
  * fix_path_separator
@@ -111,7 +112,19 @@ check_and_dump_old_cluster(bool live_check)
 	check_for_reg_data_type_usage(&old_cluster);
 	check_for_isn_and_int8_passing_mismatch(&old_cluster);
 	if (user_opts.include_logical_slots)
-		check_for_confirmed_flush_lsn(&old_cluster);
+	{
+		/*
+		 * The method used to check logical replication slots is dependent on
+		 * the value of the live_check parameter. This change was implemented
+		 * because, during a live check, it is possible for additional changes
+		 * to occur at the old node, which could cause the current WAL position
+		 * to exceed the confirmed_flush_lsn of the slot.
+		 */
+		if (live_check)
+			check_are_logical_slots_active(&old_cluster);
+		else
+			check_for_confirmed_flush_lsn(&old_cluster);
+	}
 
 	/*
 	 * PG 16 increased the size of the 'aclitem' type, which breaks the
@@ -1478,6 +1491,55 @@ check_for_parameter_settings(ClusterInfo *new_cluster)
 	check_ok();
 }
 
+/*
+ * Verify that all logical replication slots are active
+ */
+static void
+check_are_logical_slots_active(ClusterInfo *cluster)
+{
+	int			i,
+				ntups,
+				i_slotname;
+	bool		is_error = false;
+	PGresult   *res;
+	DbInfo	   *active_db = &cluster->dbarr.dbs[0];
+	PGconn	   *conn = connectToServer(cluster, active_db->db_name);
+
+	Assert(user_opts.include_logical_slots);
+
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(cluster->major_version) <= 1500)
+		return;
+
+	prep_status("Checking the status of subscriptions for logical replication slots");
+
+	res = executeQueryOrDie(conn,
+							"SELECT slot_name FROM pg_catalog.pg_replication_slots "
+							"WHERE active IS FALSE "
+							"AND temporary = false AND wal_status IN ('reserved', 'extended');");
+
+	ntups = PQntuples(res);
+	i_slotname = PQfnumber(res, "slot_name");
+
+	for (i = 0; i < ntups; i++)
+	{
+		is_error = true;
+
+		pg_log(PG_WARNING,
+			   "\nWARNING: logical replication slot \"%s\" is not active",
+			   PQgetvalue(res, i, i_slotname));
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	if (is_error)
+		pg_fatal("--include-logical-replication-slots with --check requires that "
+				 "all logical replication slots are active");
+
+	check_ok();
+}
+
 /*
  * Verify that all logical replication slots consumed all WALs, except a
  * CHECKPOINT_SHUTDOWN record.
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
index a2bb0b5e35..a46c838b60 100644
--- a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -19,6 +19,10 @@ my $old_node = PostgreSQL::Test::Cluster->new('old_node');
 $old_node->init(allows_streaming => 'logical');
 $old_node->start;
 
+# Initialize subscriber, which will be used only for --check
+my $subscriber = PostgreSQL::Test::Cluster->new('subscriber');
+$subscriber->init(allows_streaming => 'logical');
+
 # Initialize new node
 my $new_node = PostgreSQL::Test::Cluster->new('new_node');
 $new_node->init(allows_streaming => 1);
@@ -76,12 +80,68 @@ rmtree($new_node->data_dir . "/pg_upgrade_output.d");
 # non-zero value to succeed the pg_upgrade
 $new_node->append_conf('postgresql.conf', "max_replication_slots = 10");
 
-# Create a slot on old node, and generate WALs
+# Setup logical replication
 $old_node->start;
+$old_node->safe_psql('postgres',
+	"CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a");
+$old_node->safe_psql('postgres', "CREATE PUBLICATION pub FOR ALL TABLES");
+my $old_connstr = $old_node->connstr . ' dbname=postgres';
+
+$subscriber->start;
+$subscriber->safe_psql('postgres', "CREATE TABLE tbl (a int)");
+$subscriber->safe_psql('postgres',
+	"CREATE SUBSCRIPTION sub CONNECTION '$old_connstr' PUBLICATION pub WITH (copy_data = true)"
+);
+
+# Wait for initial table sync to finish
+$subscriber->wait_for_subscription_sync($old_node, 'sub');
+
+my $result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(10), 'check initial rows on subscriber');
+
+# Start a background session and open a transaction (not committed yet)
+my $bsession = $old_node->background_psql('postgres');
+$bsession->query_safe(
+	q{
+BEGIN;
+INSERT INTO tbl VALUES (generate_series(11, 20))
+});
+
+$result = $old_node->safe_psql('postgres',
+	"SELECT count(*) FROM pg_replication_slots WHERE pg_current_wal_insert_lsn() > confirmed_flush_lsn"
+);
+is($result, qq(1),
+	'check the current WAL position exceeds confirmed_flush_lsn');
+
+# Run pg_upgrade --check. In the command the status of each logical slots will
+# be checked and then this will be succeeded.
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+		'--check'
+	],
+	'run of pg_upgrade of old node');
+ok( !-d $old_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+# Cleanup
+$bsession->query_safe("ABORT");
+$bsession->quit;
+$subscriber->safe_psql('postgres', "DROP SUBSCRIPTION sub");
+
+# Create a slot on old node, and generate WALs
 $old_node->safe_psql(
 	'postgres', qq[
 	SELECT pg_create_logical_replication_slot('test_slot', 'test_decoding', false, true);
-	CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a;
+	INSERT INTO tbl VALUES (generate_series(11, 20));
 ]);
 
 $old_node->stop;
-- 
2.27.0

From 83846edcc79c6f8549b9f35319ee67b52eefc43d Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Tue, 4 Apr 2023 05:49:34 +0000
Subject: [PATCH v17 1/4] pg_upgrade: Add --include-logical-replication-slots
 option

This commit introduces a new pg_upgrade option called "--include-logical-replication-slots".
This allows nodes with logical replication slots to be upgraded. The commit can
be divided into two parts: one for pg_dump and another for pg_upgrade.

For pg_dump this commit includes a new option called "--logical-replication-slots-only".
This option can be used to dump logical replication slots. When this option is
specified, the slot_name, plugin, and two_phase parameters are extracted from
pg_replication_slots. An SQL file that executes pg_create_logical_replication_slot()
with the extracted parameters is generated.

For pg_upgrade, when '--include-logical-replication-slots' is specified, it executes
pg_dump with the new "--logical-replication-slots-only" option and restores the slots
using the pg_create_logical_replication_slots() statements that the dump
generated (see above). Note that we cannot dump replication slots at the same time
as the schema dump because we need to separate the timing of restoring replication
slots and other objects. Replication slots, in  particular, should not be restored
before executing the pg_resetwal command because it will remove WALs that are
required by the slots.

The significant advantage of this commit is that it makes it easy to continue
logical replication even after upgrading the publisher node. Previously, pg_upgrade
allowed copying publications to a new node. With this new commit, adjusting the
connection string to the new publisher will cause the apply worker on the subscriber
to connect to the new publisher automatically. This enables seamless continuation
of logical replication, even after an upgrade.

Author: Hayato Kuroda
Reviewed-by: Peter Smith, Julien Rouhaud, Vignesh C, Wang Wei
---
 doc/src/sgml/ref/pgupgrade.sgml               |  11 ++
 src/bin/pg_dump/pg_backup.h                   |   1 +
 src/bin/pg_dump/pg_dump.c                     | 155 ++++++++++++++++++
 src/bin/pg_dump/pg_dump.h                     |  14 +-
 src/bin/pg_dump/pg_dump_sort.c                |  11 +-
 src/bin/pg_upgrade/Makefile                   |   3 +
 src/bin/pg_upgrade/check.c                    |  76 +++++++++
 src/bin/pg_upgrade/dump.c                     |  24 +++
 src/bin/pg_upgrade/info.c                     | 111 ++++++++++++-
 src/bin/pg_upgrade/meson.build                |   1 +
 src/bin/pg_upgrade/option.c                   |   7 +
 src/bin/pg_upgrade/pg_upgrade.c               |  61 +++++++
 src/bin/pg_upgrade/pg_upgrade.h               |  21 +++
 .../t/003_logical_replication_slots.pl        | 146 +++++++++++++++++
 src/tools/pgindent/typedefs.list              |   3 +
 15 files changed, 641 insertions(+), 4 deletions(-)
 create mode 100644 src/bin/pg_upgrade/t/003_logical_replication_slots.pl

diff --git a/doc/src/sgml/ref/pgupgrade.sgml b/doc/src/sgml/ref/pgupgrade.sgml
index 7816b4c685..94e90ff506 100644
--- a/doc/src/sgml/ref/pgupgrade.sgml
+++ b/doc/src/sgml/ref/pgupgrade.sgml
@@ -240,6 +240,17 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--include-logical-replication-slots</option></term>
+      <listitem>
+       <para>
+        Upgrade logical replication slots. Only permanent replication slots
+        are included. Note that pg_upgrade does not check the installation of
+        plugins.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
       <term><option>-?</option></term>
       <term><option>--help</option></term>
diff --git a/src/bin/pg_dump/pg_backup.h b/src/bin/pg_dump/pg_backup.h
index aba780ef4b..0a4e931f9b 100644
--- a/src/bin/pg_dump/pg_backup.h
+++ b/src/bin/pg_dump/pg_backup.h
@@ -187,6 +187,7 @@ typedef struct _dumpOptions
 	int			use_setsessauth;
 	int			enable_row_security;
 	int			load_via_partition_root;
+	int			logical_slots_only;
 
 	/* default, if no "inclusion" switches appear, is to dump everything */
 	bool		include_everything;
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index 5dab1ba9ea..12d4066d3b 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -328,6 +328,9 @@ static void setupDumpWorker(Archive *AH);
 static TableInfo *getRootTableInfo(const TableInfo *tbinfo);
 static bool forcePartitionRootLoad(const TableInfo *tbinfo);
 
+static void getLogicalReplicationSlots(Archive *fout);
+static void dumpLogicalReplicationSlot(Archive *fout,
+									   const LogicalReplicationSlotInfo *slotinfo);
 
 int
 main(int argc, char **argv)
@@ -431,6 +434,7 @@ main(int argc, char **argv)
 		{"table-and-children", required_argument, NULL, 12},
 		{"exclude-table-and-children", required_argument, NULL, 13},
 		{"exclude-table-data-and-children", required_argument, NULL, 14},
+		{"logical-replication-slots-only", no_argument, NULL, 15},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -657,6 +661,10 @@ main(int argc, char **argv)
 										  optarg);
 				break;
 
+			case 15:			/* dump only replication slot(s) */
+				dopt.logical_slots_only = true;
+				break;
+
 			default:
 				/* getopt_long already emitted a complaint */
 				pg_log_error_hint("Try \"%s --help\" for more information.", progname);
@@ -714,6 +722,21 @@ main(int argc, char **argv)
 	if (dopt.do_nothing && dopt.dump_inserts == 0)
 		pg_fatal("option --on-conflict-do-nothing requires option --inserts, --rows-per-insert, or --column-inserts");
 
+	if (dopt.logical_slots_only)
+	{
+		if (!dopt.binary_upgrade)
+			pg_fatal("options --logical-replication-slots-only requires option --binary-upgrade");
+
+		if (dopt.dataOnly)
+			pg_fatal("options --logical-replication-slots-only and -a/--data-only cannot be used together");
+
+		if (dopt.schemaOnly)
+			pg_fatal("options --logical-replication-slots-only and -s/--schema-only cannot be used together");
+
+		if (dopt.outputClean)
+			pg_fatal("options --logical-replication-slots-only and -c/--clean cannot be used together");
+	}
+
 	/* Identify archive format to emit */
 	archiveFormat = parseArchiveFormat(format, &archiveMode);
 
@@ -876,6 +899,16 @@ main(int argc, char **argv)
 			pg_fatal("no matching extensions were found");
 	}
 
+	/*
+	 * If dump logical-replication-slots-only was requested, dump only them
+	 * and skip everything else.
+	 */
+	if (dopt.logical_slots_only)
+	{
+		getLogicalReplicationSlots(fout);
+		goto dump;
+	}
+
 	/*
 	 * Dumping LOs is the default for dumps where an inclusion switch is not
 	 * used (an "include everything" dump).  -B can be used to exclude LOs
@@ -936,6 +969,8 @@ main(int argc, char **argv)
 	if (!dopt.no_security_labels)
 		collectSecLabels(fout);
 
+dump:
+
 	/* Lastly, create dummy objects to represent the section boundaries */
 	boundaryObjs = createBoundaryObjects();
 
@@ -1109,6 +1144,12 @@ help(const char *progname)
 			 "                               servers matching PATTERN\n"));
 	printf(_("  --inserts                    dump data as INSERT commands, rather than COPY\n"));
 	printf(_("  --load-via-partition-root    load partitions via the root table\n"));
+
+	/*
+	 * The option --logical-replication-slots-only is used only by pg_upgrade
+	 * and should not be called by users, which is why it is not exposed by the
+	 * help.
+	 */
 	printf(_("  --no-comments                do not dump comments\n"));
 	printf(_("  --no-publications            do not dump publications\n"));
 	printf(_("  --no-security-labels         do not dump security label assignments\n"));
@@ -10237,6 +10278,10 @@ dumpDumpableObject(Archive *fout, DumpableObject *dobj)
 		case DO_SUBSCRIPTION:
 			dumpSubscription(fout, (const SubscriptionInfo *) dobj);
 			break;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			dumpLogicalReplicationSlot(fout,
+									   (const LogicalReplicationSlotInfo *) dobj);
+			break;
 		case DO_PRE_DATA_BOUNDARY:
 		case DO_POST_DATA_BOUNDARY:
 			/* never dumped, nothing to do */
@@ -18218,6 +18263,7 @@ addBoundaryDependencies(DumpableObject **dobjs, int numObjs,
 			case DO_PUBLICATION_REL:
 			case DO_PUBLICATION_TABLE_IN_SCHEMA:
 			case DO_SUBSCRIPTION:
+			case DO_LOGICAL_REPLICATION_SLOT:
 				/* Post-data objects: must come after the post-data boundary */
 				addObjectDependency(dobj, postDataBound->dumpId);
 				break;
@@ -18479,3 +18525,112 @@ appendReloptionsArrayAH(PQExpBuffer buffer, const char *reloptions,
 	if (!res)
 		pg_log_warning("could not parse %s array", "reloptions");
 }
+
+/*
+ * getLogicalReplicationSlots
+ *	  get information about replication slots
+ */
+static void
+getLogicalReplicationSlots(Archive *fout)
+{
+	PGresult   *res;
+	LogicalReplicationSlotInfo *slotinfo;
+	PQExpBuffer query;
+
+	int			i_slotname;
+	int			i_plugin;
+	int			i_twophase;
+	int			i,
+				ntups;
+
+	/* Check whether we should dump or not */
+	if (fout->remoteVersion < 170000)
+		return;
+
+	Assert(fout->dopt->logical_slots_only);
+
+	query = createPQExpBuffer();
+
+	resetPQExpBuffer(query);
+
+	/*
+	 * Get replication slots.
+	 *
+	 * XXX: Which information must be extracted from old node? Currently three
+	 * attributes are extracted because they are used by
+	 * pg_create_logical_replication_slot().
+	 */
+	appendPQExpBufferStr(query,
+						 "SELECT slot_name, plugin, two_phase "
+						 "FROM pg_catalog.pg_replication_slots "
+						 "WHERE database = current_database() AND temporary = false "
+						 "AND wal_status IN ('reserved', 'extended');");
+
+	res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
+
+	ntups = PQntuples(res);
+
+	i_slotname = PQfnumber(res, "slot_name");
+	i_plugin = PQfnumber(res, "plugin");
+	i_twophase = PQfnumber(res, "two_phase");
+
+	slotinfo = pg_malloc(ntups * sizeof(LogicalReplicationSlotInfo));
+
+	for (i = 0; i < ntups; i++)
+	{
+		slotinfo[i].dobj.objType = DO_LOGICAL_REPLICATION_SLOT;
+
+		slotinfo[i].dobj.catId.tableoid = InvalidOid;
+		slotinfo[i].dobj.catId.oid = InvalidOid;
+		AssignDumpId(&slotinfo[i].dobj);
+
+		slotinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_slotname));
+
+		slotinfo[i].plugin = pg_strdup(PQgetvalue(res, i, i_plugin));
+		slotinfo[i].twophase = (strcmp(PQgetvalue(res, i, i_twophase), "t") == 0);
+
+		/*
+		 * Note: Currently we do not have any options to include/exclude slots
+		 * in dumping, so all the slots must be selected.
+		 */
+		slotinfo[i].dobj.dump = DUMP_COMPONENT_DEFINITION;
+	}
+	PQclear(res);
+
+	destroyPQExpBuffer(query);
+}
+
+/*
+ * dumpLogicalReplicationSlot
+ *	  dump creation functions for the given logical replication slots
+ */
+static void
+dumpLogicalReplicationSlot(Archive *fout,
+						   const LogicalReplicationSlotInfo *slotinfo)
+{
+	Assert(fout->dopt->logical_slots_only);
+
+	if (slotinfo->dobj.dump & DUMP_COMPONENT_DEFINITION)
+	{
+		PQExpBuffer query = createPQExpBuffer();
+
+		/*
+		 * XXX: For simplification, pg_create_logical_replication_slot() is
+		 * used. Is it sufficient?
+		 */
+		appendPQExpBuffer(query, "SELECT pg_catalog.pg_create_logical_replication_slot(");
+		appendStringLiteralAH(query, slotinfo->dobj.name, fout);
+		appendPQExpBuffer(query, ", ");
+		appendStringLiteralAH(query, slotinfo->plugin, fout);
+		appendPQExpBuffer(query, ", false, %s);",
+						  slotinfo->twophase ? "true" : "false");
+
+		ArchiveEntry(fout, slotinfo->dobj.catId, slotinfo->dobj.dumpId,
+					 ARCHIVE_OPTS(.tag = slotinfo->dobj.name,
+								  .description = "REPLICATION SLOT",
+								  .section = SECTION_POST_DATA,
+								  .createStmt = query->data));
+
+		destroyPQExpBuffer(query);
+	}
+}
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index bc8f2ec36d..ed1866d9ab 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -82,7 +82,8 @@ typedef enum
 	DO_PUBLICATION,
 	DO_PUBLICATION_REL,
 	DO_PUBLICATION_TABLE_IN_SCHEMA,
-	DO_SUBSCRIPTION
+	DO_SUBSCRIPTION,
+	DO_LOGICAL_REPLICATION_SLOT
 } DumpableObjectType;
 
 /*
@@ -667,6 +668,17 @@ typedef struct _SubscriptionInfo
 	char	   *subpasswordrequired;
 } SubscriptionInfo;
 
+/*
+ * The LogicalReplicationSlotInfo struct is used to represent replication
+ * slots.
+ */
+typedef struct _LogicalReplicationSlotInfo
+{
+	DumpableObject dobj;
+	char	   *plugin;
+	bool		twophase;
+} LogicalReplicationSlotInfo;
+
 /*
  *	common utility functions
  */
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index 523a19c155..ae65443228 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -93,6 +93,7 @@ enum dbObjectTypePriorities
 	PRIO_PUBLICATION_REL,
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,
 	PRIO_SUBSCRIPTION,
+	PRIO_LOGICAL_REPLICATION_SLOT,
 	PRIO_DEFAULT_ACL,			/* done in ACL pass */
 	PRIO_EVENT_TRIGGER,			/* must be next to last! */
 	PRIO_REFRESH_MATVIEW		/* must be last! */
@@ -146,10 +147,11 @@ static const int dbObjectTypePriority[] =
 	PRIO_PUBLICATION,			/* DO_PUBLICATION */
 	PRIO_PUBLICATION_REL,		/* DO_PUBLICATION_REL */
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,	/* DO_PUBLICATION_TABLE_IN_SCHEMA */
-	PRIO_SUBSCRIPTION			/* DO_SUBSCRIPTION */
+	PRIO_SUBSCRIPTION,			/* DO_SUBSCRIPTION */
+	PRIO_LOGICAL_REPLICATION_SLOT	/* DO_LOGICAL_REPLICATION_SLOT */
 };
 
-StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_SUBSCRIPTION + 1),
+StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_LOGICAL_REPLICATION_SLOT + 1),
 				 "array length mismatch");
 
 static DumpId preDataBoundId;
@@ -1542,6 +1544,11 @@ describeDumpableObject(DumpableObject *obj, char *buf, int bufsize)
 					 "SUBSCRIPTION (ID %d OID %u)",
 					 obj->dumpId, obj->catId.oid);
 			return;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			snprintf(buf, bufsize,
+					 "LOGICAL REPLICATION SLOT (ID %d NAME %s)",
+					 obj->dumpId, obj->name);
+			return;
 		case DO_PRE_DATA_BOUNDARY:
 			snprintf(buf, bufsize,
 					 "PRE-DATA BOUNDARY  (ID %d)",
diff --git a/src/bin/pg_upgrade/Makefile b/src/bin/pg_upgrade/Makefile
index 5834513add..815d1a7ca1 100644
--- a/src/bin/pg_upgrade/Makefile
+++ b/src/bin/pg_upgrade/Makefile
@@ -3,6 +3,9 @@
 PGFILEDESC = "pg_upgrade - an in-place binary upgrade utility"
 PGAPPICON = win32
 
+# required for 003_logical_replication_slots.pl
+EXTRA_INSTALL=contrib/test_decoding
+
 subdir = src/bin/pg_upgrade
 top_builddir = ../../..
 include $(top_builddir)/src/Makefile.global
diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index 64024e3b9e..4b54f5567c 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -30,7 +30,9 @@ static void check_for_jsonb_9_4_usage(ClusterInfo *cluster);
 static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
+static void check_for_logical_replication_slots(ClusterInfo *new_cluster);
 
+static int num_slots_on_old_cluster;
 
 /*
  * fix_path_separator
@@ -89,6 +91,10 @@ check_and_dump_old_cluster(bool live_check)
 	/* Extract a list of databases and tables from the old cluster */
 	get_db_and_rel_infos(&old_cluster);
 
+	/* Additionally, extract a list of logical replication slots if required */
+	if (user_opts.include_logical_slots)
+		num_slots_on_old_cluster = get_logical_slot_infos(&old_cluster);
+
 	init_tablespaces();
 
 	get_loadable_libraries();
@@ -189,6 +195,17 @@ check_new_cluster(void)
 {
 	get_db_and_rel_infos(&new_cluster);
 
+	/*
+	 * Do additional work if --include-logical-replication-slots was specified.
+	 * This must be done before check_new_cluster_is_empty() because the
+	 * slot_arr attribute of the new_cluster will be checked in that function.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		(void) get_logical_slot_infos(&new_cluster);
+		check_for_logical_replication_slots(&new_cluster);
+	}
+
 	check_new_cluster_is_empty();
 
 	check_loadable_libraries();
@@ -364,6 +381,22 @@ check_new_cluster_is_empty(void)
 						 rel_arr->rels[relnum].nspname,
 						 rel_arr->rels[relnum].relname);
 		}
+
+		/*
+		 * If --include-logical-replication-slots is required, check the
+		 * existence of slots.
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			DbInfo *pDbInfo = &new_cluster.dbarr.dbs[dbnum];
+			LogicalSlotInfoArr *slot_arr = &pDbInfo->slot_arr;
+
+			/* if nslots > 0, report just first entry and exit */
+			if (slot_arr->nslots)
+				pg_fatal("New cluster database \"%s\" is not empty: found logical replication slot \"%s\"",
+						 pDbInfo->db_name,
+						 slot_arr->slots[0].slotname);
+		}
 	}
 }
 
@@ -1402,3 +1435,46 @@ check_for_user_defined_encoding_conversions(ClusterInfo *cluster)
 	else
 		check_ok();
 }
+
+/*
+ * Verify the parameter settings necessary for creating logical replication
+ * slots.
+ */
+static void
+check_for_logical_replication_slots(ClusterInfo *new_cluster)
+{
+	PGresult   *res;
+	PGconn	   *conn = connectToServer(new_cluster, "template1");
+	int			max_replication_slots;
+	char	   *wal_level;
+
+	/* --include-logical-replication-slots can be used since PG17. */
+	if (GET_MAJOR_VERSION(new_cluster->major_version) <= 1600)
+		return;
+
+	prep_status("Checking parameter settings for logical replication slots");
+
+	res = executeQueryOrDie(conn, "SHOW max_replication_slots;");
+	max_replication_slots = atoi(PQgetvalue(res, 0, 0));
+
+	if (max_replication_slots == 0)
+		pg_fatal("max_replication_slots must be greater than 0");
+	else if (num_slots_on_old_cluster > max_replication_slots)
+		pg_fatal("max_replication_slots must be greater than existing logical "
+				 "replication slots on old node.");
+
+	PQclear(res);
+
+	res = executeQueryOrDie(conn, "SHOW wal_level;");
+	wal_level = PQgetvalue(res, 0, 0);
+
+	if (strcmp(wal_level, "logical") != 0)
+		pg_fatal("wal_level must be \"logical\", but is set to \"%s\"",
+				 wal_level);
+
+	PQclear(res);
+
+	PQfinish(conn);
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/dump.c b/src/bin/pg_upgrade/dump.c
index 6c8c82dca8..e6b90864f5 100644
--- a/src/bin/pg_upgrade/dump.c
+++ b/src/bin/pg_upgrade/dump.c
@@ -59,6 +59,30 @@ generate_old_dump(void)
 						   log_opts.dumpdir,
 						   sql_file_name, escaped_connstr.data);
 
+		/*
+		 * Dump logical replication slots if needed.
+		 *
+		 * XXX We cannot dump replication slots at the same time as the schema
+		 * dump because we need to separate the timing of restoring
+		 * replication slots and other objects. Replication slots, in
+		 * particular, should not be restored before executing the pg_resetwal
+		 * command because it will remove WALs that are required by the slots.
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			char		slots_file_name[MAXPGPATH];
+
+			snprintf(slots_file_name, sizeof(slots_file_name),
+					 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+			parallel_exec_prog(log_file_name, NULL,
+							   "\"%s/pg_dump\" %s --logical-replication-slots-only "
+							   "--quote-all-identifiers --binary-upgrade %s "
+							   "--file=\"%s/%s\" %s",
+							   new_cluster.bindir, cluster_conn_opts(&old_cluster),
+							   log_opts.verbose ? "--verbose" : "",
+							   log_opts.dumpdir,
+							   slots_file_name, escaped_connstr.data);
+		}
 		termPQExpBuffer(&escaped_connstr);
 	}
 
diff --git a/src/bin/pg_upgrade/info.c b/src/bin/pg_upgrade/info.c
index a9988abfe1..8bc0ad2e10 100644
--- a/src/bin/pg_upgrade/info.c
+++ b/src/bin/pg_upgrade/info.c
@@ -26,6 +26,7 @@ static void get_rel_infos(ClusterInfo *cluster, DbInfo *dbinfo);
 static void free_rel_infos(RelInfoArr *rel_arr);
 static void print_db_infos(DbInfoArr *db_arr);
 static void print_rel_infos(RelInfoArr *rel_arr);
+static void print_slot_infos(LogicalSlotInfoArr *slot_arr);
 
 
 /*
@@ -394,7 +395,7 @@ get_db_infos(ClusterInfo *cluster)
 	i_spclocation = PQfnumber(res, "spclocation");
 
 	ntups = PQntuples(res);
-	dbinfos = (DbInfo *) pg_malloc(sizeof(DbInfo) * ntups);
+	dbinfos = (DbInfo *) pg_malloc0(sizeof(DbInfo) * ntups);
 
 	for (tupnum = 0; tupnum < ntups; tupnum++)
 	{
@@ -600,6 +601,96 @@ get_rel_infos(ClusterInfo *cluster, DbInfo *dbinfo)
 	dbinfo->rel_arr.nrels = num_rels;
 }
 
+/*
+ * get_logical_slot_infos_per_db()
+ *
+ * gets the LogicalSlotInfos for all the logical replication slots of the database
+ * referred to by "dbinfo".
+ */
+static void
+get_logical_slot_infos_per_db(ClusterInfo *cluster, DbInfo *dbinfo)
+{
+	PGconn	   *conn = connectToServer(cluster,
+									   dbinfo->db_name);
+	PGresult   *res;
+	LogicalSlotInfo *slotinfos = NULL;
+
+	int			num_slots;
+
+	char		query[QUERY_ALLOC];
+
+	snprintf(query, sizeof(query),
+			 "SELECT slot_name, plugin, two_phase "
+			 "FROM pg_catalog.pg_replication_slots "
+			 "WHERE database = current_database() AND temporary = false "
+			 "AND wal_status IN ('reserved', 'extended');");
+
+	res = executeQueryOrDie(conn, "%s", query);
+
+	num_slots = PQntuples(res);
+
+	if (num_slots)
+	{
+		int			slotnum;
+		int			i_slotname;
+		int			i_plugin;
+		int			i_twophase;
+
+		slotinfos = (LogicalSlotInfo *) pg_malloc(sizeof(LogicalSlotInfo) * num_slots);
+
+		i_slotname = PQfnumber(res, "slot_name");
+		i_plugin = PQfnumber(res, "plugin");
+		i_twophase = PQfnumber(res, "two_phase");
+
+		for (slotnum = 0; slotnum < num_slots; slotnum++)
+		{
+			LogicalSlotInfo *curr = &slotinfos[slotnum];
+
+			curr->slotname = pg_strdup(PQgetvalue(res, slotnum, i_slotname));
+			curr->plugin = pg_strdup(PQgetvalue(res, slotnum, i_plugin));
+			curr->two_phase = (strcmp(PQgetvalue(res, slotnum, i_twophase), "t") == 0);
+		}
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	dbinfo->slot_arr.slots = slotinfos;
+	dbinfo->slot_arr.nslots = num_slots;
+}
+
+/*
+ * get_logical_slot_infos()
+ *
+ * Higher level routine to generate LogicalSlotInfoArr for all databases.
+ */
+int
+get_logical_slot_infos(ClusterInfo *cluster)
+{
+	int			dbnum;
+	int			slotnum = 0;
+
+	if (cluster == &old_cluster)
+		pg_log(PG_VERBOSE, "\nsource databases:");
+	else
+		pg_log(PG_VERBOSE, "\ntarget databases:");
+
+	for (dbnum = 0; dbnum < cluster->dbarr.ndbs; dbnum++)
+	{
+		DbInfo	   *pDbInfo = &cluster->dbarr.dbs[dbnum];
+
+		get_logical_slot_infos_per_db(cluster, pDbInfo);
+		slotnum += pDbInfo->slot_arr.nslots;
+
+		if (log_opts.verbose)
+		{
+			pg_log(PG_VERBOSE, "Database: \"%s\"", pDbInfo->db_name);
+			print_slot_infos(&pDbInfo->slot_arr);
+		}
+	}
+
+	return slotnum;
+}
 
 static void
 free_db_and_rel_infos(DbInfoArr *db_arr)
@@ -610,6 +701,12 @@ free_db_and_rel_infos(DbInfoArr *db_arr)
 	{
 		free_rel_infos(&db_arr->dbs[dbnum].rel_arr);
 		pg_free(db_arr->dbs[dbnum].db_name);
+
+		/*
+		 * Logical replication slots must not exist on the new cluster before
+		 * doing create_logical_replication_slots().
+		 */
+		Assert(db_arr->dbs[dbnum].slot_arr.slots == NULL);
 	}
 	pg_free(db_arr->dbs);
 	db_arr->dbs = NULL;
@@ -660,3 +757,15 @@ print_rel_infos(RelInfoArr *rel_arr)
 			   rel_arr->rels[relnum].reloid,
 			   rel_arr->rels[relnum].tablespace);
 }
+
+static void
+print_slot_infos(LogicalSlotInfoArr *slot_arr)
+{
+	int			slotnum;
+
+	for (slotnum = 0; slotnum < slot_arr->nslots; slotnum++)
+		pg_log(PG_VERBOSE, "slotname: \"%s\", plugin: \"%s\", two_phase: %d",
+			   slot_arr->slots[slotnum].slotname,
+			   slot_arr->slots[slotnum].plugin,
+			   slot_arr->slots[slotnum].two_phase);
+}
diff --git a/src/bin/pg_upgrade/meson.build b/src/bin/pg_upgrade/meson.build
index 12a97f84e2..228f29b688 100644
--- a/src/bin/pg_upgrade/meson.build
+++ b/src/bin/pg_upgrade/meson.build
@@ -42,6 +42,7 @@ tests += {
     'tests': [
       't/001_basic.pl',
       't/002_pg_upgrade.pl',
+      't/003_logical_replication_slots.pl',
     ],
     'test_kwargs': {'priority': 40}, # pg_upgrade tests are slow
   },
diff --git a/src/bin/pg_upgrade/option.c b/src/bin/pg_upgrade/option.c
index 640361009e..df66a5ffe6 100644
--- a/src/bin/pg_upgrade/option.c
+++ b/src/bin/pg_upgrade/option.c
@@ -57,6 +57,7 @@ parseCommandLine(int argc, char *argv[])
 		{"verbose", no_argument, NULL, 'v'},
 		{"clone", no_argument, NULL, 1},
 		{"copy", no_argument, NULL, 2},
+		{"include-logical-replication-slots", no_argument, NULL, 3},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -199,6 +200,10 @@ parseCommandLine(int argc, char *argv[])
 				user_opts.transfer_mode = TRANSFER_MODE_COPY;
 				break;
 
+			case 3:
+				user_opts.include_logical_slots = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						os_info.progname);
@@ -289,6 +294,8 @@ usage(void)
 	printf(_("  -V, --version                 display version information, then exit\n"));
 	printf(_("  --clone                       clone instead of copying files to new cluster\n"));
 	printf(_("  --copy                        copy files to new cluster (default)\n"));
+	printf(_("  --include-logical-replication-slots\n"
+			 "                                upgrade logical replication slots\n"));
 	printf(_("  -?, --help                    show this help, then exit\n"));
 	printf(_("\n"
 			 "Before running pg_upgrade you must:\n"
diff --git a/src/bin/pg_upgrade/pg_upgrade.c b/src/bin/pg_upgrade/pg_upgrade.c
index 4562dafcff..6dd3832422 100644
--- a/src/bin/pg_upgrade/pg_upgrade.c
+++ b/src/bin/pg_upgrade/pg_upgrade.c
@@ -59,6 +59,7 @@ static void copy_xact_xlog_xid(void);
 static void set_frozenxids(bool minmxid_only);
 static void make_outputdirs(char *pgdata);
 static void setup(char *argv0, bool *live_check);
+static void create_logical_replication_slots(void);
 
 ClusterInfo old_cluster,
 			new_cluster;
@@ -188,6 +189,19 @@ main(int argc, char **argv)
 			  new_cluster.pgdata);
 	check_ok();
 
+	/*
+	 * Create logical replication slots if requested.
+	 *
+	 * Note: This must be done after doing pg_resetwal command because
+	 * pg_resetwal would remove required WALs.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		start_postmaster(&new_cluster, true);
+		create_logical_replication_slots();
+		stop_postmaster(false);
+	}
+
 	if (user_opts.do_sync)
 	{
 		prep_status("Sync data directory to disk");
@@ -860,3 +874,50 @@ set_frozenxids(bool minmxid_only)
 
 	check_ok();
 }
+
+/*
+ * create_logical_replication_slots()
+ *
+ * Similar to create_new_objects() but only restores logical replication slots.
+ */
+static void
+create_logical_replication_slots(void)
+{
+	int			dbnum;
+
+	prep_status_progress("Restoring logical replication slots in the new cluster");
+
+	for (dbnum = 0; dbnum < old_cluster.dbarr.ndbs; dbnum++)
+	{
+		char		slots_file_name[MAXPGPATH],
+					log_file_name[MAXPGPATH];
+		DbInfo	   *old_db = &old_cluster.dbarr.dbs[dbnum];
+
+		pg_log(PG_STATUS, "%s", old_db->db_name);
+
+		snprintf(slots_file_name, sizeof(slots_file_name),
+				 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+		snprintf(log_file_name, sizeof(log_file_name),
+				 DB_DUMP_LOG_FILE_MASK, old_db->db_oid);
+
+		parallel_exec_prog(log_file_name,
+						   NULL,
+						   "\"%s/psql\" %s --echo-queries --set ON_ERROR_STOP=on "
+						   "--no-psqlrc --dbname %s -f \"%s/%s\"",
+						   new_cluster.bindir,
+						   cluster_conn_opts(&new_cluster),
+						   old_db->db_name,
+						   log_opts.dumpdir,
+						   slots_file_name);
+	}
+
+	/* reap all children */
+	while (reap_child(true) == true)
+		;
+
+	end_progress_output();
+	check_ok();
+
+	/* update new_cluster info again */
+	get_logical_slot_infos(&new_cluster);
+}
diff --git a/src/bin/pg_upgrade/pg_upgrade.h b/src/bin/pg_upgrade/pg_upgrade.h
index 3eea0139c7..8034067492 100644
--- a/src/bin/pg_upgrade/pg_upgrade.h
+++ b/src/bin/pg_upgrade/pg_upgrade.h
@@ -29,6 +29,7 @@
 /* contains both global db information and CREATE DATABASE commands */
 #define GLOBALS_DUMP_FILE	"pg_upgrade_dump_globals.sql"
 #define DB_DUMP_FILE_MASK	"pg_upgrade_dump_%u.custom"
+#define DB_DUMP_LOGICAL_SLOTS_FILE_MASK	"pg_upgrade_dump_%u_logical_slots.sql"
 
 /*
  * Base directories that include all the files generated internally, from the
@@ -150,6 +151,22 @@ typedef struct
 	int			nrels;
 } RelInfoArr;
 
+/*
+ * Structure to store logical replication slot information
+ */
+typedef struct
+{
+	char	   *slotname;		/* slot name */
+	char	   *plugin;			/* plugin */
+	bool		two_phase;		/* Can the slot decode 2PC? */
+} LogicalSlotInfo;
+
+typedef struct
+{
+	int			nslots;
+	LogicalSlotInfo *slots;
+} LogicalSlotInfoArr;
+
 /*
  * The following structure represents a relation mapping.
  */
@@ -176,6 +193,7 @@ typedef struct
 	char		db_tablespace[MAXPGPATH];	/* database default tablespace
 											 * path */
 	RelInfoArr	rel_arr;		/* array of all user relinfos */
+	LogicalSlotInfoArr slot_arr;	/* array of all LogicalSlotInfo */
 } DbInfo;
 
 /*
@@ -304,6 +322,8 @@ typedef struct
 	transferMode transfer_mode; /* copy files or link them? */
 	int			jobs;			/* number of processes/threads to use */
 	char	   *socketdir;		/* directory to use for Unix sockets */
+	bool		include_logical_slots;	/* true -> dump and restore logical
+										 * replication slots */
 } UserOpts;
 
 typedef struct
@@ -400,6 +420,7 @@ FileNameMap *gen_db_file_maps(DbInfo *old_db,
 							  DbInfo *new_db, int *nmaps, const char *old_pgdata,
 							  const char *new_pgdata);
 void		get_db_and_rel_infos(ClusterInfo *cluster);
+int			get_logical_slot_infos(ClusterInfo *cluster);
 
 /* option.c */
 
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
new file mode 100644
index 0000000000..9ca266f6b2
--- /dev/null
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -0,0 +1,146 @@
+# Copyright (c) 2023, PostgreSQL Global Development Group
+
+# Tests for upgrading replication slots
+
+use strict;
+use warnings;
+
+use File::Path qw(rmtree);
+
+use PostgreSQL::Test::Cluster;
+use PostgreSQL::Test::Utils;
+use Test::More;
+
+# Can be changed to test the other modes
+my $mode = $ENV{PG_TEST_PG_UPGRADE_MODE} || '--copy';
+
+# Initialize old node
+my $old_node = PostgreSQL::Test::Cluster->new('old_node');
+$old_node->init(allows_streaming => 'logical');
+$old_node->start;
+
+# Initialize new node
+my $new_node = PostgreSQL::Test::Cluster->new('new_node');
+$new_node->init(allows_streaming => 1);
+
+my $bindir = $new_node->config_data('--bindir');
+
+$old_node->stop;
+
+# Cause a failure at the start of pg_upgrade because wal_level is replica
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with wrong wal_level');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+# Preparations for the subsequent test. The case max_replication_slots is set
+# to 0 is prohibited.
+$new_node->append_conf('postgresql.conf', "wal_level = 'logical'");
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 0");
+
+# Cause a failure at the start of pg_upgrade because max_replication_slots is 0
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with wrong max_replication_slots');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+# Preparations for the subsequent test. max_replication_slots is set to
+# non-zero value
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 1");
+
+# Create a slot on old node, and generate WALs
+$old_node->start;
+$old_node->safe_psql(
+	'postgres', qq[
+	SELECT pg_create_logical_replication_slot('test_slot', 'test_decoding', false, true);
+	SELECT pg_create_logical_replication_slot('to_be_dropped', 'test_decoding', false, true);
+	CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a;
+]);
+
+$old_node->stop;
+
+# Cause a failure at the start of pg_upgrade because max_replication_slots is
+# smaller than existing slots on old node
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with small max_replication_slots');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+# Preparations for the subsequent test. max_replication_slots is set to
+# appropriate value
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 10");
+
+# Remove an unnecessary slot and consume WALs
+$old_node->start;
+$old_node->safe_psql(
+	'postgres', qq[
+	SELECT pg_drop_replication_slot('to_be_dropped');
+	SELECT count(*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)
+]);
+$old_node->stop;
+
+# Actual run, pg_upgrade_output.d is removed at the end
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots'
+	],
+	'run of pg_upgrade of old node');
+ok( !-d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+$new_node->start;
+my $result = $new_node->safe_psql('postgres',
+	"SELECT slot_name, two_phase FROM pg_replication_slots");
+is($result, qq(test_slot|t), 'check the slot exists on new node');
+
+done_testing();
diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list
index e941fb6c82..3962322d9d 100644
--- a/src/tools/pgindent/typedefs.list
+++ b/src/tools/pgindent/typedefs.list
@@ -1498,7 +1498,10 @@ LogicalRepStreamAbortData
 LogicalRepTupleData
 LogicalRepTyp
 LogicalRepWorker
+LogicalReplicationSlotInfo
 LogicalRewriteMappingData
+LogicalSlotInfo
+LogicalSlotInfoArr
 LogicalTape
 LogicalTapeSet
 LsnReadQueue
-- 
2.27.0

From 3000080ab40a37913aa5d45f013377325bad04eb Mon Sep 17 00:00:00 2001
From: Julien Rouhaud <julien.rouhaud@free.fr>
Date: Fri, 14 Apr 2023 13:49:09 +0800
Subject: [PATCH v17 2/4] Always persist to disk logical slots during a
 shutdown checkpoint.

It's entirely possible for a logical slot to have a confirmed_flush_lsn higher
than the last value saved on disk while not being marked as dirty.  It's
currently not a problem to lose that value during a clean shutdown / restart
cycle, but a later patch adding support for pg_upgrade of publications and
logical slots will rely on that value being properly persisted to disk.

Author: Julien Rouhaud
Reviewed-by: Wang Wei
Discussion: FIXME
---
 src/backend/access/transam/xlog.c |  2 +-
 src/backend/replication/slot.c    | 26 ++++++++++++++++----------
 src/include/replication/slot.h    |  2 +-
 3 files changed, 18 insertions(+), 12 deletions(-)

diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 8b0710abe6..38dac88247 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7015,7 +7015,7 @@ static void
 CheckPointGuts(XLogRecPtr checkPointRedo, int flags)
 {
 	CheckPointRelationMap();
-	CheckPointReplicationSlots();
+	CheckPointReplicationSlots(flags & CHECKPOINT_IS_SHUTDOWN);
 	CheckPointSnapBuild();
 	CheckPointLogicalRewriteHeap();
 	CheckPointReplicationOrigin();
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index 1dc27264f6..5aed7cd190 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -109,7 +109,8 @@ static void ReplicationSlotDropPtr(ReplicationSlot *slot);
 /* internal persistency functions */
 static void RestoreSlotFromDisk(const char *name);
 static void CreateSlotOnDisk(ReplicationSlot *slot);
-static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel);
+static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+						   bool is_shutdown);
 
 /*
  * Report shared-memory space needed by ReplicationSlotsShmemInit.
@@ -783,7 +784,7 @@ ReplicationSlotSave(void)
 	Assert(MyReplicationSlot != NULL);
 
 	sprintf(path, "pg_replslot/%s", NameStr(MyReplicationSlot->data.name));
-	SaveSlotToPath(MyReplicationSlot, path, ERROR);
+	SaveSlotToPath(MyReplicationSlot, path, ERROR, false);
 }
 
 /*
@@ -1565,11 +1566,10 @@ restart:
 /*
  * Flush all replication slots to disk.
  *
- * This needn't actually be part of a checkpoint, but it's a convenient
- * location.
+ * is_shutdown is true in case of a shutdown checkpoint.
  */
 void
-CheckPointReplicationSlots(void)
+CheckPointReplicationSlots(bool is_shutdown)
 {
 	int			i;
 
@@ -1594,7 +1594,7 @@ CheckPointReplicationSlots(void)
 
 		/* save the slot to disk, locking is handled in SaveSlotToPath() */
 		sprintf(path, "pg_replslot/%s", NameStr(s->data.name));
-		SaveSlotToPath(s, path, LOG);
+		SaveSlotToPath(s, path, LOG, is_shutdown);
 	}
 	LWLockRelease(ReplicationSlotAllocationLock);
 }
@@ -1700,7 +1700,7 @@ CreateSlotOnDisk(ReplicationSlot *slot)
 
 	/* Write the actual state file. */
 	slot->dirty = true;			/* signal that we really need to write */
-	SaveSlotToPath(slot, tmppath, ERROR);
+	SaveSlotToPath(slot, tmppath, ERROR, false);
 
 	/* Rename the directory into place. */
 	if (rename(tmppath, path) != 0)
@@ -1726,7 +1726,8 @@ CreateSlotOnDisk(ReplicationSlot *slot)
  * Shared functionality between saving and creating a replication slot.
  */
 static void
-SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
+SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+			   bool is_shutdown)
 {
 	char		tmppath[MAXPGPATH];
 	char		path[MAXPGPATH];
@@ -1740,8 +1741,13 @@ SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
 	slot->just_dirtied = false;
 	SpinLockRelease(&slot->mutex);
 
-	/* and don't do anything if there's nothing to write */
-	if (!was_dirty)
+	/*
+	 * and don't do anything if there's nothing to write, unless it's this is
+	 * called for a logical slot during a shutdown checkpoint, as we want to
+	 * persist the confirmed_flush_lsn in that case, even if that's the only
+	 * modification.
+	 */
+	if (!was_dirty && (SlotIsPhysical(slot) || !is_shutdown))
 		return;
 
 	LWLockAcquire(&slot->io_in_progress_lock, LW_EXCLUSIVE);
diff --git a/src/include/replication/slot.h b/src/include/replication/slot.h
index a8a89dc784..7ca37c9f70 100644
--- a/src/include/replication/slot.h
+++ b/src/include/replication/slot.h
@@ -241,7 +241,7 @@ extern void ReplicationSlotNameForTablesync(Oid suboid, Oid relid, char *syncslo
 extern void ReplicationSlotDropAtPubNode(WalReceiverConn *wrconn, char *slotname, bool missing_ok);
 
 extern void StartupReplicationSlots(void);
-extern void CheckPointReplicationSlots(void);
+extern void CheckPointReplicationSlots(bool is_shutdown);
 
 extern void CheckSlotRequirements(void);
 extern void CheckSlotPermissions(void);
-- 
2.27.0

From 2fbe4412992e1d6821d8429249e78dceae402452 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Fri, 14 Apr 2023 08:59:03 +0000
Subject: [PATCH v17 3/4] pg_upgrade: Add check function for
 --include-logical-replication-slots option

Author: Hayato Kuroda
Reviewed-by: Wang Wei, Vignesh C
---
 src/bin/pg_upgrade/check.c                    | 70 +++++++++++++++++++
 .../t/003_logical_replication_slots.pl        | 21 ++++++
 2 files changed, 91 insertions(+)

diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index 4b54f5567c..a8888933f6 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -9,8 +9,10 @@
 
 #include "postgres_fe.h"
 
+#include "access/xlog_internal.h"
 #include "catalog/pg_authid_d.h"
 #include "catalog/pg_collation.h"
+#include "catalog/pg_control.h"
 #include "fe_utils/string_utils.h"
 #include "mb/pg_wchar.h"
 #include "pg_upgrade.h"
@@ -31,6 +33,7 @@ static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
 static void check_for_logical_replication_slots(ClusterInfo *new_cluster);
+static void check_for_confirmed_flush_lsn(ClusterInfo *cluster);
 
 static int num_slots_on_old_cluster;
 
@@ -109,6 +112,8 @@ check_and_dump_old_cluster(bool live_check)
 	check_for_composite_data_type_usage(&old_cluster);
 	check_for_reg_data_type_usage(&old_cluster);
 	check_for_isn_and_int8_passing_mismatch(&old_cluster);
+	if (user_opts.include_logical_slots)
+		check_for_confirmed_flush_lsn(&old_cluster);
 
 	/*
 	 * PG 16 increased the size of the 'aclitem' type, which breaks the
@@ -1478,3 +1483,68 @@ check_for_logical_replication_slots(ClusterInfo *new_cluster)
 
 	check_ok();
 }
+
+/*
+ * Verify that all logical replication slots consumed all WALs, except a
+ * CHECKPOINT_SHUTDOWN record.
+ */
+static void
+check_for_confirmed_flush_lsn(ClusterInfo *cluster)
+{
+	int			i,
+				ntups,
+				i_slotname;
+	bool		is_error = false;
+	PGresult   *res;
+	DbInfo	   *active_db = &cluster->dbarr.dbs[0];
+	PGconn	   *conn = connectToServer(cluster, active_db->db_name);
+
+	Assert(user_opts.include_logical_slots);
+
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(cluster->major_version) <= 1500)
+		return;
+
+	prep_status("Checking confirmed_flush_lsn for logical replication slots");
+
+	/*
+	 * Check that all logical replication slots have reached the current WAL
+	 * position, except for the CHECKPOINT_SHUTDOWN record. Even if all WALs
+	 * are consumed before shutting down the node, the checkpointer generates
+	 * a CHECKPOINT_SHUTDOWN record at shutdown, which cannot be consumed by
+	 * any slots. Therefore, we must allow for a difference between
+	 * pg_current_wal_insert_lsn() and confirmed_flush_lsn.
+	 */
+#define SHUTDOWN_RECORD_SIZE  (SizeOfXLogRecord + \
+							   SizeOfXLogRecordDataHeaderShort + \
+							   sizeof(CheckPoint))
+
+	res = executeQueryOrDie(conn,
+							"SELECT slot_name FROM pg_catalog.pg_replication_slots "
+							"WHERE (pg_catalog.pg_current_wal_insert_lsn() - confirmed_flush_lsn) > %d "
+							"AND temporary = false AND wal_status IN ('reserved', 'extended');",
+							(int) (SizeOfXLogLongPHD + SHUTDOWN_RECORD_SIZE));
+
+#undef SHUTDOWN_RECORD_SIZE
+
+	ntups = PQntuples(res);
+	i_slotname = PQfnumber(res, "slot_name");
+
+	for (i = 0; i < ntups; i++)
+	{
+		is_error = true;
+
+		pg_log(PG_WARNING,
+			   "\nWARNING: logical replication slot \"%s\" has not consumed WALs yet",
+			   PQgetvalue(res, i, i_slotname));
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	if (is_error)
+		pg_fatal("--include-logical-replication-slots requires that all "
+				 "logical replication slots consumed all the WALs");
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
index 9ca266f6b2..0bea0b5b17 100644
--- a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -112,6 +112,27 @@ rmtree($new_node->data_dir . "/pg_upgrade_output.d");
 # appropriate value
 $new_node->append_conf('postgresql.conf', "max_replication_slots = 10");
 
+# Cause a failure at the start of pg_upgrade because test_slot does not
+# finish consuming all the WALs
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with idle replication slots');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
 # Remove an unnecessary slot and consume WALs
 $old_node->start;
 $old_node->safe_psql(
-- 
2.27.0

From f9f631c391c2cc7eb2a6a1a41773626fee3a1022 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Mon, 24 Apr 2023 11:03:28 +0000
Subject: [PATCH v17 4/4] Change the method used to check logical replication
 slots during the live check

When a live check is requested, there is a possibility of additional changes
occurring, which may cause the current WAL position to exceed the confirmed_flush_lsn
of the slot. As a result, we check the confirmed_flush_lsn of each logical slot
instead. This is sufficient as all the WAL records will be sent during the publisher's
shutdown.

Author: Hayato Kuroda
Reviewed-by: Wang Wei, Vignesh C
---
 src/bin/pg_upgrade/check.c                    | 64 +++++++++++++++++-
 .../t/003_logical_replication_slots.pl        | 66 ++++++++++++++++++-
 2 files changed, 127 insertions(+), 3 deletions(-)

diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index a8888933f6..42c2e3bebc 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -34,6 +34,7 @@ static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
 static void check_for_logical_replication_slots(ClusterInfo *new_cluster);
 static void check_for_confirmed_flush_lsn(ClusterInfo *cluster);
+static void check_are_logical_slots_active(ClusterInfo *cluster);
 
 static int num_slots_on_old_cluster;
 
@@ -113,7 +114,19 @@ check_and_dump_old_cluster(bool live_check)
 	check_for_reg_data_type_usage(&old_cluster);
 	check_for_isn_and_int8_passing_mismatch(&old_cluster);
 	if (user_opts.include_logical_slots)
-		check_for_confirmed_flush_lsn(&old_cluster);
+	{
+		/*
+		 * The method used to check logical replication slots is dependent on
+		 * the value of the live_check parameter. This change was implemented
+		 * because, during a live check, it is possible for additional changes
+		 * to occur at the old node, which could cause the current WAL position
+		 * to exceed the confirmed_flush_lsn of the slot.
+		 */
+		if (live_check)
+			check_are_logical_slots_active(&old_cluster);
+		else
+			check_for_confirmed_flush_lsn(&old_cluster);
+	}
 
 	/*
 	 * PG 16 increased the size of the 'aclitem' type, which breaks the
@@ -1484,6 +1497,55 @@ check_for_logical_replication_slots(ClusterInfo *new_cluster)
 	check_ok();
 }
 
+/*
+ * Verify that all logical replication slots are active
+ */
+static void
+check_are_logical_slots_active(ClusterInfo *cluster)
+{
+	int			i,
+				ntups,
+				i_slotname;
+	bool		is_error = false;
+	PGresult   *res;
+	DbInfo	   *active_db = &cluster->dbarr.dbs[0];
+	PGconn	   *conn = connectToServer(cluster, active_db->db_name);
+
+	Assert(user_opts.include_logical_slots);
+
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(cluster->major_version) <= 1500)
+		return;
+
+	prep_status("Checking the status of subscriptions for logical replication slots");
+
+	res = executeQueryOrDie(conn,
+							"SELECT slot_name FROM pg_catalog.pg_replication_slots "
+							"WHERE active IS FALSE "
+							"AND temporary = false AND wal_status IN ('reserved', 'extended');");
+
+	ntups = PQntuples(res);
+	i_slotname = PQfnumber(res, "slot_name");
+
+	for (i = 0; i < ntups; i++)
+	{
+		is_error = true;
+
+		pg_log(PG_WARNING,
+			   "\nWARNING: logical replication slot \"%s\" is not active",
+			   PQgetvalue(res, i, i_slotname));
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	if (is_error)
+		pg_fatal("--include-logical-replication-slots with --check requires that "
+				 "all logical replication slots are active");
+
+	check_ok();
+}
+
 /*
  * Verify that all logical replication slots consumed all WALs, except a
  * CHECKPOINT_SHUTDOWN record.
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
index 0bea0b5b17..f711452737 100644
--- a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -19,6 +19,10 @@ my $old_node = PostgreSQL::Test::Cluster->new('old_node');
 $old_node->init(allows_streaming => 'logical');
 $old_node->start;
 
+# Initialize subscriber, which will be used only for --check
+my $subscriber = PostgreSQL::Test::Cluster->new('subscriber');
+$subscriber->init(allows_streaming => 'logical');
+
 # Initialize new node
 my $new_node = PostgreSQL::Test::Cluster->new('new_node');
 $new_node->init(allows_streaming => 1);
@@ -76,13 +80,70 @@ rmtree($new_node->data_dir . "/pg_upgrade_output.d");
 # non-zero value
 $new_node->append_conf('postgresql.conf', "max_replication_slots = 1");
 
-# Create a slot on old node, and generate WALs
+# Setup logical replication
 $old_node->start;
+$old_node->safe_psql('postgres',
+	"CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a");
+$old_node->safe_psql('postgres', "CREATE PUBLICATION pub FOR ALL TABLES");
+my $old_connstr = $old_node->connstr . ' dbname=postgres';
+
+$subscriber->start;
+$subscriber->safe_psql('postgres', "CREATE TABLE tbl (a int)");
+$subscriber->safe_psql('postgres',
+	"CREATE SUBSCRIPTION sub CONNECTION '$old_connstr' PUBLICATION pub WITH (copy_data = true)"
+);
+
+# Wait for initial table sync to finish
+$subscriber->wait_for_subscription_sync($old_node, 'sub');
+
+my $result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(10), 'check initial rows on subscriber');
+
+# Start a background session and open a transaction (not committed yet)
+my $bsession = $old_node->background_psql('postgres');
+$bsession->query_safe(
+	q{
+BEGIN;
+INSERT INTO tbl VALUES (generate_series(11, 20))
+});
+
+$result = $old_node->safe_psql('postgres',
+	"SELECT count(*) FROM pg_replication_slots WHERE pg_current_wal_insert_lsn() > confirmed_flush_lsn"
+);
+is($result, qq(1),
+	'check the current WAL position exceeds confirmed_flush_lsn');
+
+# Run pg_upgrade --check. In the command the status of each logical slots will
+# be checked and then this will be succeeded.
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+		'--check'
+	],
+	'run of pg_upgrade of old node');
+ok( !-d $old_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+# Cleanup
+$bsession->query_safe("ABORT");
+$bsession->quit;
+$subscriber->safe_psql('postgres', "DROP SUBSCRIPTION sub");
+$subscriber->stop();
+
+# Create a slot on old node, and generate WALs
 $old_node->safe_psql(
 	'postgres', qq[
 	SELECT pg_create_logical_replication_slot('test_slot', 'test_decoding', false, true);
 	SELECT pg_create_logical_replication_slot('to_be_dropped', 'test_decoding', false, true);
-	CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a;
+	INSERT INTO tbl VALUES (generate_series(11, 20));
 ]);
 
 $old_node->stop;
@@ -163,5 +224,6 @@ $new_node->start;
 my $result = $new_node->safe_psql('postgres',
 	"SELECT slot_name, two_phase FROM pg_replication_slots");
 is($result, qq(test_slot|t), 'check the slot exists on new node');
+$new_node->stop();
 
 done_testing();
-- 
2.27.0

From 7f8f1cb96eab09b107d9022aea3b4386796a4dce Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Fri, 21 Jul 2023 05:34:21 +0000
Subject: [PATCH 1/3] Send shutdown checkpoint record to subscriber

---
 src/backend/replication/walsender.c | 30 +++++++++++++++++++++++------
 1 file changed, 24 insertions(+), 6 deletions(-)

diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c
index d27ef2985d..fc1363ba76 100644
--- a/src/backend/replication/walsender.c
+++ b/src/backend/replication/walsender.c
@@ -187,6 +187,9 @@ static bool WalSndCaughtUp = false;
 static volatile sig_atomic_t got_SIGUSR2 = false;
 static volatile sig_atomic_t got_STOPPING = false;
 
+/* Are all the WALs flushed? */
+static bool WalsAreFlushed = false;
+
 /*
  * This is set while we are streaming. When not set
  * PROCSIG_WALSND_INIT_STOPPING signal will be handled like SIGTERM. When set,
@@ -260,7 +263,6 @@ static bool TransactionIdInRecentPast(TransactionId xid, uint32 epoch);
 static void WalSndSegmentOpen(XLogReaderState *state, XLogSegNo nextSegNo,
 							  TimeLineID *tli_p);
 
-
 /* Initialize walsender process before entering the main command loop */
 void
 InitWalSender(void)
@@ -1581,7 +1583,10 @@ WalSndWaitForWal(XLogRecPtr loc)
 		 * written, because walwriter has shut down already.
 		 */
 		if (got_STOPPING)
+		{
 			XLogBackgroundFlush();
+			WalsAreFlushed = true;
+		}
 
 		/* Update our idea of the currently flushed position. */
 		if (!RecoveryInProgress())
@@ -3100,12 +3105,20 @@ XLogSendLogical(void)
 		WalSndCaughtUp = true;
 
 	/*
-	 * If we're caught up and have been requested to stop, have WalSndLoop()
-	 * terminate the connection in an orderly manner, after writing out all
-	 * the pending data.
+	 * If we're caught up, have been requested to stop and there are no pending
+	 * records to be sent, change to stopping mode.
 	 */
-	if (WalSndCaughtUp && got_STOPPING)
-		got_SIGUSR2 = true;
+	if (WalSndCaughtUp && WalsAreFlushed && !pq_is_send_pending())
+	{
+		/*
+		 * Update the stats forcibly. pgstat_shutdown_hook reports any pending
+		 * stats at the end of the process, but it would happen after the
+		 * checkpointer exits so that it would lead assertion failure. We must
+		 * ensure all the stats are recorded before changing the state.
+		 */
+		pgstat_report_stat(true);
+		WalSndSetState(WALSNDSTATE_STOPPING);
+	}
 
 	/* Update shared memory status */
 	{
@@ -3142,6 +3155,7 @@ WalSndDone(WalSndSendDataCallback send_data)
 	replicatedPtr = XLogRecPtrIsInvalid(MyWalSnd->flush) ?
 		MyWalSnd->write : MyWalSnd->flush;
 
+
 	if (WalSndCaughtUp && sentPtr == replicatedPtr &&
 		!pq_is_send_pending())
 	{
@@ -3152,6 +3166,10 @@ WalSndDone(WalSndSendDataCallback send_data)
 		EndCommand(&qc, DestRemote, false);
 		pq_flush();
 
+		/* Mark the slot as dirty and save it to update the confirmed_flush. */
+		ReplicationSlotMarkDirty();
+		ReplicationSlotSave();
+
 		proc_exit(0);
 	}
 	if (!waiting_for_ping_response)
-- 
2.27.0

From 5287baac5bf7b78bed33ae0655fd8066bde23330 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Tue, 4 Apr 2023 05:49:34 +0000
Subject: [PATCH 2/3] pg_upgrade: Add --include-logical-replication-slots
 option

This commit introduces a new pg_upgrade option called "--include-logical-replication-slots".
This allows nodes with logical replication slots to be upgraded. The commit can
be divided into two parts: one for pg_dump and another for pg_upgrade.

For pg_dump this commit includes a new option called "--logical-replication-slots-only".
This option can be used to dump logical replication slots. When this option is
specified, the slot_name, plugin, and two_phase parameters are extracted from
pg_replication_slots. An SQL file that executes pg_create_logical_replication_slot()
with the extracted parameters is generated.

For pg_upgrade, when '--include-logical-replication-slots' is specified, it executes
pg_dump with the new "--logical-replication-slots-only" option and restores the slots
using the pg_create_logical_replication_slots() statements that the dump
generated (see above). Note that we cannot dump replication slots at the same time
as the schema dump because we need to separate the timing of restoring replication
slots and other objects. Replication slots, in  particular, should not be restored
before executing the pg_resetwal command because it will remove WALs that are
required by the slots.

The significant advantage of this commit is that it makes it easy to continue
logical replication even after upgrading the publisher node. Previously, pg_upgrade
allowed copying publications to a new node. With this new commit, adjusting the
connection string to the new publisher will cause the apply worker on the subscriber
to connect to the new publisher automatically. This enables seamless continuation
of logical replication, even after an upgrade.

Author: Hayato Kuroda
Reviewed-by: Peter Smith, Julien Rouhaud, Vignesh C, Wang Wei
---
 doc/src/sgml/ref/pgupgrade.sgml               |  11 ++
 src/bin/pg_dump/pg_backup.h                   |   1 +
 src/bin/pg_dump/pg_dump.c                     | 155 ++++++++++++++++++
 src/bin/pg_dump/pg_dump.h                     |  14 +-
 src/bin/pg_dump/pg_dump_sort.c                |  11 +-
 src/bin/pg_upgrade/Makefile                   |   3 +
 src/bin/pg_upgrade/check.c                    |  76 +++++++++
 src/bin/pg_upgrade/dump.c                     |  24 +++
 src/bin/pg_upgrade/info.c                     | 111 ++++++++++++-
 src/bin/pg_upgrade/meson.build                |   1 +
 src/bin/pg_upgrade/option.c                   |   7 +
 src/bin/pg_upgrade/pg_upgrade.c               |  61 +++++++
 src/bin/pg_upgrade/pg_upgrade.h               |  21 +++
 .../t/003_logical_replication_slots.pl        | 146 +++++++++++++++++
 src/tools/pgindent/typedefs.list              |   3 +
 15 files changed, 641 insertions(+), 4 deletions(-)
 create mode 100644 src/bin/pg_upgrade/t/003_logical_replication_slots.pl

diff --git a/doc/src/sgml/ref/pgupgrade.sgml b/doc/src/sgml/ref/pgupgrade.sgml
index 7816b4c685..94e90ff506 100644
--- a/doc/src/sgml/ref/pgupgrade.sgml
+++ b/doc/src/sgml/ref/pgupgrade.sgml
@@ -240,6 +240,17 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--include-logical-replication-slots</option></term>
+      <listitem>
+       <para>
+        Upgrade logical replication slots. Only permanent replication slots
+        are included. Note that pg_upgrade does not check the installation of
+        plugins.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
       <term><option>-?</option></term>
       <term><option>--help</option></term>
diff --git a/src/bin/pg_dump/pg_backup.h b/src/bin/pg_dump/pg_backup.h
index aba780ef4b..0a4e931f9b 100644
--- a/src/bin/pg_dump/pg_backup.h
+++ b/src/bin/pg_dump/pg_backup.h
@@ -187,6 +187,7 @@ typedef struct _dumpOptions
 	int			use_setsessauth;
 	int			enable_row_security;
 	int			load_via_partition_root;
+	int			logical_slots_only;
 
 	/* default, if no "inclusion" switches appear, is to dump everything */
 	bool		include_everything;
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index 5dab1ba9ea..12d4066d3b 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -328,6 +328,9 @@ static void setupDumpWorker(Archive *AH);
 static TableInfo *getRootTableInfo(const TableInfo *tbinfo);
 static bool forcePartitionRootLoad(const TableInfo *tbinfo);
 
+static void getLogicalReplicationSlots(Archive *fout);
+static void dumpLogicalReplicationSlot(Archive *fout,
+									   const LogicalReplicationSlotInfo *slotinfo);
 
 int
 main(int argc, char **argv)
@@ -431,6 +434,7 @@ main(int argc, char **argv)
 		{"table-and-children", required_argument, NULL, 12},
 		{"exclude-table-and-children", required_argument, NULL, 13},
 		{"exclude-table-data-and-children", required_argument, NULL, 14},
+		{"logical-replication-slots-only", no_argument, NULL, 15},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -657,6 +661,10 @@ main(int argc, char **argv)
 										  optarg);
 				break;
 
+			case 15:			/* dump only replication slot(s) */
+				dopt.logical_slots_only = true;
+				break;
+
 			default:
 				/* getopt_long already emitted a complaint */
 				pg_log_error_hint("Try \"%s --help\" for more information.", progname);
@@ -714,6 +722,21 @@ main(int argc, char **argv)
 	if (dopt.do_nothing && dopt.dump_inserts == 0)
 		pg_fatal("option --on-conflict-do-nothing requires option --inserts, --rows-per-insert, or --column-inserts");
 
+	if (dopt.logical_slots_only)
+	{
+		if (!dopt.binary_upgrade)
+			pg_fatal("options --logical-replication-slots-only requires option --binary-upgrade");
+
+		if (dopt.dataOnly)
+			pg_fatal("options --logical-replication-slots-only and -a/--data-only cannot be used together");
+
+		if (dopt.schemaOnly)
+			pg_fatal("options --logical-replication-slots-only and -s/--schema-only cannot be used together");
+
+		if (dopt.outputClean)
+			pg_fatal("options --logical-replication-slots-only and -c/--clean cannot be used together");
+	}
+
 	/* Identify archive format to emit */
 	archiveFormat = parseArchiveFormat(format, &archiveMode);
 
@@ -876,6 +899,16 @@ main(int argc, char **argv)
 			pg_fatal("no matching extensions were found");
 	}
 
+	/*
+	 * If dump logical-replication-slots-only was requested, dump only them
+	 * and skip everything else.
+	 */
+	if (dopt.logical_slots_only)
+	{
+		getLogicalReplicationSlots(fout);
+		goto dump;
+	}
+
 	/*
 	 * Dumping LOs is the default for dumps where an inclusion switch is not
 	 * used (an "include everything" dump).  -B can be used to exclude LOs
@@ -936,6 +969,8 @@ main(int argc, char **argv)
 	if (!dopt.no_security_labels)
 		collectSecLabels(fout);
 
+dump:
+
 	/* Lastly, create dummy objects to represent the section boundaries */
 	boundaryObjs = createBoundaryObjects();
 
@@ -1109,6 +1144,12 @@ help(const char *progname)
 			 "                               servers matching PATTERN\n"));
 	printf(_("  --inserts                    dump data as INSERT commands, rather than COPY\n"));
 	printf(_("  --load-via-partition-root    load partitions via the root table\n"));
+
+	/*
+	 * The option --logical-replication-slots-only is used only by pg_upgrade
+	 * and should not be called by users, which is why it is not exposed by the
+	 * help.
+	 */
 	printf(_("  --no-comments                do not dump comments\n"));
 	printf(_("  --no-publications            do not dump publications\n"));
 	printf(_("  --no-security-labels         do not dump security label assignments\n"));
@@ -10237,6 +10278,10 @@ dumpDumpableObject(Archive *fout, DumpableObject *dobj)
 		case DO_SUBSCRIPTION:
 			dumpSubscription(fout, (const SubscriptionInfo *) dobj);
 			break;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			dumpLogicalReplicationSlot(fout,
+									   (const LogicalReplicationSlotInfo *) dobj);
+			break;
 		case DO_PRE_DATA_BOUNDARY:
 		case DO_POST_DATA_BOUNDARY:
 			/* never dumped, nothing to do */
@@ -18218,6 +18263,7 @@ addBoundaryDependencies(DumpableObject **dobjs, int numObjs,
 			case DO_PUBLICATION_REL:
 			case DO_PUBLICATION_TABLE_IN_SCHEMA:
 			case DO_SUBSCRIPTION:
+			case DO_LOGICAL_REPLICATION_SLOT:
 				/* Post-data objects: must come after the post-data boundary */
 				addObjectDependency(dobj, postDataBound->dumpId);
 				break;
@@ -18479,3 +18525,112 @@ appendReloptionsArrayAH(PQExpBuffer buffer, const char *reloptions,
 	if (!res)
 		pg_log_warning("could not parse %s array", "reloptions");
 }
+
+/*
+ * getLogicalReplicationSlots
+ *	  get information about replication slots
+ */
+static void
+getLogicalReplicationSlots(Archive *fout)
+{
+	PGresult   *res;
+	LogicalReplicationSlotInfo *slotinfo;
+	PQExpBuffer query;
+
+	int			i_slotname;
+	int			i_plugin;
+	int			i_twophase;
+	int			i,
+				ntups;
+
+	/* Check whether we should dump or not */
+	if (fout->remoteVersion < 170000)
+		return;
+
+	Assert(fout->dopt->logical_slots_only);
+
+	query = createPQExpBuffer();
+
+	resetPQExpBuffer(query);
+
+	/*
+	 * Get replication slots.
+	 *
+	 * XXX: Which information must be extracted from old node? Currently three
+	 * attributes are extracted because they are used by
+	 * pg_create_logical_replication_slot().
+	 */
+	appendPQExpBufferStr(query,
+						 "SELECT slot_name, plugin, two_phase "
+						 "FROM pg_catalog.pg_replication_slots "
+						 "WHERE database = current_database() AND temporary = false "
+						 "AND wal_status IN ('reserved', 'extended');");
+
+	res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
+
+	ntups = PQntuples(res);
+
+	i_slotname = PQfnumber(res, "slot_name");
+	i_plugin = PQfnumber(res, "plugin");
+	i_twophase = PQfnumber(res, "two_phase");
+
+	slotinfo = pg_malloc(ntups * sizeof(LogicalReplicationSlotInfo));
+
+	for (i = 0; i < ntups; i++)
+	{
+		slotinfo[i].dobj.objType = DO_LOGICAL_REPLICATION_SLOT;
+
+		slotinfo[i].dobj.catId.tableoid = InvalidOid;
+		slotinfo[i].dobj.catId.oid = InvalidOid;
+		AssignDumpId(&slotinfo[i].dobj);
+
+		slotinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_slotname));
+
+		slotinfo[i].plugin = pg_strdup(PQgetvalue(res, i, i_plugin));
+		slotinfo[i].twophase = (strcmp(PQgetvalue(res, i, i_twophase), "t") == 0);
+
+		/*
+		 * Note: Currently we do not have any options to include/exclude slots
+		 * in dumping, so all the slots must be selected.
+		 */
+		slotinfo[i].dobj.dump = DUMP_COMPONENT_DEFINITION;
+	}
+	PQclear(res);
+
+	destroyPQExpBuffer(query);
+}
+
+/*
+ * dumpLogicalReplicationSlot
+ *	  dump creation functions for the given logical replication slots
+ */
+static void
+dumpLogicalReplicationSlot(Archive *fout,
+						   const LogicalReplicationSlotInfo *slotinfo)
+{
+	Assert(fout->dopt->logical_slots_only);
+
+	if (slotinfo->dobj.dump & DUMP_COMPONENT_DEFINITION)
+	{
+		PQExpBuffer query = createPQExpBuffer();
+
+		/*
+		 * XXX: For simplification, pg_create_logical_replication_slot() is
+		 * used. Is it sufficient?
+		 */
+		appendPQExpBuffer(query, "SELECT pg_catalog.pg_create_logical_replication_slot(");
+		appendStringLiteralAH(query, slotinfo->dobj.name, fout);
+		appendPQExpBuffer(query, ", ");
+		appendStringLiteralAH(query, slotinfo->plugin, fout);
+		appendPQExpBuffer(query, ", false, %s);",
+						  slotinfo->twophase ? "true" : "false");
+
+		ArchiveEntry(fout, slotinfo->dobj.catId, slotinfo->dobj.dumpId,
+					 ARCHIVE_OPTS(.tag = slotinfo->dobj.name,
+								  .description = "REPLICATION SLOT",
+								  .section = SECTION_POST_DATA,
+								  .createStmt = query->data));
+
+		destroyPQExpBuffer(query);
+	}
+}
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index bc8f2ec36d..ed1866d9ab 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -82,7 +82,8 @@ typedef enum
 	DO_PUBLICATION,
 	DO_PUBLICATION_REL,
 	DO_PUBLICATION_TABLE_IN_SCHEMA,
-	DO_SUBSCRIPTION
+	DO_SUBSCRIPTION,
+	DO_LOGICAL_REPLICATION_SLOT
 } DumpableObjectType;
 
 /*
@@ -667,6 +668,17 @@ typedef struct _SubscriptionInfo
 	char	   *subpasswordrequired;
 } SubscriptionInfo;
 
+/*
+ * The LogicalReplicationSlotInfo struct is used to represent replication
+ * slots.
+ */
+typedef struct _LogicalReplicationSlotInfo
+{
+	DumpableObject dobj;
+	char	   *plugin;
+	bool		twophase;
+} LogicalReplicationSlotInfo;
+
 /*
  *	common utility functions
  */
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index 523a19c155..ae65443228 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -93,6 +93,7 @@ enum dbObjectTypePriorities
 	PRIO_PUBLICATION_REL,
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,
 	PRIO_SUBSCRIPTION,
+	PRIO_LOGICAL_REPLICATION_SLOT,
 	PRIO_DEFAULT_ACL,			/* done in ACL pass */
 	PRIO_EVENT_TRIGGER,			/* must be next to last! */
 	PRIO_REFRESH_MATVIEW		/* must be last! */
@@ -146,10 +147,11 @@ static const int dbObjectTypePriority[] =
 	PRIO_PUBLICATION,			/* DO_PUBLICATION */
 	PRIO_PUBLICATION_REL,		/* DO_PUBLICATION_REL */
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,	/* DO_PUBLICATION_TABLE_IN_SCHEMA */
-	PRIO_SUBSCRIPTION			/* DO_SUBSCRIPTION */
+	PRIO_SUBSCRIPTION,			/* DO_SUBSCRIPTION */
+	PRIO_LOGICAL_REPLICATION_SLOT	/* DO_LOGICAL_REPLICATION_SLOT */
 };
 
-StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_SUBSCRIPTION + 1),
+StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_LOGICAL_REPLICATION_SLOT + 1),
 				 "array length mismatch");
 
 static DumpId preDataBoundId;
@@ -1542,6 +1544,11 @@ describeDumpableObject(DumpableObject *obj, char *buf, int bufsize)
 					 "SUBSCRIPTION (ID %d OID %u)",
 					 obj->dumpId, obj->catId.oid);
 			return;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			snprintf(buf, bufsize,
+					 "LOGICAL REPLICATION SLOT (ID %d NAME %s)",
+					 obj->dumpId, obj->name);
+			return;
 		case DO_PRE_DATA_BOUNDARY:
 			snprintf(buf, bufsize,
 					 "PRE-DATA BOUNDARY  (ID %d)",
diff --git a/src/bin/pg_upgrade/Makefile b/src/bin/pg_upgrade/Makefile
index 5834513add..815d1a7ca1 100644
--- a/src/bin/pg_upgrade/Makefile
+++ b/src/bin/pg_upgrade/Makefile
@@ -3,6 +3,9 @@
 PGFILEDESC = "pg_upgrade - an in-place binary upgrade utility"
 PGAPPICON = win32
 
+# required for 003_logical_replication_slots.pl
+EXTRA_INSTALL=contrib/test_decoding
+
 subdir = src/bin/pg_upgrade
 top_builddir = ../../..
 include $(top_builddir)/src/Makefile.global
diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index 64024e3b9e..4b54f5567c 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -30,7 +30,9 @@ static void check_for_jsonb_9_4_usage(ClusterInfo *cluster);
 static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
+static void check_for_logical_replication_slots(ClusterInfo *new_cluster);
 
+static int num_slots_on_old_cluster;
 
 /*
  * fix_path_separator
@@ -89,6 +91,10 @@ check_and_dump_old_cluster(bool live_check)
 	/* Extract a list of databases and tables from the old cluster */
 	get_db_and_rel_infos(&old_cluster);
 
+	/* Additionally, extract a list of logical replication slots if required */
+	if (user_opts.include_logical_slots)
+		num_slots_on_old_cluster = get_logical_slot_infos(&old_cluster);
+
 	init_tablespaces();
 
 	get_loadable_libraries();
@@ -189,6 +195,17 @@ check_new_cluster(void)
 {
 	get_db_and_rel_infos(&new_cluster);
 
+	/*
+	 * Do additional work if --include-logical-replication-slots was specified.
+	 * This must be done before check_new_cluster_is_empty() because the
+	 * slot_arr attribute of the new_cluster will be checked in that function.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		(void) get_logical_slot_infos(&new_cluster);
+		check_for_logical_replication_slots(&new_cluster);
+	}
+
 	check_new_cluster_is_empty();
 
 	check_loadable_libraries();
@@ -364,6 +381,22 @@ check_new_cluster_is_empty(void)
 						 rel_arr->rels[relnum].nspname,
 						 rel_arr->rels[relnum].relname);
 		}
+
+		/*
+		 * If --include-logical-replication-slots is required, check the
+		 * existence of slots.
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			DbInfo *pDbInfo = &new_cluster.dbarr.dbs[dbnum];
+			LogicalSlotInfoArr *slot_arr = &pDbInfo->slot_arr;
+
+			/* if nslots > 0, report just first entry and exit */
+			if (slot_arr->nslots)
+				pg_fatal("New cluster database \"%s\" is not empty: found logical replication slot \"%s\"",
+						 pDbInfo->db_name,
+						 slot_arr->slots[0].slotname);
+		}
 	}
 }
 
@@ -1402,3 +1435,46 @@ check_for_user_defined_encoding_conversions(ClusterInfo *cluster)
 	else
 		check_ok();
 }
+
+/*
+ * Verify the parameter settings necessary for creating logical replication
+ * slots.
+ */
+static void
+check_for_logical_replication_slots(ClusterInfo *new_cluster)
+{
+	PGresult   *res;
+	PGconn	   *conn = connectToServer(new_cluster, "template1");
+	int			max_replication_slots;
+	char	   *wal_level;
+
+	/* --include-logical-replication-slots can be used since PG17. */
+	if (GET_MAJOR_VERSION(new_cluster->major_version) <= 1600)
+		return;
+
+	prep_status("Checking parameter settings for logical replication slots");
+
+	res = executeQueryOrDie(conn, "SHOW max_replication_slots;");
+	max_replication_slots = atoi(PQgetvalue(res, 0, 0));
+
+	if (max_replication_slots == 0)
+		pg_fatal("max_replication_slots must be greater than 0");
+	else if (num_slots_on_old_cluster > max_replication_slots)
+		pg_fatal("max_replication_slots must be greater than existing logical "
+				 "replication slots on old node.");
+
+	PQclear(res);
+
+	res = executeQueryOrDie(conn, "SHOW wal_level;");
+	wal_level = PQgetvalue(res, 0, 0);
+
+	if (strcmp(wal_level, "logical") != 0)
+		pg_fatal("wal_level must be \"logical\", but is set to \"%s\"",
+				 wal_level);
+
+	PQclear(res);
+
+	PQfinish(conn);
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/dump.c b/src/bin/pg_upgrade/dump.c
index 6c8c82dca8..e6b90864f5 100644
--- a/src/bin/pg_upgrade/dump.c
+++ b/src/bin/pg_upgrade/dump.c
@@ -59,6 +59,30 @@ generate_old_dump(void)
 						   log_opts.dumpdir,
 						   sql_file_name, escaped_connstr.data);
 
+		/*
+		 * Dump logical replication slots if needed.
+		 *
+		 * XXX We cannot dump replication slots at the same time as the schema
+		 * dump because we need to separate the timing of restoring
+		 * replication slots and other objects. Replication slots, in
+		 * particular, should not be restored before executing the pg_resetwal
+		 * command because it will remove WALs that are required by the slots.
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			char		slots_file_name[MAXPGPATH];
+
+			snprintf(slots_file_name, sizeof(slots_file_name),
+					 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+			parallel_exec_prog(log_file_name, NULL,
+							   "\"%s/pg_dump\" %s --logical-replication-slots-only "
+							   "--quote-all-identifiers --binary-upgrade %s "
+							   "--file=\"%s/%s\" %s",
+							   new_cluster.bindir, cluster_conn_opts(&old_cluster),
+							   log_opts.verbose ? "--verbose" : "",
+							   log_opts.dumpdir,
+							   slots_file_name, escaped_connstr.data);
+		}
 		termPQExpBuffer(&escaped_connstr);
 	}
 
diff --git a/src/bin/pg_upgrade/info.c b/src/bin/pg_upgrade/info.c
index a9988abfe1..8bc0ad2e10 100644
--- a/src/bin/pg_upgrade/info.c
+++ b/src/bin/pg_upgrade/info.c
@@ -26,6 +26,7 @@ static void get_rel_infos(ClusterInfo *cluster, DbInfo *dbinfo);
 static void free_rel_infos(RelInfoArr *rel_arr);
 static void print_db_infos(DbInfoArr *db_arr);
 static void print_rel_infos(RelInfoArr *rel_arr);
+static void print_slot_infos(LogicalSlotInfoArr *slot_arr);
 
 
 /*
@@ -394,7 +395,7 @@ get_db_infos(ClusterInfo *cluster)
 	i_spclocation = PQfnumber(res, "spclocation");
 
 	ntups = PQntuples(res);
-	dbinfos = (DbInfo *) pg_malloc(sizeof(DbInfo) * ntups);
+	dbinfos = (DbInfo *) pg_malloc0(sizeof(DbInfo) * ntups);
 
 	for (tupnum = 0; tupnum < ntups; tupnum++)
 	{
@@ -600,6 +601,96 @@ get_rel_infos(ClusterInfo *cluster, DbInfo *dbinfo)
 	dbinfo->rel_arr.nrels = num_rels;
 }
 
+/*
+ * get_logical_slot_infos_per_db()
+ *
+ * gets the LogicalSlotInfos for all the logical replication slots of the database
+ * referred to by "dbinfo".
+ */
+static void
+get_logical_slot_infos_per_db(ClusterInfo *cluster, DbInfo *dbinfo)
+{
+	PGconn	   *conn = connectToServer(cluster,
+									   dbinfo->db_name);
+	PGresult   *res;
+	LogicalSlotInfo *slotinfos = NULL;
+
+	int			num_slots;
+
+	char		query[QUERY_ALLOC];
+
+	snprintf(query, sizeof(query),
+			 "SELECT slot_name, plugin, two_phase "
+			 "FROM pg_catalog.pg_replication_slots "
+			 "WHERE database = current_database() AND temporary = false "
+			 "AND wal_status IN ('reserved', 'extended');");
+
+	res = executeQueryOrDie(conn, "%s", query);
+
+	num_slots = PQntuples(res);
+
+	if (num_slots)
+	{
+		int			slotnum;
+		int			i_slotname;
+		int			i_plugin;
+		int			i_twophase;
+
+		slotinfos = (LogicalSlotInfo *) pg_malloc(sizeof(LogicalSlotInfo) * num_slots);
+
+		i_slotname = PQfnumber(res, "slot_name");
+		i_plugin = PQfnumber(res, "plugin");
+		i_twophase = PQfnumber(res, "two_phase");
+
+		for (slotnum = 0; slotnum < num_slots; slotnum++)
+		{
+			LogicalSlotInfo *curr = &slotinfos[slotnum];
+
+			curr->slotname = pg_strdup(PQgetvalue(res, slotnum, i_slotname));
+			curr->plugin = pg_strdup(PQgetvalue(res, slotnum, i_plugin));
+			curr->two_phase = (strcmp(PQgetvalue(res, slotnum, i_twophase), "t") == 0);
+		}
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	dbinfo->slot_arr.slots = slotinfos;
+	dbinfo->slot_arr.nslots = num_slots;
+}
+
+/*
+ * get_logical_slot_infos()
+ *
+ * Higher level routine to generate LogicalSlotInfoArr for all databases.
+ */
+int
+get_logical_slot_infos(ClusterInfo *cluster)
+{
+	int			dbnum;
+	int			slotnum = 0;
+
+	if (cluster == &old_cluster)
+		pg_log(PG_VERBOSE, "\nsource databases:");
+	else
+		pg_log(PG_VERBOSE, "\ntarget databases:");
+
+	for (dbnum = 0; dbnum < cluster->dbarr.ndbs; dbnum++)
+	{
+		DbInfo	   *pDbInfo = &cluster->dbarr.dbs[dbnum];
+
+		get_logical_slot_infos_per_db(cluster, pDbInfo);
+		slotnum += pDbInfo->slot_arr.nslots;
+
+		if (log_opts.verbose)
+		{
+			pg_log(PG_VERBOSE, "Database: \"%s\"", pDbInfo->db_name);
+			print_slot_infos(&pDbInfo->slot_arr);
+		}
+	}
+
+	return slotnum;
+}
 
 static void
 free_db_and_rel_infos(DbInfoArr *db_arr)
@@ -610,6 +701,12 @@ free_db_and_rel_infos(DbInfoArr *db_arr)
 	{
 		free_rel_infos(&db_arr->dbs[dbnum].rel_arr);
 		pg_free(db_arr->dbs[dbnum].db_name);
+
+		/*
+		 * Logical replication slots must not exist on the new cluster before
+		 * doing create_logical_replication_slots().
+		 */
+		Assert(db_arr->dbs[dbnum].slot_arr.slots == NULL);
 	}
 	pg_free(db_arr->dbs);
 	db_arr->dbs = NULL;
@@ -660,3 +757,15 @@ print_rel_infos(RelInfoArr *rel_arr)
 			   rel_arr->rels[relnum].reloid,
 			   rel_arr->rels[relnum].tablespace);
 }
+
+static void
+print_slot_infos(LogicalSlotInfoArr *slot_arr)
+{
+	int			slotnum;
+
+	for (slotnum = 0; slotnum < slot_arr->nslots; slotnum++)
+		pg_log(PG_VERBOSE, "slotname: \"%s\", plugin: \"%s\", two_phase: %d",
+			   slot_arr->slots[slotnum].slotname,
+			   slot_arr->slots[slotnum].plugin,
+			   slot_arr->slots[slotnum].two_phase);
+}
diff --git a/src/bin/pg_upgrade/meson.build b/src/bin/pg_upgrade/meson.build
index 12a97f84e2..228f29b688 100644
--- a/src/bin/pg_upgrade/meson.build
+++ b/src/bin/pg_upgrade/meson.build
@@ -42,6 +42,7 @@ tests += {
     'tests': [
       't/001_basic.pl',
       't/002_pg_upgrade.pl',
+      't/003_logical_replication_slots.pl',
     ],
     'test_kwargs': {'priority': 40}, # pg_upgrade tests are slow
   },
diff --git a/src/bin/pg_upgrade/option.c b/src/bin/pg_upgrade/option.c
index 640361009e..df66a5ffe6 100644
--- a/src/bin/pg_upgrade/option.c
+++ b/src/bin/pg_upgrade/option.c
@@ -57,6 +57,7 @@ parseCommandLine(int argc, char *argv[])
 		{"verbose", no_argument, NULL, 'v'},
 		{"clone", no_argument, NULL, 1},
 		{"copy", no_argument, NULL, 2},
+		{"include-logical-replication-slots", no_argument, NULL, 3},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -199,6 +200,10 @@ parseCommandLine(int argc, char *argv[])
 				user_opts.transfer_mode = TRANSFER_MODE_COPY;
 				break;
 
+			case 3:
+				user_opts.include_logical_slots = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						os_info.progname);
@@ -289,6 +294,8 @@ usage(void)
 	printf(_("  -V, --version                 display version information, then exit\n"));
 	printf(_("  --clone                       clone instead of copying files to new cluster\n"));
 	printf(_("  --copy                        copy files to new cluster (default)\n"));
+	printf(_("  --include-logical-replication-slots\n"
+			 "                                upgrade logical replication slots\n"));
 	printf(_("  -?, --help                    show this help, then exit\n"));
 	printf(_("\n"
 			 "Before running pg_upgrade you must:\n"
diff --git a/src/bin/pg_upgrade/pg_upgrade.c b/src/bin/pg_upgrade/pg_upgrade.c
index 4562dafcff..6dd3832422 100644
--- a/src/bin/pg_upgrade/pg_upgrade.c
+++ b/src/bin/pg_upgrade/pg_upgrade.c
@@ -59,6 +59,7 @@ static void copy_xact_xlog_xid(void);
 static void set_frozenxids(bool minmxid_only);
 static void make_outputdirs(char *pgdata);
 static void setup(char *argv0, bool *live_check);
+static void create_logical_replication_slots(void);
 
 ClusterInfo old_cluster,
 			new_cluster;
@@ -188,6 +189,19 @@ main(int argc, char **argv)
 			  new_cluster.pgdata);
 	check_ok();
 
+	/*
+	 * Create logical replication slots if requested.
+	 *
+	 * Note: This must be done after doing pg_resetwal command because
+	 * pg_resetwal would remove required WALs.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		start_postmaster(&new_cluster, true);
+		create_logical_replication_slots();
+		stop_postmaster(false);
+	}
+
 	if (user_opts.do_sync)
 	{
 		prep_status("Sync data directory to disk");
@@ -860,3 +874,50 @@ set_frozenxids(bool minmxid_only)
 
 	check_ok();
 }
+
+/*
+ * create_logical_replication_slots()
+ *
+ * Similar to create_new_objects() but only restores logical replication slots.
+ */
+static void
+create_logical_replication_slots(void)
+{
+	int			dbnum;
+
+	prep_status_progress("Restoring logical replication slots in the new cluster");
+
+	for (dbnum = 0; dbnum < old_cluster.dbarr.ndbs; dbnum++)
+	{
+		char		slots_file_name[MAXPGPATH],
+					log_file_name[MAXPGPATH];
+		DbInfo	   *old_db = &old_cluster.dbarr.dbs[dbnum];
+
+		pg_log(PG_STATUS, "%s", old_db->db_name);
+
+		snprintf(slots_file_name, sizeof(slots_file_name),
+				 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+		snprintf(log_file_name, sizeof(log_file_name),
+				 DB_DUMP_LOG_FILE_MASK, old_db->db_oid);
+
+		parallel_exec_prog(log_file_name,
+						   NULL,
+						   "\"%s/psql\" %s --echo-queries --set ON_ERROR_STOP=on "
+						   "--no-psqlrc --dbname %s -f \"%s/%s\"",
+						   new_cluster.bindir,
+						   cluster_conn_opts(&new_cluster),
+						   old_db->db_name,
+						   log_opts.dumpdir,
+						   slots_file_name);
+	}
+
+	/* reap all children */
+	while (reap_child(true) == true)
+		;
+
+	end_progress_output();
+	check_ok();
+
+	/* update new_cluster info again */
+	get_logical_slot_infos(&new_cluster);
+}
diff --git a/src/bin/pg_upgrade/pg_upgrade.h b/src/bin/pg_upgrade/pg_upgrade.h
index 3eea0139c7..8034067492 100644
--- a/src/bin/pg_upgrade/pg_upgrade.h
+++ b/src/bin/pg_upgrade/pg_upgrade.h
@@ -29,6 +29,7 @@
 /* contains both global db information and CREATE DATABASE commands */
 #define GLOBALS_DUMP_FILE	"pg_upgrade_dump_globals.sql"
 #define DB_DUMP_FILE_MASK	"pg_upgrade_dump_%u.custom"
+#define DB_DUMP_LOGICAL_SLOTS_FILE_MASK	"pg_upgrade_dump_%u_logical_slots.sql"
 
 /*
  * Base directories that include all the files generated internally, from the
@@ -150,6 +151,22 @@ typedef struct
 	int			nrels;
 } RelInfoArr;
 
+/*
+ * Structure to store logical replication slot information
+ */
+typedef struct
+{
+	char	   *slotname;		/* slot name */
+	char	   *plugin;			/* plugin */
+	bool		two_phase;		/* Can the slot decode 2PC? */
+} LogicalSlotInfo;
+
+typedef struct
+{
+	int			nslots;
+	LogicalSlotInfo *slots;
+} LogicalSlotInfoArr;
+
 /*
  * The following structure represents a relation mapping.
  */
@@ -176,6 +193,7 @@ typedef struct
 	char		db_tablespace[MAXPGPATH];	/* database default tablespace
 											 * path */
 	RelInfoArr	rel_arr;		/* array of all user relinfos */
+	LogicalSlotInfoArr slot_arr;	/* array of all LogicalSlotInfo */
 } DbInfo;
 
 /*
@@ -304,6 +322,8 @@ typedef struct
 	transferMode transfer_mode; /* copy files or link them? */
 	int			jobs;			/* number of processes/threads to use */
 	char	   *socketdir;		/* directory to use for Unix sockets */
+	bool		include_logical_slots;	/* true -> dump and restore logical
+										 * replication slots */
 } UserOpts;
 
 typedef struct
@@ -400,6 +420,7 @@ FileNameMap *gen_db_file_maps(DbInfo *old_db,
 							  DbInfo *new_db, int *nmaps, const char *old_pgdata,
 							  const char *new_pgdata);
 void		get_db_and_rel_infos(ClusterInfo *cluster);
+int			get_logical_slot_infos(ClusterInfo *cluster);
 
 /* option.c */
 
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
new file mode 100644
index 0000000000..9ca266f6b2
--- /dev/null
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -0,0 +1,146 @@
+# Copyright (c) 2023, PostgreSQL Global Development Group
+
+# Tests for upgrading replication slots
+
+use strict;
+use warnings;
+
+use File::Path qw(rmtree);
+
+use PostgreSQL::Test::Cluster;
+use PostgreSQL::Test::Utils;
+use Test::More;
+
+# Can be changed to test the other modes
+my $mode = $ENV{PG_TEST_PG_UPGRADE_MODE} || '--copy';
+
+# Initialize old node
+my $old_node = PostgreSQL::Test::Cluster->new('old_node');
+$old_node->init(allows_streaming => 'logical');
+$old_node->start;
+
+# Initialize new node
+my $new_node = PostgreSQL::Test::Cluster->new('new_node');
+$new_node->init(allows_streaming => 1);
+
+my $bindir = $new_node->config_data('--bindir');
+
+$old_node->stop;
+
+# Cause a failure at the start of pg_upgrade because wal_level is replica
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with wrong wal_level');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+# Preparations for the subsequent test. The case max_replication_slots is set
+# to 0 is prohibited.
+$new_node->append_conf('postgresql.conf', "wal_level = 'logical'");
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 0");
+
+# Cause a failure at the start of pg_upgrade because max_replication_slots is 0
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with wrong max_replication_slots');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+# Preparations for the subsequent test. max_replication_slots is set to
+# non-zero value
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 1");
+
+# Create a slot on old node, and generate WALs
+$old_node->start;
+$old_node->safe_psql(
+	'postgres', qq[
+	SELECT pg_create_logical_replication_slot('test_slot', 'test_decoding', false, true);
+	SELECT pg_create_logical_replication_slot('to_be_dropped', 'test_decoding', false, true);
+	CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a;
+]);
+
+$old_node->stop;
+
+# Cause a failure at the start of pg_upgrade because max_replication_slots is
+# smaller than existing slots on old node
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with small max_replication_slots');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+# Preparations for the subsequent test. max_replication_slots is set to
+# appropriate value
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 10");
+
+# Remove an unnecessary slot and consume WALs
+$old_node->start;
+$old_node->safe_psql(
+	'postgres', qq[
+	SELECT pg_drop_replication_slot('to_be_dropped');
+	SELECT count(*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)
+]);
+$old_node->stop;
+
+# Actual run, pg_upgrade_output.d is removed at the end
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots'
+	],
+	'run of pg_upgrade of old node');
+ok( !-d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+$new_node->start;
+my $result = $new_node->safe_psql('postgres',
+	"SELECT slot_name, two_phase FROM pg_replication_slots");
+is($result, qq(test_slot|t), 'check the slot exists on new node');
+
+done_testing();
diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list
index a1cf01e38e..141268bfd2 100644
--- a/src/tools/pgindent/typedefs.list
+++ b/src/tools/pgindent/typedefs.list
@@ -1498,7 +1498,10 @@ LogicalRepStreamAbortData
 LogicalRepTupleData
 LogicalRepTyp
 LogicalRepWorker
+LogicalReplicationSlotInfo
 LogicalRewriteMappingData
+LogicalSlotInfo
+LogicalSlotInfoArr
 LogicalTape
 LogicalTapeSet
 LsnReadQueue
-- 
2.27.0

From d4694a87b66e5b36d73cb18ba8eb7ad8cd223626 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Fri, 14 Apr 2023 08:59:03 +0000
Subject: [PATCH 3/3] pg_upgrade: Add check function for
 --include-logical-replication-slots option

XXX: Actually, this commit disallows to support slots which are created by user
backends. In the checking function we ensure that all the avtive slots have
confirmed_flush_lsn which is same as current WAL position, and they would not be
the same. For slots which are used by logical replication, logical walsenders
guarantee that at the shutdown. For individual slots, however, cannot be handled
by walsenders, so confirmed_flush_lsn is behind shutdown checkpoint record.

Author: Hayato Kuroda
Reviewed-by: Wang Wei, Vignesh C
---
 src/bin/pg_upgrade/check.c                    |  57 ++++++++
 .../t/003_logical_replication_slots.pl        | 134 +++++++++++-------
 2 files changed, 138 insertions(+), 53 deletions(-)

diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index 4b54f5567c..e9911cb302 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -31,6 +31,7 @@ static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
 static void check_for_logical_replication_slots(ClusterInfo *new_cluster);
+static void check_for_confirmed_flush_lsn(ClusterInfo *cluster);
 
 static int num_slots_on_old_cluster;
 
@@ -109,6 +110,8 @@ check_and_dump_old_cluster(bool live_check)
 	check_for_composite_data_type_usage(&old_cluster);
 	check_for_reg_data_type_usage(&old_cluster);
 	check_for_isn_and_int8_passing_mismatch(&old_cluster);
+	if (user_opts.include_logical_slots)
+		check_for_confirmed_flush_lsn(&old_cluster);
 
 	/*
 	 * PG 16 increased the size of the 'aclitem' type, which breaks the
@@ -1478,3 +1481,57 @@ check_for_logical_replication_slots(ClusterInfo *new_cluster)
 
 	check_ok();
 }
+
+/*
+ * Verify that all logical replication slots consumed all WALs, except a
+ * CHECKPOINT_SHUTDOWN record.
+ */
+static void
+check_for_confirmed_flush_lsn(ClusterInfo *cluster)
+{
+	int			i,
+				ntups,
+				i_slotname;
+	bool		is_error = false;
+	PGresult   *res;
+	DbInfo	   *active_db = &cluster->dbarr.dbs[0];
+	PGconn	   *conn = connectToServer(cluster, active_db->db_name);
+
+	Assert(user_opts.include_logical_slots);
+
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(cluster->major_version) <= 1500)
+		return;
+
+	prep_status("Checking confirmed_flush_lsn for logical replication slots");
+
+	/*
+	 * Check that all logical replication slots have reached the current WAL
+	 * position.
+	 */
+	res = executeQueryOrDie(conn,
+							"SELECT slot_name FROM pg_catalog.pg_replication_slots "
+							"WHERE (pg_catalog.pg_current_wal_insert_lsn() - confirmed_flush_lsn) <> 0 "
+							"AND temporary = false AND wal_status IN ('reserved', 'extended');");
+
+	ntups = PQntuples(res);
+	i_slotname = PQfnumber(res, "slot_name");
+
+	for (i = 0; i < ntups; i++)
+	{
+		is_error = true;
+
+		pg_log(PG_WARNING,
+			   "\nWARNING: logical replication slot \"%s\" has not consumed WALs yet",
+			   PQgetvalue(res, i, i_slotname));
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	if (is_error)
+		pg_fatal("--include-logical-replication-slots requires that all "
+				 "logical replication slots consumed all the WALs");
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
index 9ca266f6b2..87049d7116 100644
--- a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -15,132 +15,160 @@ use Test::More;
 my $mode = $ENV{PG_TEST_PG_UPGRADE_MODE} || '--copy';
 
 # Initialize old node
-my $old_node = PostgreSQL::Test::Cluster->new('old_node');
-$old_node->init(allows_streaming => 'logical');
-$old_node->start;
+my $old_publisher = PostgreSQL::Test::Cluster->new('old_publisher');
+$old_publisher->init(allows_streaming => 'logical');
 
 # Initialize new node
-my $new_node = PostgreSQL::Test::Cluster->new('new_node');
-$new_node->init(allows_streaming => 1);
+my $new_publisher = PostgreSQL::Test::Cluster->new('new_publisher');
+$new_publisher->init(allows_streaming => 1);
 
-my $bindir = $new_node->config_data('--bindir');
+# Initialize subscriber node
+my $subscriber = PostgreSQL::Test::Cluster->new('subscriber');
+$subscriber->init(allows_streaming => 'logical');
 
-$old_node->stop;
+my $bindir = $new_publisher->config_data('--bindir');
 
 # Cause a failure at the start of pg_upgrade because wal_level is replica
 command_fails(
 	[
 		'pg_upgrade', '--no-sync',
-		'-d',         $old_node->data_dir,
-		'-D',         $new_node->data_dir,
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
 		'-b',         $bindir,
 		'-B',         $bindir,
-		'-s',         $new_node->host,
-		'-p',         $old_node->port,
-		'-P',         $new_node->port,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
 		$mode,        '--include-logical-replication-slots',
 	],
 	'run of pg_upgrade of old node with wrong wal_level');
-ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+ok( -d $new_publisher->data_dir . "/pg_upgrade_output.d",
 	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
 
 # Clean up
-rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+rmtree($new_publisher->data_dir . "/pg_upgrade_output.d");
 
 # Preparations for the subsequent test. The case max_replication_slots is set
 # to 0 is prohibited.
-$new_node->append_conf('postgresql.conf', "wal_level = 'logical'");
-$new_node->append_conf('postgresql.conf', "max_replication_slots = 0");
+$new_publisher->append_conf('postgresql.conf', "wal_level = 'logical'");
+$new_publisher->append_conf('postgresql.conf', "max_replication_slots = 0");
 
 # Cause a failure at the start of pg_upgrade because max_replication_slots is 0
 command_fails(
 	[
 		'pg_upgrade', '--no-sync',
-		'-d',         $old_node->data_dir,
-		'-D',         $new_node->data_dir,
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
 		'-b',         $bindir,
 		'-B',         $bindir,
-		'-s',         $new_node->host,
-		'-p',         $old_node->port,
-		'-P',         $new_node->port,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
 		$mode,        '--include-logical-replication-slots',
 	],
 	'run of pg_upgrade of old node with wrong max_replication_slots');
-ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+ok( -d $new_publisher->data_dir . "/pg_upgrade_output.d",
 	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
 
 # Clean up
-rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+rmtree($new_publisher->data_dir . "/pg_upgrade_output.d");
 
 # Preparations for the subsequent test. max_replication_slots is set to
 # non-zero value
-$new_node->append_conf('postgresql.conf', "max_replication_slots = 1");
+$new_publisher->append_conf('postgresql.conf', "max_replication_slots = 1");
 
-# Create a slot on old node, and generate WALs
-$old_node->start;
-$old_node->safe_psql(
+# Create a slot on old node
+$old_publisher->start;
+$old_publisher->safe_psql(
 	'postgres', qq[
-	SELECT pg_create_logical_replication_slot('test_slot', 'test_decoding', false, true);
-	SELECT pg_create_logical_replication_slot('to_be_dropped', 'test_decoding', false, true);
+	SELECT pg_create_logical_replication_slot('test_slot1', 'test_decoding', false, true);
+	SELECT pg_create_logical_replication_slot('test_slot2', 'test_decoding', false, true);
 	CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a;
 ]);
 
-$old_node->stop;
+$old_publisher->stop;
 
 # Cause a failure at the start of pg_upgrade because max_replication_slots is
 # smaller than existing slots on old node
 command_fails(
 	[
 		'pg_upgrade', '--no-sync',
-		'-d',         $old_node->data_dir,
-		'-D',         $new_node->data_dir,
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
 		'-b',         $bindir,
 		'-B',         $bindir,
-		'-s',         $new_node->host,
-		'-p',         $old_node->port,
-		'-P',         $new_node->port,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
 		$mode,        '--include-logical-replication-slots',
 	],
 	'run of pg_upgrade of old node with small max_replication_slots');
-ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+ok( -d $new_publisher->data_dir . "/pg_upgrade_output.d",
 	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
 
 # Clean up
-rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+rmtree($new_publisher->data_dir . "/pg_upgrade_output.d");
 
 # Preparations for the subsequent test. max_replication_slots is set to
 # appropriate value
-$new_node->append_conf('postgresql.conf', "max_replication_slots = 10");
+$new_publisher->append_conf('postgresql.conf', "max_replication_slots = 10");
 
-# Remove an unnecessary slot and consume WALs
-$old_node->start;
-$old_node->safe_psql(
+$old_publisher->start;
+$old_publisher->safe_psql(
 	'postgres', qq[
-	SELECT pg_drop_replication_slot('to_be_dropped');
-	SELECT count(*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)
+	SELECT pg_drop_replication_slot('test_slot1');
+	SELECT pg_drop_replication_slot('test_slot2');
 ]);
-$old_node->stop;
+
+# Setup logical replication
+my $old_connstr = $old_publisher->connstr . ' dbname=postgres';
+$old_publisher->safe_psql('postgres', "CREATE PUBLICATION pub FOR ALL TABLES");
+$subscriber->start;
+$subscriber->safe_psql(
+	'postgres', qq[
+	CREATE TABLE tbl (a int);
+	CREATE SUBSCRIPTION sub CONNECTION '$old_connstr' PUBLICATION pub WITH (two_phase = 'true')
+]);
+
+# Wait for initial table sync to finish
+$subscriber->wait_for_subscription_sync($old_publisher, 'sub');
+
+$old_publisher->stop;
 
 # Actual run, pg_upgrade_output.d is removed at the end
 command_ok(
 	[
 		'pg_upgrade', '--no-sync',
-		'-d',         $old_node->data_dir,
-		'-D',         $new_node->data_dir,
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
 		'-b',         $bindir,
 		'-B',         $bindir,
-		'-s',         $new_node->host,
-		'-p',         $old_node->port,
-		'-P',         $new_node->port,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
 		$mode,        '--include-logical-replication-slots'
 	],
 	'run of pg_upgrade of old node');
-ok( !-d $new_node->data_dir . "/pg_upgrade_output.d",
+ok( !-d $new_publisher->data_dir . "/pg_upgrade_output.d",
 	"pg_upgrade_output.d/ removed after pg_upgrade success");
 
-$new_node->start;
-my $result = $new_node->safe_psql('postgres',
+$new_publisher->start;
+my $result = $new_publisher->safe_psql('postgres',
 	"SELECT slot_name, two_phase FROM pg_replication_slots");
-is($result, qq(test_slot|t), 'check the slot exists on new node');
+is($result, qq(sub|t), 'check the slot exists on new node');
+
+# Change the connection
+$subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub DISABLE");
+my $new_connstr = $new_publisher->connstr . ' dbname=postgres';
+$subscriber->safe_psql('postgres',
+	"ALTER SUBSCRIPTION sub CONNECTION '$new_connstr'");
+$subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub ENABLE");
+
+# Check whether changes on the new publisher get replicated to the subscriber
+$new_publisher->safe_psql('postgres',
+	"INSERT INTO tbl VALUES (generate_series(11, 20))");
+$new_publisher->wait_for_catchup('sub');
+$result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(20), 'check changes are shipped to subscriber');
 
 done_testing();
-- 
2.27.0

From 49360788fea01756aa9578da3dad34786bd1281a Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Fri, 14 Apr 2023 08:59:03 +0000
Subject: [PATCH 4/4] pg_upgrade: Add check function for
 --include-logical-replication-slots option

XXX: Actually, this commit disallows to support slots which are created by user
backends. In the checking function we ensure that all the avtive slots have
confirmed_flush_lsn which is same as current WAL position, and they would not be
the same. For slots which are used by logical replication, logical walsenders
guarantee that at the shutdown. For individual slots, however, cannot be handled
by walsenders, so confirmed_flush_lsn is behind shutdown checkpoint record.

Author: Hayato Kuroda
Reviewed-by: Wang Wei, Vignesh C
---
 src/bin/pg_upgrade/check.c                    |  59 ++++++++
 .../t/003_logical_replication_slots.pl        | 134 +++++++++++-------
 2 files changed, 140 insertions(+), 53 deletions(-)

diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index 4b54f5567c..264895721a 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -31,6 +31,7 @@ static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
 static void check_for_logical_replication_slots(ClusterInfo *new_cluster);
+static void check_for_confirmed_flush_lsn(ClusterInfo *cluster);
 
 static int num_slots_on_old_cluster;
 
@@ -109,6 +110,8 @@ check_and_dump_old_cluster(bool live_check)
 	check_for_composite_data_type_usage(&old_cluster);
 	check_for_reg_data_type_usage(&old_cluster);
 	check_for_isn_and_int8_passing_mismatch(&old_cluster);
+	if (user_opts.include_logical_slots)
+		check_for_confirmed_flush_lsn(&old_cluster);
 
 	/*
 	 * PG 16 increased the size of the 'aclitem' type, which breaks the
@@ -1478,3 +1481,59 @@ check_for_logical_replication_slots(ClusterInfo *new_cluster)
 
 	check_ok();
 }
+
+/*
+ * Verify that all logical replication slots consumed all WALs, except a
+ * CHECKPOINT_SHUTDOWN record.
+ */
+static void
+check_for_confirmed_flush_lsn(ClusterInfo *cluster)
+{
+	int			i,
+				ntups,
+				i_slotname;
+	bool		is_error = false;
+	PGresult   *res;
+	DbInfo	   *active_db = &cluster->dbarr.dbs[0];
+	PGconn	   *conn = connectToServer(cluster, active_db->db_name);
+
+	Assert(user_opts.include_logical_slots);
+
+	/* --include-logical-replication-slots can be used since PG16. */
+	if (GET_MAJOR_VERSION(cluster->major_version) <= 1500)
+		return;
+
+	prep_status("Checking confirmed_flush_lsn for logical replication slots");
+
+	/*
+	 * Check that all logical replication slots have reached the current WAL
+	 * position.
+	 */
+	res = executeQueryOrDie(conn,
+							"SELECT slot_name FROM pg_catalog.pg_replication_slots "
+							"WHERE (SELECT count(record_type) "
+							"		FROM pg_catalog.pg_get_wal_records_content(confirmed_flush_lsn, pg_catalog.pg_current_wal_insert_lsn()) "
+							"		WHERE record_type != 'CHECKPOINT_SHUTDOWN') <> 0 "
+							"AND temporary = false AND wal_status IN ('reserved', 'extended');");
+
+	ntups = PQntuples(res);
+	i_slotname = PQfnumber(res, "slot_name");
+
+	for (i = 0; i < ntups; i++)
+	{
+		is_error = true;
+
+		pg_log(PG_WARNING,
+			   "\nWARNING: logical replication slot \"%s\" has not consumed WALs yet",
+			   PQgetvalue(res, i, i_slotname));
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	if (is_error)
+		pg_fatal("--include-logical-replication-slots requires that all "
+				 "logical replication slots consumed all the WALs");
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
index 9ca266f6b2..87049d7116 100644
--- a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -15,132 +15,160 @@ use Test::More;
 my $mode = $ENV{PG_TEST_PG_UPGRADE_MODE} || '--copy';
 
 # Initialize old node
-my $old_node = PostgreSQL::Test::Cluster->new('old_node');
-$old_node->init(allows_streaming => 'logical');
-$old_node->start;
+my $old_publisher = PostgreSQL::Test::Cluster->new('old_publisher');
+$old_publisher->init(allows_streaming => 'logical');
 
 # Initialize new node
-my $new_node = PostgreSQL::Test::Cluster->new('new_node');
-$new_node->init(allows_streaming => 1);
+my $new_publisher = PostgreSQL::Test::Cluster->new('new_publisher');
+$new_publisher->init(allows_streaming => 1);
 
-my $bindir = $new_node->config_data('--bindir');
+# Initialize subscriber node
+my $subscriber = PostgreSQL::Test::Cluster->new('subscriber');
+$subscriber->init(allows_streaming => 'logical');
 
-$old_node->stop;
+my $bindir = $new_publisher->config_data('--bindir');
 
 # Cause a failure at the start of pg_upgrade because wal_level is replica
 command_fails(
 	[
 		'pg_upgrade', '--no-sync',
-		'-d',         $old_node->data_dir,
-		'-D',         $new_node->data_dir,
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
 		'-b',         $bindir,
 		'-B',         $bindir,
-		'-s',         $new_node->host,
-		'-p',         $old_node->port,
-		'-P',         $new_node->port,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
 		$mode,        '--include-logical-replication-slots',
 	],
 	'run of pg_upgrade of old node with wrong wal_level');
-ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+ok( -d $new_publisher->data_dir . "/pg_upgrade_output.d",
 	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
 
 # Clean up
-rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+rmtree($new_publisher->data_dir . "/pg_upgrade_output.d");
 
 # Preparations for the subsequent test. The case max_replication_slots is set
 # to 0 is prohibited.
-$new_node->append_conf('postgresql.conf', "wal_level = 'logical'");
-$new_node->append_conf('postgresql.conf', "max_replication_slots = 0");
+$new_publisher->append_conf('postgresql.conf', "wal_level = 'logical'");
+$new_publisher->append_conf('postgresql.conf', "max_replication_slots = 0");
 
 # Cause a failure at the start of pg_upgrade because max_replication_slots is 0
 command_fails(
 	[
 		'pg_upgrade', '--no-sync',
-		'-d',         $old_node->data_dir,
-		'-D',         $new_node->data_dir,
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
 		'-b',         $bindir,
 		'-B',         $bindir,
-		'-s',         $new_node->host,
-		'-p',         $old_node->port,
-		'-P',         $new_node->port,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
 		$mode,        '--include-logical-replication-slots',
 	],
 	'run of pg_upgrade of old node with wrong max_replication_slots');
-ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+ok( -d $new_publisher->data_dir . "/pg_upgrade_output.d",
 	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
 
 # Clean up
-rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+rmtree($new_publisher->data_dir . "/pg_upgrade_output.d");
 
 # Preparations for the subsequent test. max_replication_slots is set to
 # non-zero value
-$new_node->append_conf('postgresql.conf', "max_replication_slots = 1");
+$new_publisher->append_conf('postgresql.conf', "max_replication_slots = 1");
 
-# Create a slot on old node, and generate WALs
-$old_node->start;
-$old_node->safe_psql(
+# Create a slot on old node
+$old_publisher->start;
+$old_publisher->safe_psql(
 	'postgres', qq[
-	SELECT pg_create_logical_replication_slot('test_slot', 'test_decoding', false, true);
-	SELECT pg_create_logical_replication_slot('to_be_dropped', 'test_decoding', false, true);
+	SELECT pg_create_logical_replication_slot('test_slot1', 'test_decoding', false, true);
+	SELECT pg_create_logical_replication_slot('test_slot2', 'test_decoding', false, true);
 	CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a;
 ]);
 
-$old_node->stop;
+$old_publisher->stop;
 
 # Cause a failure at the start of pg_upgrade because max_replication_slots is
 # smaller than existing slots on old node
 command_fails(
 	[
 		'pg_upgrade', '--no-sync',
-		'-d',         $old_node->data_dir,
-		'-D',         $new_node->data_dir,
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
 		'-b',         $bindir,
 		'-B',         $bindir,
-		'-s',         $new_node->host,
-		'-p',         $old_node->port,
-		'-P',         $new_node->port,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
 		$mode,        '--include-logical-replication-slots',
 	],
 	'run of pg_upgrade of old node with small max_replication_slots');
-ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+ok( -d $new_publisher->data_dir . "/pg_upgrade_output.d",
 	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
 
 # Clean up
-rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+rmtree($new_publisher->data_dir . "/pg_upgrade_output.d");
 
 # Preparations for the subsequent test. max_replication_slots is set to
 # appropriate value
-$new_node->append_conf('postgresql.conf', "max_replication_slots = 10");
+$new_publisher->append_conf('postgresql.conf', "max_replication_slots = 10");
 
-# Remove an unnecessary slot and consume WALs
-$old_node->start;
-$old_node->safe_psql(
+$old_publisher->start;
+$old_publisher->safe_psql(
 	'postgres', qq[
-	SELECT pg_drop_replication_slot('to_be_dropped');
-	SELECT count(*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)
+	SELECT pg_drop_replication_slot('test_slot1');
+	SELECT pg_drop_replication_slot('test_slot2');
 ]);
-$old_node->stop;
+
+# Setup logical replication
+my $old_connstr = $old_publisher->connstr . ' dbname=postgres';
+$old_publisher->safe_psql('postgres', "CREATE PUBLICATION pub FOR ALL TABLES");
+$subscriber->start;
+$subscriber->safe_psql(
+	'postgres', qq[
+	CREATE TABLE tbl (a int);
+	CREATE SUBSCRIPTION sub CONNECTION '$old_connstr' PUBLICATION pub WITH (two_phase = 'true')
+]);
+
+# Wait for initial table sync to finish
+$subscriber->wait_for_subscription_sync($old_publisher, 'sub');
+
+$old_publisher->stop;
 
 # Actual run, pg_upgrade_output.d is removed at the end
 command_ok(
 	[
 		'pg_upgrade', '--no-sync',
-		'-d',         $old_node->data_dir,
-		'-D',         $new_node->data_dir,
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
 		'-b',         $bindir,
 		'-B',         $bindir,
-		'-s',         $new_node->host,
-		'-p',         $old_node->port,
-		'-P',         $new_node->port,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
 		$mode,        '--include-logical-replication-slots'
 	],
 	'run of pg_upgrade of old node');
-ok( !-d $new_node->data_dir . "/pg_upgrade_output.d",
+ok( !-d $new_publisher->data_dir . "/pg_upgrade_output.d",
 	"pg_upgrade_output.d/ removed after pg_upgrade success");
 
-$new_node->start;
-my $result = $new_node->safe_psql('postgres',
+$new_publisher->start;
+my $result = $new_publisher->safe_psql('postgres',
 	"SELECT slot_name, two_phase FROM pg_replication_slots");
-is($result, qq(test_slot|t), 'check the slot exists on new node');
+is($result, qq(sub|t), 'check the slot exists on new node');
+
+# Change the connection
+$subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub DISABLE");
+my $new_connstr = $new_publisher->connstr . ' dbname=postgres';
+$subscriber->safe_psql('postgres',
+	"ALTER SUBSCRIPTION sub CONNECTION '$new_connstr'");
+$subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub ENABLE");
+
+# Check whether changes on the new publisher get replicated to the subscriber
+$new_publisher->safe_psql('postgres',
+	"INSERT INTO tbl VALUES (generate_series(11, 20))");
+$new_publisher->wait_for_catchup('sub');
+$result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(20), 'check changes are shipped to subscriber');
 
 done_testing();
-- 
2.34.1

From 8cb9043ece78ba5235b11b7277b608a37044c559 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Tue, 4 Apr 2023 05:49:34 +0000
Subject: [PATCH 2/3] pg_upgrade: Add --include-logical-replication-slots
 option

This commit introduces a new pg_upgrade option called "--include-logical-replication-slots".
This allows nodes with logical replication slots to be upgraded. The commit can
be divided into two parts: one for pg_dump and another for pg_upgrade.

For pg_dump this commit includes a new option called "--logical-replication-slots-only".
This option can be used to dump logical replication slots. When this option is
specified, the slot_name, plugin, and two_phase parameters are extracted from
pg_replication_slots. An SQL file that executes pg_create_logical_replication_slot()
with the extracted parameters is generated.

For pg_upgrade, when '--include-logical-replication-slots' is specified, it executes
pg_dump with the new "--logical-replication-slots-only" option and restores the slots
using the pg_create_logical_replication_slots() statements that the dump
generated (see above). Note that we cannot dump replication slots at the same time
as the schema dump because we need to separate the timing of restoring replication
slots and other objects. Replication slots, in  particular, should not be restored
before executing the pg_resetwal command because it will remove WALs that are
required by the slots.

The significant advantage of this commit is that it makes it easy to continue
logical replication even after upgrading the publisher node. Previously, pg_upgrade
allowed copying publications to a new node. With this new commit, adjusting the
connection string to the new publisher will cause the apply worker on the subscriber
to connect to the new publisher automatically. This enables seamless continuation
of logical replication, even after an upgrade.

Author: Hayato Kuroda
Reviewed-by: Peter Smith, Julien Rouhaud, Vignesh C, Wang Wei
---
 doc/src/sgml/ref/pgupgrade.sgml               |  11 ++
 src/bin/pg_dump/pg_backup.h                   |   1 +
 src/bin/pg_dump/pg_dump.c                     | 155 ++++++++++++++++++
 src/bin/pg_dump/pg_dump.h                     |  14 +-
 src/bin/pg_dump/pg_dump_sort.c                |  11 +-
 src/bin/pg_upgrade/Makefile                   |   3 +
 src/bin/pg_upgrade/check.c                    |  76 +++++++++
 src/bin/pg_upgrade/dump.c                     |  24 +++
 src/bin/pg_upgrade/info.c                     | 111 ++++++++++++-
 src/bin/pg_upgrade/meson.build                |   1 +
 src/bin/pg_upgrade/option.c                   |   7 +
 src/bin/pg_upgrade/pg_upgrade.c               |  61 +++++++
 src/bin/pg_upgrade/pg_upgrade.h               |  21 +++
 .../t/003_logical_replication_slots.pl        | 146 +++++++++++++++++
 src/tools/pgindent/typedefs.list              |   3 +
 15 files changed, 641 insertions(+), 4 deletions(-)
 create mode 100644 src/bin/pg_upgrade/t/003_logical_replication_slots.pl

diff --git a/doc/src/sgml/ref/pgupgrade.sgml b/doc/src/sgml/ref/pgupgrade.sgml
index 7816b4c685..94e90ff506 100644
--- a/doc/src/sgml/ref/pgupgrade.sgml
+++ b/doc/src/sgml/ref/pgupgrade.sgml
@@ -240,6 +240,17 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--include-logical-replication-slots</option></term>
+      <listitem>
+       <para>
+        Upgrade logical replication slots. Only permanent replication slots
+        are included. Note that pg_upgrade does not check the installation of
+        plugins.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
       <term><option>-?</option></term>
       <term><option>--help</option></term>
diff --git a/src/bin/pg_dump/pg_backup.h b/src/bin/pg_dump/pg_backup.h
index aba780ef4b..0a4e931f9b 100644
--- a/src/bin/pg_dump/pg_backup.h
+++ b/src/bin/pg_dump/pg_backup.h
@@ -187,6 +187,7 @@ typedef struct _dumpOptions
 	int			use_setsessauth;
 	int			enable_row_security;
 	int			load_via_partition_root;
+	int			logical_slots_only;
 
 	/* default, if no "inclusion" switches appear, is to dump everything */
 	bool		include_everything;
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index 5dab1ba9ea..12d4066d3b 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -328,6 +328,9 @@ static void setupDumpWorker(Archive *AH);
 static TableInfo *getRootTableInfo(const TableInfo *tbinfo);
 static bool forcePartitionRootLoad(const TableInfo *tbinfo);
 
+static void getLogicalReplicationSlots(Archive *fout);
+static void dumpLogicalReplicationSlot(Archive *fout,
+									   const LogicalReplicationSlotInfo *slotinfo);
 
 int
 main(int argc, char **argv)
@@ -431,6 +434,7 @@ main(int argc, char **argv)
 		{"table-and-children", required_argument, NULL, 12},
 		{"exclude-table-and-children", required_argument, NULL, 13},
 		{"exclude-table-data-and-children", required_argument, NULL, 14},
+		{"logical-replication-slots-only", no_argument, NULL, 15},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -657,6 +661,10 @@ main(int argc, char **argv)
 										  optarg);
 				break;
 
+			case 15:			/* dump only replication slot(s) */
+				dopt.logical_slots_only = true;
+				break;
+
 			default:
 				/* getopt_long already emitted a complaint */
 				pg_log_error_hint("Try \"%s --help\" for more information.", progname);
@@ -714,6 +722,21 @@ main(int argc, char **argv)
 	if (dopt.do_nothing && dopt.dump_inserts == 0)
 		pg_fatal("option --on-conflict-do-nothing requires option --inserts, --rows-per-insert, or --column-inserts");
 
+	if (dopt.logical_slots_only)
+	{
+		if (!dopt.binary_upgrade)
+			pg_fatal("options --logical-replication-slots-only requires option --binary-upgrade");
+
+		if (dopt.dataOnly)
+			pg_fatal("options --logical-replication-slots-only and -a/--data-only cannot be used together");
+
+		if (dopt.schemaOnly)
+			pg_fatal("options --logical-replication-slots-only and -s/--schema-only cannot be used together");
+
+		if (dopt.outputClean)
+			pg_fatal("options --logical-replication-slots-only and -c/--clean cannot be used together");
+	}
+
 	/* Identify archive format to emit */
 	archiveFormat = parseArchiveFormat(format, &archiveMode);
 
@@ -876,6 +899,16 @@ main(int argc, char **argv)
 			pg_fatal("no matching extensions were found");
 	}
 
+	/*
+	 * If dump logical-replication-slots-only was requested, dump only them
+	 * and skip everything else.
+	 */
+	if (dopt.logical_slots_only)
+	{
+		getLogicalReplicationSlots(fout);
+		goto dump;
+	}
+
 	/*
 	 * Dumping LOs is the default for dumps where an inclusion switch is not
 	 * used (an "include everything" dump).  -B can be used to exclude LOs
@@ -936,6 +969,8 @@ main(int argc, char **argv)
 	if (!dopt.no_security_labels)
 		collectSecLabels(fout);
 
+dump:
+
 	/* Lastly, create dummy objects to represent the section boundaries */
 	boundaryObjs = createBoundaryObjects();
 
@@ -1109,6 +1144,12 @@ help(const char *progname)
 			 "                               servers matching PATTERN\n"));
 	printf(_("  --inserts                    dump data as INSERT commands, rather than COPY\n"));
 	printf(_("  --load-via-partition-root    load partitions via the root table\n"));
+
+	/*
+	 * The option --logical-replication-slots-only is used only by pg_upgrade
+	 * and should not be called by users, which is why it is not exposed by the
+	 * help.
+	 */
 	printf(_("  --no-comments                do not dump comments\n"));
 	printf(_("  --no-publications            do not dump publications\n"));
 	printf(_("  --no-security-labels         do not dump security label assignments\n"));
@@ -10237,6 +10278,10 @@ dumpDumpableObject(Archive *fout, DumpableObject *dobj)
 		case DO_SUBSCRIPTION:
 			dumpSubscription(fout, (const SubscriptionInfo *) dobj);
 			break;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			dumpLogicalReplicationSlot(fout,
+									   (const LogicalReplicationSlotInfo *) dobj);
+			break;
 		case DO_PRE_DATA_BOUNDARY:
 		case DO_POST_DATA_BOUNDARY:
 			/* never dumped, nothing to do */
@@ -18218,6 +18263,7 @@ addBoundaryDependencies(DumpableObject **dobjs, int numObjs,
 			case DO_PUBLICATION_REL:
 			case DO_PUBLICATION_TABLE_IN_SCHEMA:
 			case DO_SUBSCRIPTION:
+			case DO_LOGICAL_REPLICATION_SLOT:
 				/* Post-data objects: must come after the post-data boundary */
 				addObjectDependency(dobj, postDataBound->dumpId);
 				break;
@@ -18479,3 +18525,112 @@ appendReloptionsArrayAH(PQExpBuffer buffer, const char *reloptions,
 	if (!res)
 		pg_log_warning("could not parse %s array", "reloptions");
 }
+
+/*
+ * getLogicalReplicationSlots
+ *	  get information about replication slots
+ */
+static void
+getLogicalReplicationSlots(Archive *fout)
+{
+	PGresult   *res;
+	LogicalReplicationSlotInfo *slotinfo;
+	PQExpBuffer query;
+
+	int			i_slotname;
+	int			i_plugin;
+	int			i_twophase;
+	int			i,
+				ntups;
+
+	/* Check whether we should dump or not */
+	if (fout->remoteVersion < 170000)
+		return;
+
+	Assert(fout->dopt->logical_slots_only);
+
+	query = createPQExpBuffer();
+
+	resetPQExpBuffer(query);
+
+	/*
+	 * Get replication slots.
+	 *
+	 * XXX: Which information must be extracted from old node? Currently three
+	 * attributes are extracted because they are used by
+	 * pg_create_logical_replication_slot().
+	 */
+	appendPQExpBufferStr(query,
+						 "SELECT slot_name, plugin, two_phase "
+						 "FROM pg_catalog.pg_replication_slots "
+						 "WHERE database = current_database() AND temporary = false "
+						 "AND wal_status IN ('reserved', 'extended');");
+
+	res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
+
+	ntups = PQntuples(res);
+
+	i_slotname = PQfnumber(res, "slot_name");
+	i_plugin = PQfnumber(res, "plugin");
+	i_twophase = PQfnumber(res, "two_phase");
+
+	slotinfo = pg_malloc(ntups * sizeof(LogicalReplicationSlotInfo));
+
+	for (i = 0; i < ntups; i++)
+	{
+		slotinfo[i].dobj.objType = DO_LOGICAL_REPLICATION_SLOT;
+
+		slotinfo[i].dobj.catId.tableoid = InvalidOid;
+		slotinfo[i].dobj.catId.oid = InvalidOid;
+		AssignDumpId(&slotinfo[i].dobj);
+
+		slotinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_slotname));
+
+		slotinfo[i].plugin = pg_strdup(PQgetvalue(res, i, i_plugin));
+		slotinfo[i].twophase = (strcmp(PQgetvalue(res, i, i_twophase), "t") == 0);
+
+		/*
+		 * Note: Currently we do not have any options to include/exclude slots
+		 * in dumping, so all the slots must be selected.
+		 */
+		slotinfo[i].dobj.dump = DUMP_COMPONENT_DEFINITION;
+	}
+	PQclear(res);
+
+	destroyPQExpBuffer(query);
+}
+
+/*
+ * dumpLogicalReplicationSlot
+ *	  dump creation functions for the given logical replication slots
+ */
+static void
+dumpLogicalReplicationSlot(Archive *fout,
+						   const LogicalReplicationSlotInfo *slotinfo)
+{
+	Assert(fout->dopt->logical_slots_only);
+
+	if (slotinfo->dobj.dump & DUMP_COMPONENT_DEFINITION)
+	{
+		PQExpBuffer query = createPQExpBuffer();
+
+		/*
+		 * XXX: For simplification, pg_create_logical_replication_slot() is
+		 * used. Is it sufficient?
+		 */
+		appendPQExpBuffer(query, "SELECT pg_catalog.pg_create_logical_replication_slot(");
+		appendStringLiteralAH(query, slotinfo->dobj.name, fout);
+		appendPQExpBuffer(query, ", ");
+		appendStringLiteralAH(query, slotinfo->plugin, fout);
+		appendPQExpBuffer(query, ", false, %s);",
+						  slotinfo->twophase ? "true" : "false");
+
+		ArchiveEntry(fout, slotinfo->dobj.catId, slotinfo->dobj.dumpId,
+					 ARCHIVE_OPTS(.tag = slotinfo->dobj.name,
+								  .description = "REPLICATION SLOT",
+								  .section = SECTION_POST_DATA,
+								  .createStmt = query->data));
+
+		destroyPQExpBuffer(query);
+	}
+}
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index bc8f2ec36d..ed1866d9ab 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -82,7 +82,8 @@ typedef enum
 	DO_PUBLICATION,
 	DO_PUBLICATION_REL,
 	DO_PUBLICATION_TABLE_IN_SCHEMA,
-	DO_SUBSCRIPTION
+	DO_SUBSCRIPTION,
+	DO_LOGICAL_REPLICATION_SLOT
 } DumpableObjectType;
 
 /*
@@ -667,6 +668,17 @@ typedef struct _SubscriptionInfo
 	char	   *subpasswordrequired;
 } SubscriptionInfo;
 
+/*
+ * The LogicalReplicationSlotInfo struct is used to represent replication
+ * slots.
+ */
+typedef struct _LogicalReplicationSlotInfo
+{
+	DumpableObject dobj;
+	char	   *plugin;
+	bool		twophase;
+} LogicalReplicationSlotInfo;
+
 /*
  *	common utility functions
  */
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index 523a19c155..ae65443228 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -93,6 +93,7 @@ enum dbObjectTypePriorities
 	PRIO_PUBLICATION_REL,
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,
 	PRIO_SUBSCRIPTION,
+	PRIO_LOGICAL_REPLICATION_SLOT,
 	PRIO_DEFAULT_ACL,			/* done in ACL pass */
 	PRIO_EVENT_TRIGGER,			/* must be next to last! */
 	PRIO_REFRESH_MATVIEW		/* must be last! */
@@ -146,10 +147,11 @@ static const int dbObjectTypePriority[] =
 	PRIO_PUBLICATION,			/* DO_PUBLICATION */
 	PRIO_PUBLICATION_REL,		/* DO_PUBLICATION_REL */
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,	/* DO_PUBLICATION_TABLE_IN_SCHEMA */
-	PRIO_SUBSCRIPTION			/* DO_SUBSCRIPTION */
+	PRIO_SUBSCRIPTION,			/* DO_SUBSCRIPTION */
+	PRIO_LOGICAL_REPLICATION_SLOT	/* DO_LOGICAL_REPLICATION_SLOT */
 };
 
-StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_SUBSCRIPTION + 1),
+StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_LOGICAL_REPLICATION_SLOT + 1),
 				 "array length mismatch");
 
 static DumpId preDataBoundId;
@@ -1542,6 +1544,11 @@ describeDumpableObject(DumpableObject *obj, char *buf, int bufsize)
 					 "SUBSCRIPTION (ID %d OID %u)",
 					 obj->dumpId, obj->catId.oid);
 			return;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			snprintf(buf, bufsize,
+					 "LOGICAL REPLICATION SLOT (ID %d NAME %s)",
+					 obj->dumpId, obj->name);
+			return;
 		case DO_PRE_DATA_BOUNDARY:
 			snprintf(buf, bufsize,
 					 "PRE-DATA BOUNDARY  (ID %d)",
diff --git a/src/bin/pg_upgrade/Makefile b/src/bin/pg_upgrade/Makefile
index 5834513add..815d1a7ca1 100644
--- a/src/bin/pg_upgrade/Makefile
+++ b/src/bin/pg_upgrade/Makefile
@@ -3,6 +3,9 @@
 PGFILEDESC = "pg_upgrade - an in-place binary upgrade utility"
 PGAPPICON = win32
 
+# required for 003_logical_replication_slots.pl
+EXTRA_INSTALL=contrib/test_decoding
+
 subdir = src/bin/pg_upgrade
 top_builddir = ../../..
 include $(top_builddir)/src/Makefile.global
diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index 64024e3b9e..4b54f5567c 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -30,7 +30,9 @@ static void check_for_jsonb_9_4_usage(ClusterInfo *cluster);
 static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
+static void check_for_logical_replication_slots(ClusterInfo *new_cluster);
 
+static int num_slots_on_old_cluster;
 
 /*
  * fix_path_separator
@@ -89,6 +91,10 @@ check_and_dump_old_cluster(bool live_check)
 	/* Extract a list of databases and tables from the old cluster */
 	get_db_and_rel_infos(&old_cluster);
 
+	/* Additionally, extract a list of logical replication slots if required */
+	if (user_opts.include_logical_slots)
+		num_slots_on_old_cluster = get_logical_slot_infos(&old_cluster);
+
 	init_tablespaces();
 
 	get_loadable_libraries();
@@ -189,6 +195,17 @@ check_new_cluster(void)
 {
 	get_db_and_rel_infos(&new_cluster);
 
+	/*
+	 * Do additional work if --include-logical-replication-slots was specified.
+	 * This must be done before check_new_cluster_is_empty() because the
+	 * slot_arr attribute of the new_cluster will be checked in that function.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		(void) get_logical_slot_infos(&new_cluster);
+		check_for_logical_replication_slots(&new_cluster);
+	}
+
 	check_new_cluster_is_empty();
 
 	check_loadable_libraries();
@@ -364,6 +381,22 @@ check_new_cluster_is_empty(void)
 						 rel_arr->rels[relnum].nspname,
 						 rel_arr->rels[relnum].relname);
 		}
+
+		/*
+		 * If --include-logical-replication-slots is required, check the
+		 * existence of slots.
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			DbInfo *pDbInfo = &new_cluster.dbarr.dbs[dbnum];
+			LogicalSlotInfoArr *slot_arr = &pDbInfo->slot_arr;
+
+			/* if nslots > 0, report just first entry and exit */
+			if (slot_arr->nslots)
+				pg_fatal("New cluster database \"%s\" is not empty: found logical replication slot \"%s\"",
+						 pDbInfo->db_name,
+						 slot_arr->slots[0].slotname);
+		}
 	}
 }
 
@@ -1402,3 +1435,46 @@ check_for_user_defined_encoding_conversions(ClusterInfo *cluster)
 	else
 		check_ok();
 }
+
+/*
+ * Verify the parameter settings necessary for creating logical replication
+ * slots.
+ */
+static void
+check_for_logical_replication_slots(ClusterInfo *new_cluster)
+{
+	PGresult   *res;
+	PGconn	   *conn = connectToServer(new_cluster, "template1");
+	int			max_replication_slots;
+	char	   *wal_level;
+
+	/* --include-logical-replication-slots can be used since PG17. */
+	if (GET_MAJOR_VERSION(new_cluster->major_version) <= 1600)
+		return;
+
+	prep_status("Checking parameter settings for logical replication slots");
+
+	res = executeQueryOrDie(conn, "SHOW max_replication_slots;");
+	max_replication_slots = atoi(PQgetvalue(res, 0, 0));
+
+	if (max_replication_slots == 0)
+		pg_fatal("max_replication_slots must be greater than 0");
+	else if (num_slots_on_old_cluster > max_replication_slots)
+		pg_fatal("max_replication_slots must be greater than existing logical "
+				 "replication slots on old node.");
+
+	PQclear(res);
+
+	res = executeQueryOrDie(conn, "SHOW wal_level;");
+	wal_level = PQgetvalue(res, 0, 0);
+
+	if (strcmp(wal_level, "logical") != 0)
+		pg_fatal("wal_level must be \"logical\", but is set to \"%s\"",
+				 wal_level);
+
+	PQclear(res);
+
+	PQfinish(conn);
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/dump.c b/src/bin/pg_upgrade/dump.c
index 6c8c82dca8..e6b90864f5 100644
--- a/src/bin/pg_upgrade/dump.c
+++ b/src/bin/pg_upgrade/dump.c
@@ -59,6 +59,30 @@ generate_old_dump(void)
 						   log_opts.dumpdir,
 						   sql_file_name, escaped_connstr.data);
 
+		/*
+		 * Dump logical replication slots if needed.
+		 *
+		 * XXX We cannot dump replication slots at the same time as the schema
+		 * dump because we need to separate the timing of restoring
+		 * replication slots and other objects. Replication slots, in
+		 * particular, should not be restored before executing the pg_resetwal
+		 * command because it will remove WALs that are required by the slots.
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			char		slots_file_name[MAXPGPATH];
+
+			snprintf(slots_file_name, sizeof(slots_file_name),
+					 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+			parallel_exec_prog(log_file_name, NULL,
+							   "\"%s/pg_dump\" %s --logical-replication-slots-only "
+							   "--quote-all-identifiers --binary-upgrade %s "
+							   "--file=\"%s/%s\" %s",
+							   new_cluster.bindir, cluster_conn_opts(&old_cluster),
+							   log_opts.verbose ? "--verbose" : "",
+							   log_opts.dumpdir,
+							   slots_file_name, escaped_connstr.data);
+		}
 		termPQExpBuffer(&escaped_connstr);
 	}
 
diff --git a/src/bin/pg_upgrade/info.c b/src/bin/pg_upgrade/info.c
index a9988abfe1..8bc0ad2e10 100644
--- a/src/bin/pg_upgrade/info.c
+++ b/src/bin/pg_upgrade/info.c
@@ -26,6 +26,7 @@ static void get_rel_infos(ClusterInfo *cluster, DbInfo *dbinfo);
 static void free_rel_infos(RelInfoArr *rel_arr);
 static void print_db_infos(DbInfoArr *db_arr);
 static void print_rel_infos(RelInfoArr *rel_arr);
+static void print_slot_infos(LogicalSlotInfoArr *slot_arr);
 
 
 /*
@@ -394,7 +395,7 @@ get_db_infos(ClusterInfo *cluster)
 	i_spclocation = PQfnumber(res, "spclocation");
 
 	ntups = PQntuples(res);
-	dbinfos = (DbInfo *) pg_malloc(sizeof(DbInfo) * ntups);
+	dbinfos = (DbInfo *) pg_malloc0(sizeof(DbInfo) * ntups);
 
 	for (tupnum = 0; tupnum < ntups; tupnum++)
 	{
@@ -600,6 +601,96 @@ get_rel_infos(ClusterInfo *cluster, DbInfo *dbinfo)
 	dbinfo->rel_arr.nrels = num_rels;
 }
 
+/*
+ * get_logical_slot_infos_per_db()
+ *
+ * gets the LogicalSlotInfos for all the logical replication slots of the database
+ * referred to by "dbinfo".
+ */
+static void
+get_logical_slot_infos_per_db(ClusterInfo *cluster, DbInfo *dbinfo)
+{
+	PGconn	   *conn = connectToServer(cluster,
+									   dbinfo->db_name);
+	PGresult   *res;
+	LogicalSlotInfo *slotinfos = NULL;
+
+	int			num_slots;
+
+	char		query[QUERY_ALLOC];
+
+	snprintf(query, sizeof(query),
+			 "SELECT slot_name, plugin, two_phase "
+			 "FROM pg_catalog.pg_replication_slots "
+			 "WHERE database = current_database() AND temporary = false "
+			 "AND wal_status IN ('reserved', 'extended');");
+
+	res = executeQueryOrDie(conn, "%s", query);
+
+	num_slots = PQntuples(res);
+
+	if (num_slots)
+	{
+		int			slotnum;
+		int			i_slotname;
+		int			i_plugin;
+		int			i_twophase;
+
+		slotinfos = (LogicalSlotInfo *) pg_malloc(sizeof(LogicalSlotInfo) * num_slots);
+
+		i_slotname = PQfnumber(res, "slot_name");
+		i_plugin = PQfnumber(res, "plugin");
+		i_twophase = PQfnumber(res, "two_phase");
+
+		for (slotnum = 0; slotnum < num_slots; slotnum++)
+		{
+			LogicalSlotInfo *curr = &slotinfos[slotnum];
+
+			curr->slotname = pg_strdup(PQgetvalue(res, slotnum, i_slotname));
+			curr->plugin = pg_strdup(PQgetvalue(res, slotnum, i_plugin));
+			curr->two_phase = (strcmp(PQgetvalue(res, slotnum, i_twophase), "t") == 0);
+		}
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	dbinfo->slot_arr.slots = slotinfos;
+	dbinfo->slot_arr.nslots = num_slots;
+}
+
+/*
+ * get_logical_slot_infos()
+ *
+ * Higher level routine to generate LogicalSlotInfoArr for all databases.
+ */
+int
+get_logical_slot_infos(ClusterInfo *cluster)
+{
+	int			dbnum;
+	int			slotnum = 0;
+
+	if (cluster == &old_cluster)
+		pg_log(PG_VERBOSE, "\nsource databases:");
+	else
+		pg_log(PG_VERBOSE, "\ntarget databases:");
+
+	for (dbnum = 0; dbnum < cluster->dbarr.ndbs; dbnum++)
+	{
+		DbInfo	   *pDbInfo = &cluster->dbarr.dbs[dbnum];
+
+		get_logical_slot_infos_per_db(cluster, pDbInfo);
+		slotnum += pDbInfo->slot_arr.nslots;
+
+		if (log_opts.verbose)
+		{
+			pg_log(PG_VERBOSE, "Database: \"%s\"", pDbInfo->db_name);
+			print_slot_infos(&pDbInfo->slot_arr);
+		}
+	}
+
+	return slotnum;
+}
 
 static void
 free_db_and_rel_infos(DbInfoArr *db_arr)
@@ -610,6 +701,12 @@ free_db_and_rel_infos(DbInfoArr *db_arr)
 	{
 		free_rel_infos(&db_arr->dbs[dbnum].rel_arr);
 		pg_free(db_arr->dbs[dbnum].db_name);
+
+		/*
+		 * Logical replication slots must not exist on the new cluster before
+		 * doing create_logical_replication_slots().
+		 */
+		Assert(db_arr->dbs[dbnum].slot_arr.slots == NULL);
 	}
 	pg_free(db_arr->dbs);
 	db_arr->dbs = NULL;
@@ -660,3 +757,15 @@ print_rel_infos(RelInfoArr *rel_arr)
 			   rel_arr->rels[relnum].reloid,
 			   rel_arr->rels[relnum].tablespace);
 }
+
+static void
+print_slot_infos(LogicalSlotInfoArr *slot_arr)
+{
+	int			slotnum;
+
+	for (slotnum = 0; slotnum < slot_arr->nslots; slotnum++)
+		pg_log(PG_VERBOSE, "slotname: \"%s\", plugin: \"%s\", two_phase: %d",
+			   slot_arr->slots[slotnum].slotname,
+			   slot_arr->slots[slotnum].plugin,
+			   slot_arr->slots[slotnum].two_phase);
+}
diff --git a/src/bin/pg_upgrade/meson.build b/src/bin/pg_upgrade/meson.build
index 12a97f84e2..228f29b688 100644
--- a/src/bin/pg_upgrade/meson.build
+++ b/src/bin/pg_upgrade/meson.build
@@ -42,6 +42,7 @@ tests += {
     'tests': [
       't/001_basic.pl',
       't/002_pg_upgrade.pl',
+      't/003_logical_replication_slots.pl',
     ],
     'test_kwargs': {'priority': 40}, # pg_upgrade tests are slow
   },
diff --git a/src/bin/pg_upgrade/option.c b/src/bin/pg_upgrade/option.c
index 640361009e..df66a5ffe6 100644
--- a/src/bin/pg_upgrade/option.c
+++ b/src/bin/pg_upgrade/option.c
@@ -57,6 +57,7 @@ parseCommandLine(int argc, char *argv[])
 		{"verbose", no_argument, NULL, 'v'},
 		{"clone", no_argument, NULL, 1},
 		{"copy", no_argument, NULL, 2},
+		{"include-logical-replication-slots", no_argument, NULL, 3},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -199,6 +200,10 @@ parseCommandLine(int argc, char *argv[])
 				user_opts.transfer_mode = TRANSFER_MODE_COPY;
 				break;
 
+			case 3:
+				user_opts.include_logical_slots = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						os_info.progname);
@@ -289,6 +294,8 @@ usage(void)
 	printf(_("  -V, --version                 display version information, then exit\n"));
 	printf(_("  --clone                       clone instead of copying files to new cluster\n"));
 	printf(_("  --copy                        copy files to new cluster (default)\n"));
+	printf(_("  --include-logical-replication-slots\n"
+			 "                                upgrade logical replication slots\n"));
 	printf(_("  -?, --help                    show this help, then exit\n"));
 	printf(_("\n"
 			 "Before running pg_upgrade you must:\n"
diff --git a/src/bin/pg_upgrade/pg_upgrade.c b/src/bin/pg_upgrade/pg_upgrade.c
index 4562dafcff..6dd3832422 100644
--- a/src/bin/pg_upgrade/pg_upgrade.c
+++ b/src/bin/pg_upgrade/pg_upgrade.c
@@ -59,6 +59,7 @@ static void copy_xact_xlog_xid(void);
 static void set_frozenxids(bool minmxid_only);
 static void make_outputdirs(char *pgdata);
 static void setup(char *argv0, bool *live_check);
+static void create_logical_replication_slots(void);
 
 ClusterInfo old_cluster,
 			new_cluster;
@@ -188,6 +189,19 @@ main(int argc, char **argv)
 			  new_cluster.pgdata);
 	check_ok();
 
+	/*
+	 * Create logical replication slots if requested.
+	 *
+	 * Note: This must be done after doing pg_resetwal command because
+	 * pg_resetwal would remove required WALs.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		start_postmaster(&new_cluster, true);
+		create_logical_replication_slots();
+		stop_postmaster(false);
+	}
+
 	if (user_opts.do_sync)
 	{
 		prep_status("Sync data directory to disk");
@@ -860,3 +874,50 @@ set_frozenxids(bool minmxid_only)
 
 	check_ok();
 }
+
+/*
+ * create_logical_replication_slots()
+ *
+ * Similar to create_new_objects() but only restores logical replication slots.
+ */
+static void
+create_logical_replication_slots(void)
+{
+	int			dbnum;
+
+	prep_status_progress("Restoring logical replication slots in the new cluster");
+
+	for (dbnum = 0; dbnum < old_cluster.dbarr.ndbs; dbnum++)
+	{
+		char		slots_file_name[MAXPGPATH],
+					log_file_name[MAXPGPATH];
+		DbInfo	   *old_db = &old_cluster.dbarr.dbs[dbnum];
+
+		pg_log(PG_STATUS, "%s", old_db->db_name);
+
+		snprintf(slots_file_name, sizeof(slots_file_name),
+				 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+		snprintf(log_file_name, sizeof(log_file_name),
+				 DB_DUMP_LOG_FILE_MASK, old_db->db_oid);
+
+		parallel_exec_prog(log_file_name,
+						   NULL,
+						   "\"%s/psql\" %s --echo-queries --set ON_ERROR_STOP=on "
+						   "--no-psqlrc --dbname %s -f \"%s/%s\"",
+						   new_cluster.bindir,
+						   cluster_conn_opts(&new_cluster),
+						   old_db->db_name,
+						   log_opts.dumpdir,
+						   slots_file_name);
+	}
+
+	/* reap all children */
+	while (reap_child(true) == true)
+		;
+
+	end_progress_output();
+	check_ok();
+
+	/* update new_cluster info again */
+	get_logical_slot_infos(&new_cluster);
+}
diff --git a/src/bin/pg_upgrade/pg_upgrade.h b/src/bin/pg_upgrade/pg_upgrade.h
index 3eea0139c7..8034067492 100644
--- a/src/bin/pg_upgrade/pg_upgrade.h
+++ b/src/bin/pg_upgrade/pg_upgrade.h
@@ -29,6 +29,7 @@
 /* contains both global db information and CREATE DATABASE commands */
 #define GLOBALS_DUMP_FILE	"pg_upgrade_dump_globals.sql"
 #define DB_DUMP_FILE_MASK	"pg_upgrade_dump_%u.custom"
+#define DB_DUMP_LOGICAL_SLOTS_FILE_MASK	"pg_upgrade_dump_%u_logical_slots.sql"
 
 /*
  * Base directories that include all the files generated internally, from the
@@ -150,6 +151,22 @@ typedef struct
 	int			nrels;
 } RelInfoArr;
 
+/*
+ * Structure to store logical replication slot information
+ */
+typedef struct
+{
+	char	   *slotname;		/* slot name */
+	char	   *plugin;			/* plugin */
+	bool		two_phase;		/* Can the slot decode 2PC? */
+} LogicalSlotInfo;
+
+typedef struct
+{
+	int			nslots;
+	LogicalSlotInfo *slots;
+} LogicalSlotInfoArr;
+
 /*
  * The following structure represents a relation mapping.
  */
@@ -176,6 +193,7 @@ typedef struct
 	char		db_tablespace[MAXPGPATH];	/* database default tablespace
 											 * path */
 	RelInfoArr	rel_arr;		/* array of all user relinfos */
+	LogicalSlotInfoArr slot_arr;	/* array of all LogicalSlotInfo */
 } DbInfo;
 
 /*
@@ -304,6 +322,8 @@ typedef struct
 	transferMode transfer_mode; /* copy files or link them? */
 	int			jobs;			/* number of processes/threads to use */
 	char	   *socketdir;		/* directory to use for Unix sockets */
+	bool		include_logical_slots;	/* true -> dump and restore logical
+										 * replication slots */
 } UserOpts;
 
 typedef struct
@@ -400,6 +420,7 @@ FileNameMap *gen_db_file_maps(DbInfo *old_db,
 							  DbInfo *new_db, int *nmaps, const char *old_pgdata,
 							  const char *new_pgdata);
 void		get_db_and_rel_infos(ClusterInfo *cluster);
+int			get_logical_slot_infos(ClusterInfo *cluster);
 
 /* option.c */
 
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
new file mode 100644
index 0000000000..9ca266f6b2
--- /dev/null
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -0,0 +1,146 @@
+# Copyright (c) 2023, PostgreSQL Global Development Group
+
+# Tests for upgrading replication slots
+
+use strict;
+use warnings;
+
+use File::Path qw(rmtree);
+
+use PostgreSQL::Test::Cluster;
+use PostgreSQL::Test::Utils;
+use Test::More;
+
+# Can be changed to test the other modes
+my $mode = $ENV{PG_TEST_PG_UPGRADE_MODE} || '--copy';
+
+# Initialize old node
+my $old_node = PostgreSQL::Test::Cluster->new('old_node');
+$old_node->init(allows_streaming => 'logical');
+$old_node->start;
+
+# Initialize new node
+my $new_node = PostgreSQL::Test::Cluster->new('new_node');
+$new_node->init(allows_streaming => 1);
+
+my $bindir = $new_node->config_data('--bindir');
+
+$old_node->stop;
+
+# Cause a failure at the start of pg_upgrade because wal_level is replica
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with wrong wal_level');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+# Preparations for the subsequent test. The case max_replication_slots is set
+# to 0 is prohibited.
+$new_node->append_conf('postgresql.conf', "wal_level = 'logical'");
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 0");
+
+# Cause a failure at the start of pg_upgrade because max_replication_slots is 0
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with wrong max_replication_slots');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+# Preparations for the subsequent test. max_replication_slots is set to
+# non-zero value
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 1");
+
+# Create a slot on old node, and generate WALs
+$old_node->start;
+$old_node->safe_psql(
+	'postgres', qq[
+	SELECT pg_create_logical_replication_slot('test_slot', 'test_decoding', false, true);
+	SELECT pg_create_logical_replication_slot('to_be_dropped', 'test_decoding', false, true);
+	CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a;
+]);
+
+$old_node->stop;
+
+# Cause a failure at the start of pg_upgrade because max_replication_slots is
+# smaller than existing slots on old node
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with small max_replication_slots');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+# Preparations for the subsequent test. max_replication_slots is set to
+# appropriate value
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 10");
+
+# Remove an unnecessary slot and consume WALs
+$old_node->start;
+$old_node->safe_psql(
+	'postgres', qq[
+	SELECT pg_drop_replication_slot('to_be_dropped');
+	SELECT count(*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)
+]);
+$old_node->stop;
+
+# Actual run, pg_upgrade_output.d is removed at the end
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots'
+	],
+	'run of pg_upgrade of old node');
+ok( !-d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+$new_node->start;
+my $result = $new_node->safe_psql('postgres',
+	"SELECT slot_name, two_phase FROM pg_replication_slots");
+is($result, qq(test_slot|t), 'check the slot exists on new node');
+
+done_testing();
diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list
index 05814136c6..5241b7bf8e 100644
--- a/src/tools/pgindent/typedefs.list
+++ b/src/tools/pgindent/typedefs.list
@@ -1497,7 +1497,10 @@ LogicalRepStreamAbortData
 LogicalRepTupleData
 LogicalRepTyp
 LogicalRepWorker
+LogicalReplicationSlotInfo
 LogicalRewriteMappingData
+LogicalSlotInfo
+LogicalSlotInfoArr
 LogicalTape
 LogicalTapeSet
 LsnReadQueue
-- 
2.34.1

From ab8839138521af35f6b00f530a39d10ef8ead555 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Fri, 21 Jul 2023 05:34:21 +0000
Subject: [PATCH 1/3] Send shutdown checkpoint record to subscriber

---
 src/backend/replication/walsender.c | 30 +++++++++++++++++++++++------
 1 file changed, 24 insertions(+), 6 deletions(-)

diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c
index d27ef2985d..fc1363ba76 100644
--- a/src/backend/replication/walsender.c
+++ b/src/backend/replication/walsender.c
@@ -187,6 +187,9 @@ static bool WalSndCaughtUp = false;
 static volatile sig_atomic_t got_SIGUSR2 = false;
 static volatile sig_atomic_t got_STOPPING = false;
 
+/* Are all the WALs flushed? */
+static bool WalsAreFlushed = false;
+
 /*
  * This is set while we are streaming. When not set
  * PROCSIG_WALSND_INIT_STOPPING signal will be handled like SIGTERM. When set,
@@ -260,7 +263,6 @@ static bool TransactionIdInRecentPast(TransactionId xid, uint32 epoch);
 static void WalSndSegmentOpen(XLogReaderState *state, XLogSegNo nextSegNo,
 							  TimeLineID *tli_p);
 
-
 /* Initialize walsender process before entering the main command loop */
 void
 InitWalSender(void)
@@ -1581,7 +1583,10 @@ WalSndWaitForWal(XLogRecPtr loc)
 		 * written, because walwriter has shut down already.
 		 */
 		if (got_STOPPING)
+		{
 			XLogBackgroundFlush();
+			WalsAreFlushed = true;
+		}
 
 		/* Update our idea of the currently flushed position. */
 		if (!RecoveryInProgress())
@@ -3100,12 +3105,20 @@ XLogSendLogical(void)
 		WalSndCaughtUp = true;
 
 	/*
-	 * If we're caught up and have been requested to stop, have WalSndLoop()
-	 * terminate the connection in an orderly manner, after writing out all
-	 * the pending data.
+	 * If we're caught up, have been requested to stop and there are no pending
+	 * records to be sent, change to stopping mode.
 	 */
-	if (WalSndCaughtUp && got_STOPPING)
-		got_SIGUSR2 = true;
+	if (WalSndCaughtUp && WalsAreFlushed && !pq_is_send_pending())
+	{
+		/*
+		 * Update the stats forcibly. pgstat_shutdown_hook reports any pending
+		 * stats at the end of the process, but it would happen after the
+		 * checkpointer exits so that it would lead assertion failure. We must
+		 * ensure all the stats are recorded before changing the state.
+		 */
+		pgstat_report_stat(true);
+		WalSndSetState(WALSNDSTATE_STOPPING);
+	}
 
 	/* Update shared memory status */
 	{
@@ -3142,6 +3155,7 @@ WalSndDone(WalSndSendDataCallback send_data)
 	replicatedPtr = XLogRecPtrIsInvalid(MyWalSnd->flush) ?
 		MyWalSnd->write : MyWalSnd->flush;
 
+
 	if (WalSndCaughtUp && sentPtr == replicatedPtr &&
 		!pq_is_send_pending())
 	{
@@ -3152,6 +3166,10 @@ WalSndDone(WalSndSendDataCallback send_data)
 		EndCommand(&qc, DestRemote, false);
 		pq_flush();
 
+		/* Mark the slot as dirty and save it to update the confirmed_flush. */
+		ReplicationSlotMarkDirty();
+		ReplicationSlotSave();
+
 		proc_exit(0);
 	}
 	if (!waiting_for_ping_response)
-- 
2.34.1

From df324a756b175ad9014d8dfe439fea1fa8137029 Mon Sep 17 00:00:00 2001
From: Vignesh C <vignesh21@gmail.com>
Date: Fri, 28 Jul 2023 16:50:52 +0530
Subject: [PATCH 3/3] Move pg_get_wal_records_info functionality from
 pg_walinspect to backend.

Upgrade of publications required pg_get_wal_records_info to check that
there are no WAL record other thatn CHECKPOINT_SHUTDOWN WAL record to be
consumed. Hence moved pg_get_wal_records_info functionality as
pg_get_wal_records_content to the backend so that it can be called from
pg_upgrade.
---
 contrib/pg_walinspect/Makefile                |   2 +-
 contrib/pg_walinspect/meson.build             |   1 +
 .../pg_walinspect/pg_walinspect--1.1--1.2.sql |  32 ++++
 contrib/pg_walinspect/pg_walinspect.c         | 148 +-----------------
 contrib/pg_walinspect/pg_walinspect.control   |   2 +-
 doc/src/sgml/func.sgml                        |  52 ++++++
 src/backend/access/transam/xlog.c             |  27 ++++
 src/backend/access/transam/xlogfuncs.c        | 125 +++++++++++++++
 src/backend/access/transam/xlogutils.c        | 117 ++++++++++++++
 src/backend/catalog/system_functions.sql      |   4 +
 src/include/access/xlog.h                     |   2 +
 src/include/access/xlogutils.h                |   4 +
 src/include/catalog/pg_proc.dat               |   9 ++
 src/test/modules/brin/t/02_wal_consistency.pl |   2 +-
 .../modules/test_custom_rmgrs/t/001_basic.pl  |   2 +-
 src/test/regress/expected/misc_functions.out  |  79 ++++++++++
 src/test/regress/sql/misc_functions.sql       |  57 +++++++
 17 files changed, 515 insertions(+), 150 deletions(-)
 create mode 100644 contrib/pg_walinspect/pg_walinspect--1.1--1.2.sql

diff --git a/contrib/pg_walinspect/Makefile b/contrib/pg_walinspect/Makefile
index 22090f7716..5cc7d81b42 100644
--- a/contrib/pg_walinspect/Makefile
+++ b/contrib/pg_walinspect/Makefile
@@ -7,7 +7,7 @@ OBJS = \
 PGFILEDESC = "pg_walinspect - functions to inspect contents of PostgreSQL Write-Ahead Log"
 
 EXTENSION = pg_walinspect
-DATA = pg_walinspect--1.0.sql pg_walinspect--1.0--1.1.sql
+DATA = pg_walinspect--1.0.sql pg_walinspect--1.0--1.1.sql pg_walinspect--1.1--1.2.sql
 
 REGRESS = pg_walinspect oldextversions
 
diff --git a/contrib/pg_walinspect/meson.build b/contrib/pg_walinspect/meson.build
index 80059f6119..8f7a99a493 100644
--- a/contrib/pg_walinspect/meson.build
+++ b/contrib/pg_walinspect/meson.build
@@ -20,6 +20,7 @@ install_data(
   'pg_walinspect.control',
   'pg_walinspect--1.0.sql',
   'pg_walinspect--1.0--1.1.sql',
+  'pg_walinspect--1.1--1.2.sql',
   kwargs: contrib_data_args,
 )
 
diff --git a/contrib/pg_walinspect/pg_walinspect--1.1--1.2.sql b/contrib/pg_walinspect/pg_walinspect--1.1--1.2.sql
new file mode 100644
index 0000000000..41ec538623
--- /dev/null
+++ b/contrib/pg_walinspect/pg_walinspect--1.1--1.2.sql
@@ -0,0 +1,32 @@
+/* contrib/pg_walinspect/pg_walinspect--1.1--1.2.sql */
+
+-- complain if script is sourced in psql, rather than via CREATE EXTENSION
+\echo Use "ALTER EXTENSION pg_walinspect UPDATE TO 1.2" to load this file. \quit
+
+/* These functions are now in the backend and callers should update to use those */
+ALTER EXTENSION pg_walinspect DROP FUNCTION pg_get_wal_records_info;
+DROP FUNCTION pg_get_wal_records_info(pg_lsn, pg_lsn);
+
+--
+-- pg_get_wal_records_info()
+--
+CREATE FUNCTION pg_get_wal_records_info(IN start_lsn pg_lsn,
+    IN end_lsn pg_lsn,
+    OUT start_lsn pg_lsn,
+    OUT end_lsn pg_lsn,
+    OUT prev_lsn pg_lsn,
+    OUT xid xid,
+    OUT resource_manager text,
+    OUT record_type text,
+    OUT record_length int4,
+    OUT main_data_length int4,
+    OUT fpi_length int4,
+    OUT description text,
+    OUT block_ref text
+)
+RETURNS SETOF record
+AS 'pg_get_wal_records_content'
+LANGUAGE INTERNAL STRICT PARALLEL SAFE;
+
+REVOKE EXECUTE ON FUNCTION pg_get_wal_records_info(pg_lsn, pg_lsn) FROM PUBLIC;
+GRANT EXECUTE ON FUNCTION pg_get_wal_records_info(pg_lsn, pg_lsn) TO pg_read_server_files;
diff --git a/contrib/pg_walinspect/pg_walinspect.c b/contrib/pg_walinspect/pg_walinspect.c
index 796a74f322..6e572d3436 100644
--- a/contrib/pg_walinspect/pg_walinspect.c
+++ b/contrib/pg_walinspect/pg_walinspect.c
@@ -38,10 +38,6 @@ PG_FUNCTION_INFO_V1(pg_get_wal_records_info_till_end_of_wal);
 PG_FUNCTION_INFO_V1(pg_get_wal_stats);
 PG_FUNCTION_INFO_V1(pg_get_wal_stats_till_end_of_wal);
 
-static void ValidateInputLSNs(XLogRecPtr start_lsn, XLogRecPtr *end_lsn);
-static XLogRecPtr GetCurrentLSN(void);
-static XLogReaderState *InitXLogReaderState(XLogRecPtr lsn);
-static XLogRecord *ReadNextXLogRecord(XLogReaderState *xlogreader);
 static void GetWALRecordInfo(XLogReaderState *record, Datum *values,
 							 bool *nulls, uint32 ncols);
 static void GetWALRecordsInfo(FunctionCallInfo fcinfo,
@@ -62,120 +58,6 @@ static void GetWalStats(FunctionCallInfo fcinfo,
 static void GetWALBlockInfo(FunctionCallInfo fcinfo, XLogReaderState *record,
 							bool show_data);
 
-/*
- * Return the LSN up to which the server has WAL.
- */
-static XLogRecPtr
-GetCurrentLSN(void)
-{
-	XLogRecPtr	curr_lsn;
-
-	/*
-	 * We determine the current LSN of the server similar to how page_read
-	 * callback read_local_xlog_page_no_wait does.
-	 */
-	if (!RecoveryInProgress())
-		curr_lsn = GetFlushRecPtr(NULL);
-	else
-		curr_lsn = GetXLogReplayRecPtr(NULL);
-
-	Assert(!XLogRecPtrIsInvalid(curr_lsn));
-
-	return curr_lsn;
-}
-
-/*
- * Initialize WAL reader and identify first valid LSN.
- */
-static XLogReaderState *
-InitXLogReaderState(XLogRecPtr lsn)
-{
-	XLogReaderState *xlogreader;
-	ReadLocalXLogPageNoWaitPrivate *private_data;
-	XLogRecPtr	first_valid_record;
-
-	/*
-	 * Reading WAL below the first page of the first segments isn't allowed.
-	 * This is a bootstrap WAL page and the page_read callback fails to read
-	 * it.
-	 */
-	if (lsn < XLOG_BLCKSZ)
-		ereport(ERROR,
-				(errmsg("could not read WAL at LSN %X/%X",
-						LSN_FORMAT_ARGS(lsn))));
-
-	private_data = (ReadLocalXLogPageNoWaitPrivate *)
-		palloc0(sizeof(ReadLocalXLogPageNoWaitPrivate));
-
-	xlogreader = XLogReaderAllocate(wal_segment_size, NULL,
-									XL_ROUTINE(.page_read = &read_local_xlog_page_no_wait,
-											   .segment_open = &wal_segment_open,
-											   .segment_close = &wal_segment_close),
-									private_data);
-
-	if (xlogreader == NULL)
-		ereport(ERROR,
-				(errcode(ERRCODE_OUT_OF_MEMORY),
-				 errmsg("out of memory"),
-				 errdetail("Failed while allocating a WAL reading processor.")));
-
-	/* first find a valid recptr to start from */
-	first_valid_record = XLogFindNextRecord(xlogreader, lsn);
-
-	if (XLogRecPtrIsInvalid(first_valid_record))
-		ereport(ERROR,
-				(errmsg("could not find a valid record after %X/%X",
-						LSN_FORMAT_ARGS(lsn))));
-
-	return xlogreader;
-}
-
-/*
- * Read next WAL record.
- *
- * By design, to be less intrusive in a running system, no slot is allocated
- * to reserve the WAL we're about to read. Therefore this function can
- * encounter read errors for historical WAL.
- *
- * We guard against ordinary errors trying to read WAL that hasn't been
- * written yet by limiting end_lsn to the flushed WAL, but that can also
- * encounter errors if the flush pointer falls in the middle of a record. In
- * that case we'll return NULL.
- */
-static XLogRecord *
-ReadNextXLogRecord(XLogReaderState *xlogreader)
-{
-	XLogRecord *record;
-	char	   *errormsg;
-
-	record = XLogReadRecord(xlogreader, &errormsg);
-
-	if (record == NULL)
-	{
-		ReadLocalXLogPageNoWaitPrivate *private_data;
-
-		/* return NULL, if end of WAL is reached */
-		private_data = (ReadLocalXLogPageNoWaitPrivate *)
-			xlogreader->private_data;
-
-		if (private_data->end_of_wal)
-			return NULL;
-
-		if (errormsg)
-			ereport(ERROR,
-					(errcode_for_file_access(),
-					 errmsg("could not read WAL at %X/%X: %s",
-							LSN_FORMAT_ARGS(xlogreader->EndRecPtr), errormsg)));
-		else
-			ereport(ERROR,
-					(errcode_for_file_access(),
-					 errmsg("could not read WAL at %X/%X",
-							LSN_FORMAT_ARGS(xlogreader->EndRecPtr))));
-	}
-
-	return record;
-}
-
 /*
  * Output values that make up a row describing caller's WAL record.
  *
@@ -502,33 +384,6 @@ pg_get_wal_record_info(PG_FUNCTION_ARGS)
 #undef PG_GET_WAL_RECORD_INFO_COLS
 }
 
-/*
- * Validate start and end LSNs coming from the function inputs.
- *
- * If end_lsn is found to be higher than the current LSN reported by the
- * cluster, use the current LSN as the upper bound.
- */
-static void
-ValidateInputLSNs(XLogRecPtr start_lsn, XLogRecPtr *end_lsn)
-{
-	XLogRecPtr	curr_lsn = GetCurrentLSN();
-
-	if (start_lsn > curr_lsn)
-		ereport(ERROR,
-				(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
-				 errmsg("WAL start LSN must be less than current LSN"),
-				 errdetail("Current WAL LSN on the database system is at %X/%X.",
-						   LSN_FORMAT_ARGS(curr_lsn))));
-
-	if (start_lsn > *end_lsn)
-		ereport(ERROR,
-				(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
-				 errmsg("WAL start LSN must be less than end LSN")));
-
-	if (*end_lsn > curr_lsn)
-		*end_lsn = curr_lsn;
-}
-
 /*
  * Get info of all WAL records between start LSN and end LSN.
  */
@@ -582,7 +437,8 @@ GetWALRecordsInfo(FunctionCallInfo fcinfo, XLogRecPtr start_lsn,
 }
 
 /*
- * Get info of all WAL records between start LSN and end LSN.
+ * The following functions have been removed in newer versions in 1.2, but
+ * they are kept around for compatibility.
  */
 Datum
 pg_get_wal_records_info(PG_FUNCTION_ARGS)
diff --git a/contrib/pg_walinspect/pg_walinspect.control b/contrib/pg_walinspect/pg_walinspect.control
index efa3cb2cfe..5f574b865b 100644
--- a/contrib/pg_walinspect/pg_walinspect.control
+++ b/contrib/pg_walinspect/pg_walinspect.control
@@ -1,5 +1,5 @@
 # pg_walinspect extension
 comment = 'functions to inspect contents of PostgreSQL Write-Ahead Log'
-default_version = '1.1'
+default_version = '1.2'
 module_pathname = '$libdir/pg_walinspect'
 relocatable = true
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index b94827674c..a8866feaa3 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -26670,6 +26670,37 @@ LOG:  Grand total: 1651920 bytes in 201 blocks; 622360 free (88 chunks); 1029560
         get the replication lag.
        </para></entry>
       </row>
+
+      <row>
+       <entry role="func_table_entry"><para role="func_signature">
+        <indexterm>
+         <primary>pg_get_wal_records_content</primary>
+        </indexterm>
+        <function>pg_get_wal_records_content</function> ( <parameter>lsn1</parameter> <type>pg_lsn</type>, <parameter>lsn2</parameter> <type>pg_lsn</type> )
+        <returnvalue>record</returnvalue>
+        ( <parameter>start_lsn</parameter> <type>pg_lsn</type>,
+        <parameter>end_lsn</parameter> <type>pg_lsn</type>,
+        <parameter>prev_lsn</parameter> <type>pg_lsn</type>,
+        <parameter>xid</parameter> <type>xid</type>,
+        <parameter>resource_manager</parameter> <type>text</type>,
+        <parameter>record_type</parameter> <type>text</type>,
+        <parameter>record_length</parameter> <type>integer</type>,
+        <parameter>main_data_length</parameter> <type>integer</type>,
+        <parameter>fpi_length</parameter> <type>integer</type>,
+        <parameter>description</parameter> <type>text</type>,
+        <parameter>block_ref</parameter> <type>text</type>
+         )
+       </para>
+       <para>
+        Gets information of all the valid WAL records between
+        <replaceable>start_lsn</replaceable> and <replaceable>end_lsn</replaceable>.
+        Returns one row per WAL record. This can be used
+        with <structname>pg_stat_replication</structname> or some of the
+        functions shown in <xref linkend="functions-admin-backup-table"/> to
+        get the replication lag WAL records content. The function raises an
+        error if <replaceable>start_lsn</replaceable> is not available.
+       </para></entry>
+      </row>
      </tbody>
     </tgroup>
    </table>
@@ -26728,6 +26759,27 @@ postgres=# SELECT '0/0'::pg_lsn + pd.segment_number * ps.setting::int + :offset
 </programlisting>
    </para>
 
+   <para>
+    <function>pg_get_wal_records_content</function> gets information of all the valid WAL records between
+     <replaceable>start_lsn</replaceable> and <replaceable>end_lsn</replaceable>.
+     Returns one row per WAL record.  For example:
+<screen>
+postgres=# SELECT * FROM pg_get_wal_records_content('0/1E913618', '0/1E913740') LIMIT 1;
+-[ RECORD 1 ]----+--------------------------------------------------------------
+start_lsn        | 0/1E913618
+end_lsn          | 0/1E913650
+prev_lsn         | 0/1E9135A0
+xid              | 0
+resource_manager | Standby
+record_type      | RUNNING_XACTS
+record_length    | 50
+main_data_length | 24
+fpi_length       | 0
+description      | nextXid 33775 latestCompletedXid 33774 oldestRunningXid 33775
+block_ref        |
+</screen>
+   </para>
+
   </sect2>
 
   <sect2 id="functions-recovery-control">
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 60c0b7ec3a..6115df57fd 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -8988,3 +8988,30 @@ SetWalWriterSleeping(bool sleeping)
 	XLogCtl->WalWriterSleeping = sleeping;
 	SpinLockRelease(&XLogCtl->info_lck);
 }
+
+/*
+ * Validate start and end LSNs coming from the function inputs.
+ *
+ * If end_lsn is found to be higher than the current LSN reported by the
+ * cluster, use the current LSN as the upper bound.
+ */
+void
+ValidateInputLSNs(XLogRecPtr start_lsn, XLogRecPtr *end_lsn)
+{
+	XLogRecPtr	curr_lsn = GetCurrentLSN();
+
+	if (start_lsn > curr_lsn)
+		ereport(ERROR,
+				(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
+				 errmsg("WAL start LSN must be less than current LSN"),
+				 errdetail("Current WAL LSN on the database system is at %X/%X.",
+						   LSN_FORMAT_ARGS(curr_lsn))));
+
+	if (start_lsn > *end_lsn)
+		ereport(ERROR,
+				(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
+				 errmsg("WAL start LSN must be less than end LSN")));
+
+	if (*end_lsn > curr_lsn)
+		*end_lsn = curr_lsn;
+}
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 5044ff0643..bb66eb26b6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -754,3 +754,128 @@ pg_promote(PG_FUNCTION_ARGS)
 						   wait_seconds)));
 	PG_RETURN_BOOL(false);
 }
+
+/*
+ * Get content of all WAL records between start LSN and end LSN.
+ */
+Datum
+pg_get_wal_records_content(PG_FUNCTION_ARGS)
+{
+#define PG_GET_WAL_RECORDS_INFO_COLS 11
+	FuncCallContext *funcctx;
+	XLogReaderState *xlogreader;
+
+	XLogRecPtr	start_lsn = PG_GETARG_LSN(0);
+	XLogRecPtr	end_lsn = PG_GETARG_LSN(1);
+
+	if (SRF_IS_FIRSTCALL())
+	{
+		MemoryContext oldcontext;
+		TupleDesc	tupdesc;
+
+		ValidateInputLSNs(start_lsn, &end_lsn);
+
+		/* create a function context for cross-call persistence */
+		funcctx = SRF_FIRSTCALL_INIT();
+
+		/*
+		 * Switch to memory context appropriate for multiple function calls
+		 */
+		oldcontext = MemoryContextSwitchTo(funcctx->multi_call_memory_ctx);
+
+		/* build tupdesc for result tuples */
+		tupdesc = CreateTemplateTupleDesc(11);
+		TupleDescInitEntry(tupdesc, (AttrNumber) 1, "start_lsn",
+						   PG_LSNOID, -1, 0);
+		TupleDescInitEntry(tupdesc, (AttrNumber) 2, "end_lsn",
+						   PG_LSNOID, -1, 0);
+		TupleDescInitEntry(tupdesc, (AttrNumber) 3, "prev_lsn",
+						   PG_LSNOID, -1, 0);
+		TupleDescInitEntry(tupdesc, (AttrNumber) 4, "xid",
+						   XIDOID, -1, 0);
+		TupleDescInitEntry(tupdesc, (AttrNumber) 5, "resource_manager",
+						   TEXTOID, -1, 0);
+		TupleDescInitEntry(tupdesc, (AttrNumber) 6, "record_type",
+						   TEXTOID, -1, 0);
+		TupleDescInitEntry(tupdesc, (AttrNumber) 7, "record_length",
+						   INT4OID, -1, 0);
+		TupleDescInitEntry(tupdesc, (AttrNumber) 8, "main_data_length",
+						   INT4OID, -1, 0);
+		TupleDescInitEntry(tupdesc, (AttrNumber) 9, "fpi_length",
+						   INT4OID, -1, 0);
+		TupleDescInitEntry(tupdesc, (AttrNumber) 10, "description",
+						   TEXTOID, -1, 0);
+		TupleDescInitEntry(tupdesc, (AttrNumber) 11, "block_ref",
+						   TEXTOID, -1, 0);
+
+		funcctx->tuple_desc = BlessTupleDesc(tupdesc);
+
+		if (start_lsn < end_lsn)
+			funcctx->user_fctx = InitXLogReaderState(start_lsn);
+		else
+			funcctx->user_fctx = NULL;
+
+		MemoryContextSwitchTo(oldcontext);
+	}
+
+	funcctx = SRF_PERCALL_SETUP();
+	xlogreader = (XLogReaderState *) funcctx->user_fctx;
+
+	while (xlogreader && ReadNextXLogRecord(xlogreader) &&
+		   xlogreader->EndRecPtr <= end_lsn)
+	{
+		Datum		values[11] = {0};
+		bool		nulls[11] = {0};
+		HeapTuple	tuple;
+		Datum		result;
+		XLogReaderState *record = xlogreader;
+		const char *record_type;
+		StringInfoData rec_desc;
+		StringInfoData rec_blk_ref;
+		int			i = 0;
+		uint32		fpi_len = 0;
+		RmgrData	desc;
+
+		desc = GetRmgr(XLogRecGetRmid(record));
+		record_type = desc.rm_identify(XLogRecGetInfo(record));
+
+		if (record_type == NULL)
+			record_type = psprintf("UNKNOWN (%x)", XLogRecGetInfo(record) & ~XLR_INFO_MASK);
+
+		initStringInfo(&rec_desc);
+		desc.rm_desc(&rec_desc, record);
+
+		if (XLogRecHasAnyBlockRefs(record))
+		{
+			initStringInfo(&rec_blk_ref);
+			XLogRecGetBlockRefInfo(record, false, true, &rec_blk_ref, &fpi_len);
+		}
+
+		values[i++] = LSNGetDatum(record->ReadRecPtr);
+		values[i++] = LSNGetDatum(record->EndRecPtr);
+		values[i++] = LSNGetDatum(XLogRecGetPrev(record));
+		values[i++] = TransactionIdGetDatum(XLogRecGetXid(record));
+		values[i++] = CStringGetTextDatum(desc.rm_name);
+		values[i++] = CStringGetTextDatum(record_type);
+		values[i++] = UInt32GetDatum(XLogRecGetTotalLen(record));
+		values[i++] = UInt32GetDatum(XLogRecGetDataLen(record));
+		values[i++] = UInt32GetDatum(fpi_len);
+
+		if (rec_desc.len > 0)
+			values[i++] = CStringGetTextDatum(rec_desc.data);
+		else
+			nulls[i++] = true;
+
+		if (XLogRecHasAnyBlockRefs(record))
+			values[i++] = CStringGetTextDatum(rec_blk_ref.data);
+		else
+			nulls[i++] = true;
+
+		tuple = heap_form_tuple(funcctx->tuple_desc, values, nulls);
+		result = HeapTupleGetDatum(tuple);
+		SRF_RETURN_NEXT(funcctx, result);
+	}
+
+	SRF_RETURN_DONE(funcctx);
+#undef PG_GET_WAL_RECORDS_INFO_COLS
+}
diff --git a/src/backend/access/transam/xlogutils.c b/src/backend/access/transam/xlogutils.c
index e174a2a891..90eab3e2f3 100644
--- a/src/backend/access/transam/xlogutils.c
+++ b/src/backend/access/transam/xlogutils.c
@@ -1048,3 +1048,120 @@ WALReadRaiseError(WALReadError *errinfo)
 						errinfo->wre_req)));
 	}
 }
+
+/*
+ * Return the LSN up to which the server has WAL.
+ */
+XLogRecPtr
+GetCurrentLSN(void)
+{
+	XLogRecPtr	curr_lsn;
+
+	/*
+	 * We determine the current LSN of the server similar to how page_read
+	 * callback read_local_xlog_page_no_wait does.
+	 */
+	if (!RecoveryInProgress())
+		curr_lsn = GetFlushRecPtr(NULL);
+	else
+		curr_lsn = GetXLogReplayRecPtr(NULL);
+
+	Assert(!XLogRecPtrIsInvalid(curr_lsn));
+
+	return curr_lsn;
+}
+
+/*
+ * Initialize WAL reader and identify first valid LSN.
+ */
+XLogReaderState *
+InitXLogReaderState(XLogRecPtr lsn)
+{
+	XLogReaderState *xlogreader;
+	ReadLocalXLogPageNoWaitPrivate *private_data;
+	XLogRecPtr	first_valid_record;
+
+	/*
+	 * Reading WAL below the first page of the first segments isn't allowed.
+	 * This is a bootstrap WAL page and the page_read callback fails to read
+	 * it.
+	 */
+	if (lsn < XLOG_BLCKSZ)
+		ereport(ERROR,
+				(errmsg("could not read WAL at LSN %X/%X",
+						LSN_FORMAT_ARGS(lsn))));
+
+	private_data = (ReadLocalXLogPageNoWaitPrivate *)
+		palloc0(sizeof(ReadLocalXLogPageNoWaitPrivate));
+
+	xlogreader = XLogReaderAllocate(wal_segment_size, NULL,
+									XL_ROUTINE(.page_read = &read_local_xlog_page_no_wait,
+											   .segment_open = &wal_segment_open,
+											   .segment_close = &wal_segment_close),
+									private_data);
+
+	if (xlogreader == NULL)
+		ereport(ERROR,
+				(errcode(ERRCODE_OUT_OF_MEMORY),
+				 errmsg("out of memory"),
+				 errdetail("Failed while allocating a WAL reading processor.")));
+
+	/* first find a valid recptr to start from */
+	first_valid_record = XLogFindNextRecord(xlogreader, lsn);
+
+	if (XLogRecPtrIsInvalid(first_valid_record))
+	{
+		ereport(ERROR,
+				(errmsg("could not find a valid record after %X/%X",
+						LSN_FORMAT_ARGS(lsn))));
+	}
+
+	return xlogreader;
+}
+
+
+/*
+ * Read next WAL record.
+ *
+ * By design, to be less intrusive in a running system, no slot is allocated
+ * to reserve the WAL we're about to read. Therefore this function can
+ * encounter read errors for historical WAL.
+ *
+ * We guard against ordinary errors trying to read WAL that hasn't been
+ * written yet by limiting end_lsn to the flushed WAL, but that can also
+ * encounter errors if the flush pointer falls in the middle of a record. In
+ * that case we'll return NULL.
+ */
+XLogRecord *
+ReadNextXLogRecord(XLogReaderState *xlogreader)
+{
+	XLogRecord *record;
+	char	   *errormsg;
+
+	record = XLogReadRecord(xlogreader, &errormsg);
+
+	if (record == NULL)
+	{
+		ReadLocalXLogPageNoWaitPrivate *private_data;
+
+		/* return NULL, if end of WAL is reached */
+		private_data = (ReadLocalXLogPageNoWaitPrivate *)
+			xlogreader->private_data;
+
+		if (private_data->end_of_wal)
+			return NULL;
+
+		if (errormsg)
+			ereport(ERROR,
+					(errcode_for_file_access(),
+					 errmsg("could not read WAL at %X/%X: %s",
+							LSN_FORMAT_ARGS(xlogreader->EndRecPtr), errormsg)));
+		else
+			ereport(ERROR,
+					(errcode_for_file_access(),
+					 errmsg("could not read WAL at %X/%X",
+							LSN_FORMAT_ARGS(xlogreader->EndRecPtr))));
+	}
+
+	return record;
+}
diff --git a/src/backend/catalog/system_functions.sql b/src/backend/catalog/system_functions.sql
index 07c0d89c4f..1fed0def9b 100644
--- a/src/backend/catalog/system_functions.sql
+++ b/src/backend/catalog/system_functions.sql
@@ -616,6 +616,8 @@ REVOKE EXECUTE ON FUNCTION pg_backup_stop(boolean) FROM public;
 
 REVOKE EXECUTE ON FUNCTION pg_create_restore_point(text) FROM public;
 
+REVOKE EXECUTE ON FUNCTION pg_get_wal_records_content(pg_lsn, pg_lsn) FROM public;
+
 REVOKE EXECUTE ON FUNCTION pg_switch_wal() FROM public;
 
 REVOKE EXECUTE ON FUNCTION pg_log_standby_snapshot() FROM public;
@@ -726,6 +728,8 @@ REVOKE EXECUTE ON FUNCTION pg_ls_replslotdir(text) FROM PUBLIC;
 -- We also set up some things as accessible to standard roles.
 --
 
+GRANT EXECUTE ON FUNCTION pg_get_wal_records_content(pg_lsn, pg_lsn) TO pg_read_server_files;
+
 GRANT EXECUTE ON FUNCTION pg_ls_logdir() TO pg_monitor;
 
 GRANT EXECUTE ON FUNCTION pg_ls_waldir() TO pg_monitor;
diff --git a/src/include/access/xlog.h b/src/include/access/xlog.h
index 48ca852381..4ec1fc9d82 100644
--- a/src/include/access/xlog.h
+++ b/src/include/access/xlog.h
@@ -259,6 +259,8 @@ extern void SetInstallXLogFileSegmentActive(void);
 extern bool IsInstallXLogFileSegmentActive(void);
 extern void XLogShutdownWalRcv(void);
 
+extern void ValidateInputLSNs(XLogRecPtr start_lsn, XLogRecPtr *end_lsn);
+
 /*
  * Routines to start, stop, and get status of a base backup.
  */
diff --git a/src/include/access/xlogutils.h b/src/include/access/xlogutils.h
index 5b77b11f50..77e2974ab3 100644
--- a/src/include/access/xlogutils.h
+++ b/src/include/access/xlogutils.h
@@ -115,4 +115,8 @@ extern void XLogReadDetermineTimeline(XLogReaderState *state,
 
 extern void WALReadRaiseError(WALReadError *errinfo);
 
+extern XLogRecPtr GetCurrentLSN(void);
+extern XLogReaderState *InitXLogReaderState(XLogRecPtr lsn);
+extern XLogRecord *ReadNextXLogRecord(XLogReaderState *xlogreader);
+
 #endif
diff --git a/src/include/catalog/pg_proc.dat b/src/include/catalog/pg_proc.dat
index 6996073989..d45696fca0 100644
--- a/src/include/catalog/pg_proc.dat
+++ b/src/include/catalog/pg_proc.dat
@@ -6489,6 +6489,15 @@
   proargnames => '{rm_id, rm_name, rm_builtin}',
   prosrc => 'pg_get_wal_resource_managers' },
 
+{ oid => '3813',
+  descr => 'Info of WAL conents between start LSN and end LSN',
+  proname => 'pg_get_wal_records_content', prorows => '10', proretset => 't',
+  provolatile => 's', prorettype => 'record', proargtypes => 'pg_lsn pg_lsn',
+  proallargtypes => '{pg_lsn,pg_lsn,pg_lsn,pg_lsn,pg_lsn,xid,text,text,int4,int4,int4,text,text}',
+  proargmodes => '{i,i,o,o,o,o,o,o,o,o,o,o,o}',
+  proargnames => '{start_lsn,end_lsn,start_lsn,end_lsn,prev_lsn,xid,resource_manager,record_type,record_length,main_data_length,fpi_length,description,block_ref}',
+  prosrc => 'pg_get_wal_records_content' },
+
 { oid => '2621', descr => 'reload configuration files',
   proname => 'pg_reload_conf', provolatile => 'v', prorettype => 'bool',
   proargtypes => '', prosrc => 'pg_reload_conf' },
diff --git a/src/test/modules/brin/t/02_wal_consistency.pl b/src/test/modules/brin/t/02_wal_consistency.pl
index 8b2b244feb..288bbd7a60 100644
--- a/src/test/modules/brin/t/02_wal_consistency.pl
+++ b/src/test/modules/brin/t/02_wal_consistency.pl
@@ -64,7 +64,7 @@ my $end_lsn = $whiskey->lsn('flush');
 
 my ($ret, $out, $err) = $whiskey->psql(
 	'postgres', qq{
-	select count(*) from pg_get_wal_records_info('$start_lsn', '$end_lsn')
+	select count(*) from pg_get_wal_records_content('$start_lsn', '$end_lsn')
 	where resource_manager = 'BRIN' AND
 	record_type ILIKE '%revmap%'
 	});
diff --git a/src/test/modules/test_custom_rmgrs/t/001_basic.pl b/src/test/modules/test_custom_rmgrs/t/001_basic.pl
index 50655d3788..4000de0560 100644
--- a/src/test/modules/test_custom_rmgrs/t/001_basic.pl
+++ b/src/test/modules/test_custom_rmgrs/t/001_basic.pl
@@ -54,7 +54,7 @@ my $expected =
   qq($record_end_lsn|test_custom_rmgrs|TEST_CUSTOM_RMGRS_MESSAGE|0|payload (10 bytes): payload123);
 my $result = $node->safe_psql(
 	'postgres',
-	qq[SELECT end_lsn, resource_manager, record_type, fpi_length, description FROM pg_get_wal_records_info('$start_lsn', '$end_lsn')
+	qq[SELECT end_lsn, resource_manager, record_type, fpi_length, description FROM pg_get_wal_records_content('$start_lsn', '$end_lsn')
 		WHERE resource_manager = 'test_custom_rmgrs';]);
 is($result, $expected,
 	'custom WAL resource manager has successfully written a WAL record');
diff --git a/src/test/regress/expected/misc_functions.out b/src/test/regress/expected/misc_functions.out
index c669948370..e5e10cddf9 100644
--- a/src/test/regress/expected/misc_functions.out
+++ b/src/test/regress/expected/misc_functions.out
@@ -642,3 +642,82 @@ SELECT segment_number > 0 AS ok_segment_number, timeline_id
  t                 |  4294967295
 (1 row)
 
+-- pg_get_wal_records_content
+CREATE TABLE sample_tbl(col1 int, col2 int);
+SELECT pg_current_wal_lsn() AS wal_lsn1 \gset
+INSERT INTO sample_tbl SELECT * FROM generate_series(1, 2);
+SELECT pg_current_wal_lsn() AS wal_lsn2 \gset
+-- Mask DETAIL messages as these could refer to current LSN positions.
+\set VERBOSITY terse
+-- Invalid start LSN.
+SELECT * FROM pg_get_wal_records_content('0/0', :'wal_lsn1');
+ERROR:  could not read WAL at LSN 0/0
+-- Start LSN > End LSN.
+SELECT * FROM pg_get_wal_records_content(:'wal_lsn2', :'wal_lsn1');
+ERROR:  WAL start LSN must be less than end LSN
+-- Success with end LSNs.
+SELECT COUNT(*) >= 1 AS ok FROM pg_get_wal_records_content(:'wal_lsn1', 'FFFFFFFF/FFFFFFFF');
+ ok 
+----
+ t
+(1 row)
+
+-- Failures with start LSNs.
+SELECT * FROM pg_get_wal_records_content('FFFFFFFF/FFFFFFFE', 'FFFFFFFF/FFFFFFFF');
+ERROR:  WAL start LSN must be less than current LSN
+SELECT COUNT(*) >= 1 AS ok FROM pg_get_wal_records_content(:'wal_lsn1', :'wal_lsn2');
+ ok 
+----
+ t
+(1 row)
+
+-- Test for filtering out WAL records of a particular table
+SELECT oid AS sample_tbl_oid FROM pg_class WHERE relname = 'sample_tbl' \gset
+SELECT COUNT(*) >= 1 AS ok FROM pg_get_wal_records_content(:'wal_lsn1', :'wal_lsn2')
+			WHERE block_ref LIKE concat('%', :'sample_tbl_oid', '%') AND resource_manager = 'Heap';
+ ok 
+----
+ t
+(1 row)
+
+-- Test for filtering out WAL records based on resource_manager and
+-- record_type
+SELECT COUNT(*) >= 1 AS ok FROM pg_get_wal_records_content(:'wal_lsn1', :'wal_lsn2')
+			WHERE resource_manager = 'Heap' AND record_type = 'INSERT';
+ ok 
+----
+ t
+(1 row)
+
+\set VERBOSITY default
+-- Tests for permissions
+CREATE ROLE regress_pg_get_wal;
+SELECT has_function_privilege('regress_pg_get_wal',
+  'pg_get_wal_records_content(pg_lsn, pg_lsn) ', 'EXECUTE'); -- no
+ has_function_privilege 
+------------------------
+ f
+(1 row)
+
+-- Functions accessible by users with role pg_read_server_files.
+GRANT pg_read_server_files TO regress_pg_get_wal;
+SELECT has_function_privilege('regress_pg_get_wal',
+  'pg_get_wal_records_content(pg_lsn, pg_lsn) ', 'EXECUTE'); -- yes
+ has_function_privilege 
+------------------------
+ t
+(1 row)
+
+-- Superuser can grant execute to other users.
+GRANT EXECUTE ON FUNCTION pg_get_wal_records_content(pg_lsn, pg_lsn)
+  TO regress_pg_get_wal;
+SELECT has_function_privilege('regress_pg_get_wal',
+  'pg_get_wal_records_content(pg_lsn, pg_lsn) ', 'EXECUTE'); -- yes
+ has_function_privilege 
+------------------------
+ t
+(1 row)
+
+REVOKE EXECUTE ON FUNCTION pg_get_wal_records_content(pg_lsn, pg_lsn)
+  FROM regress_pg_get_wal;
+DROP ROLE regress_pg_get_wal;
diff --git a/src/test/regress/sql/misc_functions.sql b/src/test/regress/sql/misc_functions.sql
index b57f01f3e9..40cee85dbc 100644
--- a/src/test/regress/sql/misc_functions.sql
+++ b/src/test/regress/sql/misc_functions.sql
@@ -237,3 +237,60 @@ SELECT segment_number > 0 AS ok_segment_number, timeline_id
   FROM pg_split_walfile_name('000000010000000100000000');
 SELECT segment_number > 0 AS ok_segment_number, timeline_id
   FROM pg_split_walfile_name('ffffffFF00000001000000af');
+
+-- pg_get_wal_records_content
+CREATE TABLE sample_tbl(col1 int, col2 int);
+SELECT pg_current_wal_lsn() AS wal_lsn1 \gset
+INSERT INTO sample_tbl SELECT * FROM generate_series(1, 2);
+SELECT pg_current_wal_lsn() AS wal_lsn2 \gset
+
+-- Mask DETAIL messages as these could refer to current LSN positions.
+\set VERBOSITY terse
+
+-- Invalid start LSN.
+SELECT * FROM pg_get_wal_records_content('0/0', :'wal_lsn1');
+-- Start LSN > End LSN.
+SELECT * FROM pg_get_wal_records_content(:'wal_lsn2', :'wal_lsn1');
+-- Success with end LSNs.
+SELECT COUNT(*) >= 1 AS ok FROM pg_get_wal_records_content(:'wal_lsn1', 'FFFFFFFF/FFFFFFFF');
+-- Failures with start LSNs.
+SELECT * FROM pg_get_wal_records_content('FFFFFFFF/FFFFFFFE', 'FFFFFFFF/FFFFFFFF');
+
+SELECT COUNT(*) >= 1 AS ok FROM pg_get_wal_records_content(:'wal_lsn1', :'wal_lsn2');
+
+-- Test for filtering out WAL records of a particular table
+SELECT oid AS sample_tbl_oid FROM pg_class WHERE relname = 'sample_tbl' \gset
+
+SELECT COUNT(*) >= 1 AS ok FROM pg_get_wal_records_content(:'wal_lsn1', :'wal_lsn2')
+			WHERE block_ref LIKE concat('%', :'sample_tbl_oid', '%') AND resource_manager = 'Heap';
+
+-- Test for filtering out WAL records based on resource_manager and
+-- record_type
+
+SELECT COUNT(*) >= 1 AS ok FROM pg_get_wal_records_content(:'wal_lsn1', :'wal_lsn2')
+			WHERE resource_manager = 'Heap' AND record_type = 'INSERT';
+
+\set VERBOSITY default
+
+-- Tests for permissions
+CREATE ROLE regress_pg_get_wal;
+SELECT has_function_privilege('regress_pg_get_wal',
+  'pg_get_wal_records_content(pg_lsn, pg_lsn) ', 'EXECUTE'); -- no
+
+-- Functions accessible by users with role pg_read_server_files.
+GRANT pg_read_server_files TO regress_pg_get_wal;
+
+SELECT has_function_privilege('regress_pg_get_wal',
+  'pg_get_wal_records_content(pg_lsn, pg_lsn) ', 'EXECUTE'); -- yes
+
+-- Superuser can grant execute to other users.
+GRANT EXECUTE ON FUNCTION pg_get_wal_records_content(pg_lsn, pg_lsn)
+  TO regress_pg_get_wal;
+
+SELECT has_function_privilege('regress_pg_get_wal',
+  'pg_get_wal_records_content(pg_lsn, pg_lsn) ', 'EXECUTE'); -- yes
+
+REVOKE EXECUTE ON FUNCTION pg_get_wal_records_content(pg_lsn, pg_lsn)
+  FROM regress_pg_get_wal;
+
+DROP ROLE regress_pg_get_wal;
-- 
2.34.1

From 2b522ec87876c141ae94915a398609ea9374f638 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Tue, 4 Apr 2023 05:49:34 +0000
Subject: [PATCH v18 1/4] pg_upgrade: Add --include-logical-replication-slots
 option

This commit introduces a new pg_upgrade option called "--include-logical-replication-slots".
This allows nodes with logical replication slots to be upgraded. The commit can
be divided into two parts: one for pg_dump and another for pg_upgrade.

For pg_dump this commit includes a new option called "--logical-replication-slots-only".
This option can be used to dump logical replication slots. When this option is
specified, the slot_name, plugin, and two_phase parameters are extracted from
pg_replication_slots. An SQL file that executes pg_create_logical_replication_slot()
with the extracted parameters is generated.

For pg_upgrade, when '--include-logical-replication-slots' is specified, it executes
pg_dump with the new "--logical-replication-slots-only" option and restores the slots
using the pg_create_logical_replication_slots() statements that the dump
generated (see above). Note that we cannot dump replication slots at the same time
as the schema dump because we need to separate the timing of restoring replication
slots and other objects. Replication slots, in  particular, should not be restored
before executing the pg_resetwal command because it will remove WALs that are
required by the slots.

The significant advantage of this commit is that it makes it easy to continue
logical replication even after upgrading the publisher node. Previously, pg_upgrade
allowed copying publications to a new node. With this new commit, adjusting the
connection string to the new publisher will cause the apply worker on the subscriber
to connect to the new publisher automatically. This enables seamless continuation
of logical replication, even after an upgrade.

Author: Hayato Kuroda
Reviewed-by: Peter Smith, Julien Rouhaud, Vignesh C, Wang Wei
---
 doc/src/sgml/ref/pgupgrade.sgml               |  39 +++++
 src/bin/pg_dump/pg_backup.h                   |   1 +
 src/bin/pg_dump/pg_dump.c                     | 143 +++++++++++++++++
 src/bin/pg_dump/pg_dump.h                     |  14 +-
 src/bin/pg_dump/pg_dump_sort.c                |  11 +-
 src/bin/pg_upgrade/Makefile                   |   3 +
 src/bin/pg_upgrade/check.c                    |  77 +++++++++
 src/bin/pg_upgrade/dump.c                     |  24 +++
 src/bin/pg_upgrade/info.c                     | 111 ++++++++++++-
 src/bin/pg_upgrade/meson.build                |   1 +
 src/bin/pg_upgrade/option.c                   |   7 +
 src/bin/pg_upgrade/pg_upgrade.c               |  61 ++++++++
 src/bin/pg_upgrade/pg_upgrade.h               |  21 +++
 .../t/003_logical_replication_slots.pl        | 146 ++++++++++++++++++
 src/tools/pgindent/typedefs.list              |   3 +
 15 files changed, 658 insertions(+), 4 deletions(-)
 create mode 100644 src/bin/pg_upgrade/t/003_logical_replication_slots.pl

diff --git a/doc/src/sgml/ref/pgupgrade.sgml b/doc/src/sgml/ref/pgupgrade.sgml
index 7816b4c685..9e62d6cb1d 100644
--- a/doc/src/sgml/ref/pgupgrade.sgml
+++ b/doc/src/sgml/ref/pgupgrade.sgml
@@ -240,6 +240,17 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>--include-logical-replication-slots</option></term>
+      <listitem>
+       <para>
+        Upgrade logical replication slots. Only permanent and usable
+        replication slots are included. Note that pg_upgrade does not check the
+        installation of plugins.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
       <term><option>-?</option></term>
       <term><option>--help</option></term>
@@ -402,6 +413,34 @@ NET STOP postgresql-&majorversion;
     </para>
    </step>
 
+   <step>
+    <title>Prepare for publisher upgrades</title>
+
+    <para>
+     If you are upgrading a publisher node, it is recommended to use the
+     <option>--include-logical-replication-slots</option> option. This helps
+     avoid the need for manually defining the same replication slot on the new
+     publisher.
+    </para>
+
+    <para>
+     Before you start upgrading the publisher node, ensure that the
+     subscription is temporarily disabled. After the upgrade is complete,
+     execute the
+     <link linkend="sql-altersubscription"><command>ALTER SUBSCRIPTION ... DISABLE</command></link>
+     command to update the connection string, and then re-enable the
+     subscription.
+    </para>
+
+    <para>
+     Note that only the replication slots with a wal_status different from
+     <literal>lost</literal> will be replicated to the new node. If there are
+     any subscriptions referring to a <literal>lost</literal> slot, they must
+     either be dropped or the slot_name should be changed to <literal>NONE</literal>
+     prior to upgrading.
+    </para>
+   </step>
+
    <step>
     <title>Run <application>pg_upgrade</application></title>
 
diff --git a/src/bin/pg_dump/pg_backup.h b/src/bin/pg_dump/pg_backup.h
index aba780ef4b..0a4e931f9b 100644
--- a/src/bin/pg_dump/pg_backup.h
+++ b/src/bin/pg_dump/pg_backup.h
@@ -187,6 +187,7 @@ typedef struct _dumpOptions
 	int			use_setsessauth;
 	int			enable_row_security;
 	int			load_via_partition_root;
+	int			logical_slots_only;
 
 	/* default, if no "inclusion" switches appear, is to dump everything */
 	bool		include_everything;
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index 5dab1ba9ea..c6d999b2cd 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -328,6 +328,9 @@ static void setupDumpWorker(Archive *AH);
 static TableInfo *getRootTableInfo(const TableInfo *tbinfo);
 static bool forcePartitionRootLoad(const TableInfo *tbinfo);
 
+static void getLogicalReplicationSlots(Archive *fout);
+static void dumpLogicalReplicationSlot(Archive *fout,
+									   const LogicalReplicationSlotInfo *slotinfo);
 
 int
 main(int argc, char **argv)
@@ -431,6 +434,7 @@ main(int argc, char **argv)
 		{"table-and-children", required_argument, NULL, 12},
 		{"exclude-table-and-children", required_argument, NULL, 13},
 		{"exclude-table-data-and-children", required_argument, NULL, 14},
+		{"logical-replication-slots-only", no_argument, NULL, 15},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -657,6 +661,10 @@ main(int argc, char **argv)
 										  optarg);
 				break;
 
+			case 15:			/* dump only replication slot(s) */
+				dopt.logical_slots_only = true;
+				break;
+
 			default:
 				/* getopt_long already emitted a complaint */
 				pg_log_error_hint("Try \"%s --help\" for more information.", progname);
@@ -714,6 +722,9 @@ main(int argc, char **argv)
 	if (dopt.do_nothing && dopt.dump_inserts == 0)
 		pg_fatal("option --on-conflict-do-nothing requires option --inserts, --rows-per-insert, or --column-inserts");
 
+	if (dopt.logical_slots_only && !dopt.binary_upgrade)
+		pg_fatal("options --logical-replication-slots-only requires option --binary-upgrade");
+
 	/* Identify archive format to emit */
 	archiveFormat = parseArchiveFormat(format, &archiveMode);
 
@@ -876,6 +887,16 @@ main(int argc, char **argv)
 			pg_fatal("no matching extensions were found");
 	}
 
+	/*
+	 * If dump logical-replication-slots-only was requested, dump only them
+	 * and skip everything else.
+	 */
+	if (dopt.logical_slots_only)
+	{
+		getLogicalReplicationSlots(fout);
+		goto dump;
+	}
+
 	/*
 	 * Dumping LOs is the default for dumps where an inclusion switch is not
 	 * used (an "include everything" dump).  -B can be used to exclude LOs
@@ -936,6 +957,8 @@ main(int argc, char **argv)
 	if (!dopt.no_security_labels)
 		collectSecLabels(fout);
 
+dump:
+
 	/* Lastly, create dummy objects to represent the section boundaries */
 	boundaryObjs = createBoundaryObjects();
 
@@ -1109,6 +1132,12 @@ help(const char *progname)
 			 "                               servers matching PATTERN\n"));
 	printf(_("  --inserts                    dump data as INSERT commands, rather than COPY\n"));
 	printf(_("  --load-via-partition-root    load partitions via the root table\n"));
+
+	/*
+	 * The option --logical-replication-slots-only is used only by pg_upgrade
+	 * and should not be called by users, which is why it is not exposed by
+	 * the help.
+	 */
 	printf(_("  --no-comments                do not dump comments\n"));
 	printf(_("  --no-publications            do not dump publications\n"));
 	printf(_("  --no-security-labels         do not dump security label assignments\n"));
@@ -10237,6 +10266,10 @@ dumpDumpableObject(Archive *fout, DumpableObject *dobj)
 		case DO_SUBSCRIPTION:
 			dumpSubscription(fout, (const SubscriptionInfo *) dobj);
 			break;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			dumpLogicalReplicationSlot(fout,
+									   (const LogicalReplicationSlotInfo *) dobj);
+			break;
 		case DO_PRE_DATA_BOUNDARY:
 		case DO_POST_DATA_BOUNDARY:
 			/* never dumped, nothing to do */
@@ -18218,6 +18251,7 @@ addBoundaryDependencies(DumpableObject **dobjs, int numObjs,
 			case DO_PUBLICATION_REL:
 			case DO_PUBLICATION_TABLE_IN_SCHEMA:
 			case DO_SUBSCRIPTION:
+			case DO_LOGICAL_REPLICATION_SLOT:
 				/* Post-data objects: must come after the post-data boundary */
 				addObjectDependency(dobj, postDataBound->dumpId);
 				break;
@@ -18479,3 +18513,112 @@ appendReloptionsArrayAH(PQExpBuffer buffer, const char *reloptions,
 	if (!res)
 		pg_log_warning("could not parse %s array", "reloptions");
 }
+
+/*
+ * getLogicalReplicationSlots
+ *	  get information about replication slots
+ */
+static void
+getLogicalReplicationSlots(Archive *fout)
+{
+	PGresult   *res;
+	LogicalReplicationSlotInfo *slotinfo;
+	PQExpBuffer query;
+
+	int			i_slotname;
+	int			i_plugin;
+	int			i_twophase;
+	int			i,
+				ntups;
+
+	/* Check whether we should dump or not */
+	if (fout->remoteVersion < 170000)
+		return;
+
+	Assert(fout->dopt->logical_slots_only);
+
+	query = createPQExpBuffer();
+
+	resetPQExpBuffer(query);
+
+	/*
+	 * Get replication slots.
+	 *
+	 * XXX: Which information must be extracted from old node? Currently three
+	 * attributes are extracted because they are used by
+	 * pg_create_logical_replication_slot().
+	 */
+	appendPQExpBufferStr(query,
+						 "SELECT slot_name, plugin, two_phase "
+						 "FROM pg_catalog.pg_replication_slots "
+						 "WHERE database = current_database() AND temporary = false "
+						 "AND wal_status IN ('reserved', 'extended');");
+
+	res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
+
+	ntups = PQntuples(res);
+
+	i_slotname = PQfnumber(res, "slot_name");
+	i_plugin = PQfnumber(res, "plugin");
+	i_twophase = PQfnumber(res, "two_phase");
+
+	slotinfo = pg_malloc(ntups * sizeof(LogicalReplicationSlotInfo));
+
+	for (i = 0; i < ntups; i++)
+	{
+		slotinfo[i].dobj.objType = DO_LOGICAL_REPLICATION_SLOT;
+
+		slotinfo[i].dobj.catId.tableoid = InvalidOid;
+		slotinfo[i].dobj.catId.oid = InvalidOid;
+		AssignDumpId(&slotinfo[i].dobj);
+
+		slotinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_slotname));
+
+		slotinfo[i].plugin = pg_strdup(PQgetvalue(res, i, i_plugin));
+		slotinfo[i].twophase = (strcmp(PQgetvalue(res, i, i_twophase), "t") == 0);
+
+		/*
+		 * Note: Currently we do not have any options to include/exclude slots
+		 * in dumping, so all the slots must be selected.
+		 */
+		slotinfo[i].dobj.dump = DUMP_COMPONENT_DEFINITION;
+	}
+	PQclear(res);
+
+	destroyPQExpBuffer(query);
+}
+
+/*
+ * dumpLogicalReplicationSlot
+ *	  dump creation functions for the given logical replication slots
+ */
+static void
+dumpLogicalReplicationSlot(Archive *fout,
+						   const LogicalReplicationSlotInfo *slotinfo)
+{
+	Assert(fout->dopt->logical_slots_only);
+
+	if (slotinfo->dobj.dump & DUMP_COMPONENT_DEFINITION)
+	{
+		PQExpBuffer query = createPQExpBuffer();
+
+		/*
+		 * XXX: For simplification, pg_create_logical_replication_slot() is
+		 * used. Is it sufficient?
+		 */
+		appendPQExpBuffer(query, "SELECT pg_catalog.pg_create_logical_replication_slot(");
+		appendStringLiteralAH(query, slotinfo->dobj.name, fout);
+		appendPQExpBuffer(query, ", ");
+		appendStringLiteralAH(query, slotinfo->plugin, fout);
+		appendPQExpBuffer(query, ", false, %s);",
+						  slotinfo->twophase ? "true" : "false");
+
+		ArchiveEntry(fout, slotinfo->dobj.catId, slotinfo->dobj.dumpId,
+					 ARCHIVE_OPTS(.tag = slotinfo->dobj.name,
+								  .description = "REPLICATION SLOT",
+								  .section = SECTION_POST_DATA,
+								  .createStmt = query->data));
+
+		destroyPQExpBuffer(query);
+	}
+}
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index bc8f2ec36d..ed1866d9ab 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -82,7 +82,8 @@ typedef enum
 	DO_PUBLICATION,
 	DO_PUBLICATION_REL,
 	DO_PUBLICATION_TABLE_IN_SCHEMA,
-	DO_SUBSCRIPTION
+	DO_SUBSCRIPTION,
+	DO_LOGICAL_REPLICATION_SLOT
 } DumpableObjectType;
 
 /*
@@ -667,6 +668,17 @@ typedef struct _SubscriptionInfo
 	char	   *subpasswordrequired;
 } SubscriptionInfo;
 
+/*
+ * The LogicalReplicationSlotInfo struct is used to represent replication
+ * slots.
+ */
+typedef struct _LogicalReplicationSlotInfo
+{
+	DumpableObject dobj;
+	char	   *plugin;
+	bool		twophase;
+} LogicalReplicationSlotInfo;
+
 /*
  *	common utility functions
  */
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index 523a19c155..ae65443228 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -93,6 +93,7 @@ enum dbObjectTypePriorities
 	PRIO_PUBLICATION_REL,
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,
 	PRIO_SUBSCRIPTION,
+	PRIO_LOGICAL_REPLICATION_SLOT,
 	PRIO_DEFAULT_ACL,			/* done in ACL pass */
 	PRIO_EVENT_TRIGGER,			/* must be next to last! */
 	PRIO_REFRESH_MATVIEW		/* must be last! */
@@ -146,10 +147,11 @@ static const int dbObjectTypePriority[] =
 	PRIO_PUBLICATION,			/* DO_PUBLICATION */
 	PRIO_PUBLICATION_REL,		/* DO_PUBLICATION_REL */
 	PRIO_PUBLICATION_TABLE_IN_SCHEMA,	/* DO_PUBLICATION_TABLE_IN_SCHEMA */
-	PRIO_SUBSCRIPTION			/* DO_SUBSCRIPTION */
+	PRIO_SUBSCRIPTION,			/* DO_SUBSCRIPTION */
+	PRIO_LOGICAL_REPLICATION_SLOT	/* DO_LOGICAL_REPLICATION_SLOT */
 };
 
-StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_SUBSCRIPTION + 1),
+StaticAssertDecl(lengthof(dbObjectTypePriority) == (DO_LOGICAL_REPLICATION_SLOT + 1),
 				 "array length mismatch");
 
 static DumpId preDataBoundId;
@@ -1542,6 +1544,11 @@ describeDumpableObject(DumpableObject *obj, char *buf, int bufsize)
 					 "SUBSCRIPTION (ID %d OID %u)",
 					 obj->dumpId, obj->catId.oid);
 			return;
+		case DO_LOGICAL_REPLICATION_SLOT:
+			snprintf(buf, bufsize,
+					 "LOGICAL REPLICATION SLOT (ID %d NAME %s)",
+					 obj->dumpId, obj->name);
+			return;
 		case DO_PRE_DATA_BOUNDARY:
 			snprintf(buf, bufsize,
 					 "PRE-DATA BOUNDARY  (ID %d)",
diff --git a/src/bin/pg_upgrade/Makefile b/src/bin/pg_upgrade/Makefile
index 5834513add..815d1a7ca1 100644
--- a/src/bin/pg_upgrade/Makefile
+++ b/src/bin/pg_upgrade/Makefile
@@ -3,6 +3,9 @@
 PGFILEDESC = "pg_upgrade - an in-place binary upgrade utility"
 PGAPPICON = win32
 
+# required for 003_logical_replication_slots.pl
+EXTRA_INSTALL=contrib/test_decoding
+
 subdir = src/bin/pg_upgrade
 top_builddir = ../../..
 include $(top_builddir)/src/Makefile.global
diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index 64024e3b9e..e7a42620f3 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -30,7 +30,9 @@ static void check_for_jsonb_9_4_usage(ClusterInfo *cluster);
 static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
+static void check_for_logical_replication_slots(ClusterInfo *new_cluster);
 
+static int	num_slots_on_old_cluster;
 
 /*
  * fix_path_separator
@@ -89,6 +91,10 @@ check_and_dump_old_cluster(bool live_check)
 	/* Extract a list of databases and tables from the old cluster */
 	get_db_and_rel_infos(&old_cluster);
 
+	/* Additionally, extract a list of logical replication slots if required */
+	if (user_opts.include_logical_slots)
+		num_slots_on_old_cluster = get_logical_slot_infos(&old_cluster);
+
 	init_tablespaces();
 
 	get_loadable_libraries();
@@ -189,6 +195,18 @@ check_new_cluster(void)
 {
 	get_db_and_rel_infos(&new_cluster);
 
+	/*
+	 * Do additional work if --include-logical-replication-slots was
+	 * specified. This must be done before check_new_cluster_is_empty()
+	 * because the slot_arr attribute of the new_cluster will be checked in
+	 * that function.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		(void) get_logical_slot_infos(&new_cluster);
+		check_for_logical_replication_slots(&new_cluster);
+	}
+
 	check_new_cluster_is_empty();
 
 	check_loadable_libraries();
@@ -364,6 +382,22 @@ check_new_cluster_is_empty(void)
 						 rel_arr->rels[relnum].nspname,
 						 rel_arr->rels[relnum].relname);
 		}
+
+		/*
+		 * If --include-logical-replication-slots is required, check the
+		 * existence of slots.
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			DbInfo	   *pDbInfo = &new_cluster.dbarr.dbs[dbnum];
+			LogicalSlotInfoArr *slot_arr = &pDbInfo->slot_arr;
+
+			/* if nslots > 0, report just first entry and exit */
+			if (slot_arr->nslots)
+				pg_fatal("New cluster database \"%s\" is not empty: found logical replication slot \"%s\"",
+						 pDbInfo->db_name,
+						 slot_arr->slots[0].slotname);
+		}
 	}
 }
 
@@ -1402,3 +1436,46 @@ check_for_user_defined_encoding_conversions(ClusterInfo *cluster)
 	else
 		check_ok();
 }
+
+/*
+ * Verify the parameter settings necessary for creating logical replication
+ * slots.
+ */
+static void
+check_for_logical_replication_slots(ClusterInfo *new_cluster)
+{
+	PGresult   *res;
+	PGconn	   *conn = connectToServer(new_cluster, "template1");
+	int			max_replication_slots;
+	char	   *wal_level;
+
+	/* --include-logical-replication-slots can be used since PG17. */
+	if (GET_MAJOR_VERSION(new_cluster->major_version) <= 1600)
+		return;
+
+	prep_status("Checking parameter settings for logical replication slots");
+
+	res = executeQueryOrDie(conn, "SHOW max_replication_slots;");
+	max_replication_slots = atoi(PQgetvalue(res, 0, 0));
+
+	if (max_replication_slots == 0)
+		pg_fatal("max_replication_slots must be greater than 0");
+	else if (num_slots_on_old_cluster > max_replication_slots)
+		pg_fatal("max_replication_slots must be greater than existing logical "
+				 "replication slots on old node.");
+
+	PQclear(res);
+
+	res = executeQueryOrDie(conn, "SHOW wal_level;");
+	wal_level = PQgetvalue(res, 0, 0);
+
+	if (strcmp(wal_level, "logical") != 0)
+		pg_fatal("wal_level must be \"logical\", but is set to \"%s\"",
+				 wal_level);
+
+	PQclear(res);
+
+	PQfinish(conn);
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/dump.c b/src/bin/pg_upgrade/dump.c
index 6c8c82dca8..e6b90864f5 100644
--- a/src/bin/pg_upgrade/dump.c
+++ b/src/bin/pg_upgrade/dump.c
@@ -59,6 +59,30 @@ generate_old_dump(void)
 						   log_opts.dumpdir,
 						   sql_file_name, escaped_connstr.data);
 
+		/*
+		 * Dump logical replication slots if needed.
+		 *
+		 * XXX We cannot dump replication slots at the same time as the schema
+		 * dump because we need to separate the timing of restoring
+		 * replication slots and other objects. Replication slots, in
+		 * particular, should not be restored before executing the pg_resetwal
+		 * command because it will remove WALs that are required by the slots.
+		 */
+		if (user_opts.include_logical_slots)
+		{
+			char		slots_file_name[MAXPGPATH];
+
+			snprintf(slots_file_name, sizeof(slots_file_name),
+					 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+			parallel_exec_prog(log_file_name, NULL,
+							   "\"%s/pg_dump\" %s --logical-replication-slots-only "
+							   "--quote-all-identifiers --binary-upgrade %s "
+							   "--file=\"%s/%s\" %s",
+							   new_cluster.bindir, cluster_conn_opts(&old_cluster),
+							   log_opts.verbose ? "--verbose" : "",
+							   log_opts.dumpdir,
+							   slots_file_name, escaped_connstr.data);
+		}
 		termPQExpBuffer(&escaped_connstr);
 	}
 
diff --git a/src/bin/pg_upgrade/info.c b/src/bin/pg_upgrade/info.c
index a9988abfe1..8bc0ad2e10 100644
--- a/src/bin/pg_upgrade/info.c
+++ b/src/bin/pg_upgrade/info.c
@@ -26,6 +26,7 @@ static void get_rel_infos(ClusterInfo *cluster, DbInfo *dbinfo);
 static void free_rel_infos(RelInfoArr *rel_arr);
 static void print_db_infos(DbInfoArr *db_arr);
 static void print_rel_infos(RelInfoArr *rel_arr);
+static void print_slot_infos(LogicalSlotInfoArr *slot_arr);
 
 
 /*
@@ -394,7 +395,7 @@ get_db_infos(ClusterInfo *cluster)
 	i_spclocation = PQfnumber(res, "spclocation");
 
 	ntups = PQntuples(res);
-	dbinfos = (DbInfo *) pg_malloc(sizeof(DbInfo) * ntups);
+	dbinfos = (DbInfo *) pg_malloc0(sizeof(DbInfo) * ntups);
 
 	for (tupnum = 0; tupnum < ntups; tupnum++)
 	{
@@ -600,6 +601,96 @@ get_rel_infos(ClusterInfo *cluster, DbInfo *dbinfo)
 	dbinfo->rel_arr.nrels = num_rels;
 }
 
+/*
+ * get_logical_slot_infos_per_db()
+ *
+ * gets the LogicalSlotInfos for all the logical replication slots of the database
+ * referred to by "dbinfo".
+ */
+static void
+get_logical_slot_infos_per_db(ClusterInfo *cluster, DbInfo *dbinfo)
+{
+	PGconn	   *conn = connectToServer(cluster,
+									   dbinfo->db_name);
+	PGresult   *res;
+	LogicalSlotInfo *slotinfos = NULL;
+
+	int			num_slots;
+
+	char		query[QUERY_ALLOC];
+
+	snprintf(query, sizeof(query),
+			 "SELECT slot_name, plugin, two_phase "
+			 "FROM pg_catalog.pg_replication_slots "
+			 "WHERE database = current_database() AND temporary = false "
+			 "AND wal_status IN ('reserved', 'extended');");
+
+	res = executeQueryOrDie(conn, "%s", query);
+
+	num_slots = PQntuples(res);
+
+	if (num_slots)
+	{
+		int			slotnum;
+		int			i_slotname;
+		int			i_plugin;
+		int			i_twophase;
+
+		slotinfos = (LogicalSlotInfo *) pg_malloc(sizeof(LogicalSlotInfo) * num_slots);
+
+		i_slotname = PQfnumber(res, "slot_name");
+		i_plugin = PQfnumber(res, "plugin");
+		i_twophase = PQfnumber(res, "two_phase");
+
+		for (slotnum = 0; slotnum < num_slots; slotnum++)
+		{
+			LogicalSlotInfo *curr = &slotinfos[slotnum];
+
+			curr->slotname = pg_strdup(PQgetvalue(res, slotnum, i_slotname));
+			curr->plugin = pg_strdup(PQgetvalue(res, slotnum, i_plugin));
+			curr->two_phase = (strcmp(PQgetvalue(res, slotnum, i_twophase), "t") == 0);
+		}
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	dbinfo->slot_arr.slots = slotinfos;
+	dbinfo->slot_arr.nslots = num_slots;
+}
+
+/*
+ * get_logical_slot_infos()
+ *
+ * Higher level routine to generate LogicalSlotInfoArr for all databases.
+ */
+int
+get_logical_slot_infos(ClusterInfo *cluster)
+{
+	int			dbnum;
+	int			slotnum = 0;
+
+	if (cluster == &old_cluster)
+		pg_log(PG_VERBOSE, "\nsource databases:");
+	else
+		pg_log(PG_VERBOSE, "\ntarget databases:");
+
+	for (dbnum = 0; dbnum < cluster->dbarr.ndbs; dbnum++)
+	{
+		DbInfo	   *pDbInfo = &cluster->dbarr.dbs[dbnum];
+
+		get_logical_slot_infos_per_db(cluster, pDbInfo);
+		slotnum += pDbInfo->slot_arr.nslots;
+
+		if (log_opts.verbose)
+		{
+			pg_log(PG_VERBOSE, "Database: \"%s\"", pDbInfo->db_name);
+			print_slot_infos(&pDbInfo->slot_arr);
+		}
+	}
+
+	return slotnum;
+}
 
 static void
 free_db_and_rel_infos(DbInfoArr *db_arr)
@@ -610,6 +701,12 @@ free_db_and_rel_infos(DbInfoArr *db_arr)
 	{
 		free_rel_infos(&db_arr->dbs[dbnum].rel_arr);
 		pg_free(db_arr->dbs[dbnum].db_name);
+
+		/*
+		 * Logical replication slots must not exist on the new cluster before
+		 * doing create_logical_replication_slots().
+		 */
+		Assert(db_arr->dbs[dbnum].slot_arr.slots == NULL);
 	}
 	pg_free(db_arr->dbs);
 	db_arr->dbs = NULL;
@@ -660,3 +757,15 @@ print_rel_infos(RelInfoArr *rel_arr)
 			   rel_arr->rels[relnum].reloid,
 			   rel_arr->rels[relnum].tablespace);
 }
+
+static void
+print_slot_infos(LogicalSlotInfoArr *slot_arr)
+{
+	int			slotnum;
+
+	for (slotnum = 0; slotnum < slot_arr->nslots; slotnum++)
+		pg_log(PG_VERBOSE, "slotname: \"%s\", plugin: \"%s\", two_phase: %d",
+			   slot_arr->slots[slotnum].slotname,
+			   slot_arr->slots[slotnum].plugin,
+			   slot_arr->slots[slotnum].two_phase);
+}
diff --git a/src/bin/pg_upgrade/meson.build b/src/bin/pg_upgrade/meson.build
index 12a97f84e2..228f29b688 100644
--- a/src/bin/pg_upgrade/meson.build
+++ b/src/bin/pg_upgrade/meson.build
@@ -42,6 +42,7 @@ tests += {
     'tests': [
       't/001_basic.pl',
       't/002_pg_upgrade.pl',
+      't/003_logical_replication_slots.pl',
     ],
     'test_kwargs': {'priority': 40}, # pg_upgrade tests are slow
   },
diff --git a/src/bin/pg_upgrade/option.c b/src/bin/pg_upgrade/option.c
index 640361009e..df66a5ffe6 100644
--- a/src/bin/pg_upgrade/option.c
+++ b/src/bin/pg_upgrade/option.c
@@ -57,6 +57,7 @@ parseCommandLine(int argc, char *argv[])
 		{"verbose", no_argument, NULL, 'v'},
 		{"clone", no_argument, NULL, 1},
 		{"copy", no_argument, NULL, 2},
+		{"include-logical-replication-slots", no_argument, NULL, 3},
 
 		{NULL, 0, NULL, 0}
 	};
@@ -199,6 +200,10 @@ parseCommandLine(int argc, char *argv[])
 				user_opts.transfer_mode = TRANSFER_MODE_COPY;
 				break;
 
+			case 3:
+				user_opts.include_logical_slots = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						os_info.progname);
@@ -289,6 +294,8 @@ usage(void)
 	printf(_("  -V, --version                 display version information, then exit\n"));
 	printf(_("  --clone                       clone instead of copying files to new cluster\n"));
 	printf(_("  --copy                        copy files to new cluster (default)\n"));
+	printf(_("  --include-logical-replication-slots\n"
+			 "                                upgrade logical replication slots\n"));
 	printf(_("  -?, --help                    show this help, then exit\n"));
 	printf(_("\n"
 			 "Before running pg_upgrade you must:\n"
diff --git a/src/bin/pg_upgrade/pg_upgrade.c b/src/bin/pg_upgrade/pg_upgrade.c
index 4562dafcff..6dd3832422 100644
--- a/src/bin/pg_upgrade/pg_upgrade.c
+++ b/src/bin/pg_upgrade/pg_upgrade.c
@@ -59,6 +59,7 @@ static void copy_xact_xlog_xid(void);
 static void set_frozenxids(bool minmxid_only);
 static void make_outputdirs(char *pgdata);
 static void setup(char *argv0, bool *live_check);
+static void create_logical_replication_slots(void);
 
 ClusterInfo old_cluster,
 			new_cluster;
@@ -188,6 +189,19 @@ main(int argc, char **argv)
 			  new_cluster.pgdata);
 	check_ok();
 
+	/*
+	 * Create logical replication slots if requested.
+	 *
+	 * Note: This must be done after doing pg_resetwal command because
+	 * pg_resetwal would remove required WALs.
+	 */
+	if (user_opts.include_logical_slots)
+	{
+		start_postmaster(&new_cluster, true);
+		create_logical_replication_slots();
+		stop_postmaster(false);
+	}
+
 	if (user_opts.do_sync)
 	{
 		prep_status("Sync data directory to disk");
@@ -860,3 +874,50 @@ set_frozenxids(bool minmxid_only)
 
 	check_ok();
 }
+
+/*
+ * create_logical_replication_slots()
+ *
+ * Similar to create_new_objects() but only restores logical replication slots.
+ */
+static void
+create_logical_replication_slots(void)
+{
+	int			dbnum;
+
+	prep_status_progress("Restoring logical replication slots in the new cluster");
+
+	for (dbnum = 0; dbnum < old_cluster.dbarr.ndbs; dbnum++)
+	{
+		char		slots_file_name[MAXPGPATH],
+					log_file_name[MAXPGPATH];
+		DbInfo	   *old_db = &old_cluster.dbarr.dbs[dbnum];
+
+		pg_log(PG_STATUS, "%s", old_db->db_name);
+
+		snprintf(slots_file_name, sizeof(slots_file_name),
+				 DB_DUMP_LOGICAL_SLOTS_FILE_MASK, old_db->db_oid);
+		snprintf(log_file_name, sizeof(log_file_name),
+				 DB_DUMP_LOG_FILE_MASK, old_db->db_oid);
+
+		parallel_exec_prog(log_file_name,
+						   NULL,
+						   "\"%s/psql\" %s --echo-queries --set ON_ERROR_STOP=on "
+						   "--no-psqlrc --dbname %s -f \"%s/%s\"",
+						   new_cluster.bindir,
+						   cluster_conn_opts(&new_cluster),
+						   old_db->db_name,
+						   log_opts.dumpdir,
+						   slots_file_name);
+	}
+
+	/* reap all children */
+	while (reap_child(true) == true)
+		;
+
+	end_progress_output();
+	check_ok();
+
+	/* update new_cluster info again */
+	get_logical_slot_infos(&new_cluster);
+}
diff --git a/src/bin/pg_upgrade/pg_upgrade.h b/src/bin/pg_upgrade/pg_upgrade.h
index 3eea0139c7..8034067492 100644
--- a/src/bin/pg_upgrade/pg_upgrade.h
+++ b/src/bin/pg_upgrade/pg_upgrade.h
@@ -29,6 +29,7 @@
 /* contains both global db information and CREATE DATABASE commands */
 #define GLOBALS_DUMP_FILE	"pg_upgrade_dump_globals.sql"
 #define DB_DUMP_FILE_MASK	"pg_upgrade_dump_%u.custom"
+#define DB_DUMP_LOGICAL_SLOTS_FILE_MASK	"pg_upgrade_dump_%u_logical_slots.sql"
 
 /*
  * Base directories that include all the files generated internally, from the
@@ -150,6 +151,22 @@ typedef struct
 	int			nrels;
 } RelInfoArr;
 
+/*
+ * Structure to store logical replication slot information
+ */
+typedef struct
+{
+	char	   *slotname;		/* slot name */
+	char	   *plugin;			/* plugin */
+	bool		two_phase;		/* Can the slot decode 2PC? */
+} LogicalSlotInfo;
+
+typedef struct
+{
+	int			nslots;
+	LogicalSlotInfo *slots;
+} LogicalSlotInfoArr;
+
 /*
  * The following structure represents a relation mapping.
  */
@@ -176,6 +193,7 @@ typedef struct
 	char		db_tablespace[MAXPGPATH];	/* database default tablespace
 											 * path */
 	RelInfoArr	rel_arr;		/* array of all user relinfos */
+	LogicalSlotInfoArr slot_arr;	/* array of all LogicalSlotInfo */
 } DbInfo;
 
 /*
@@ -304,6 +322,8 @@ typedef struct
 	transferMode transfer_mode; /* copy files or link them? */
 	int			jobs;			/* number of processes/threads to use */
 	char	   *socketdir;		/* directory to use for Unix sockets */
+	bool		include_logical_slots;	/* true -> dump and restore logical
+										 * replication slots */
 } UserOpts;
 
 typedef struct
@@ -400,6 +420,7 @@ FileNameMap *gen_db_file_maps(DbInfo *old_db,
 							  DbInfo *new_db, int *nmaps, const char *old_pgdata,
 							  const char *new_pgdata);
 void		get_db_and_rel_infos(ClusterInfo *cluster);
+int			get_logical_slot_infos(ClusterInfo *cluster);
 
 /* option.c */
 
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
new file mode 100644
index 0000000000..9ca266f6b2
--- /dev/null
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -0,0 +1,146 @@
+# Copyright (c) 2023, PostgreSQL Global Development Group
+
+# Tests for upgrading replication slots
+
+use strict;
+use warnings;
+
+use File::Path qw(rmtree);
+
+use PostgreSQL::Test::Cluster;
+use PostgreSQL::Test::Utils;
+use Test::More;
+
+# Can be changed to test the other modes
+my $mode = $ENV{PG_TEST_PG_UPGRADE_MODE} || '--copy';
+
+# Initialize old node
+my $old_node = PostgreSQL::Test::Cluster->new('old_node');
+$old_node->init(allows_streaming => 'logical');
+$old_node->start;
+
+# Initialize new node
+my $new_node = PostgreSQL::Test::Cluster->new('new_node');
+$new_node->init(allows_streaming => 1);
+
+my $bindir = $new_node->config_data('--bindir');
+
+$old_node->stop;
+
+# Cause a failure at the start of pg_upgrade because wal_level is replica
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with wrong wal_level');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+# Preparations for the subsequent test. The case max_replication_slots is set
+# to 0 is prohibited.
+$new_node->append_conf('postgresql.conf', "wal_level = 'logical'");
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 0");
+
+# Cause a failure at the start of pg_upgrade because max_replication_slots is 0
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with wrong max_replication_slots');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+# Preparations for the subsequent test. max_replication_slots is set to
+# non-zero value
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 1");
+
+# Create a slot on old node, and generate WALs
+$old_node->start;
+$old_node->safe_psql(
+	'postgres', qq[
+	SELECT pg_create_logical_replication_slot('test_slot', 'test_decoding', false, true);
+	SELECT pg_create_logical_replication_slot('to_be_dropped', 'test_decoding', false, true);
+	CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a;
+]);
+
+$old_node->stop;
+
+# Cause a failure at the start of pg_upgrade because max_replication_slots is
+# smaller than existing slots on old node
+command_fails(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots',
+	],
+	'run of pg_upgrade of old node with small max_replication_slots');
+ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
+
+# Clean up
+rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+
+# Preparations for the subsequent test. max_replication_slots is set to
+# appropriate value
+$new_node->append_conf('postgresql.conf', "max_replication_slots = 10");
+
+# Remove an unnecessary slot and consume WALs
+$old_node->start;
+$old_node->safe_psql(
+	'postgres', qq[
+	SELECT pg_drop_replication_slot('to_be_dropped');
+	SELECT count(*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)
+]);
+$old_node->stop;
+
+# Actual run, pg_upgrade_output.d is removed at the end
+command_ok(
+	[
+		'pg_upgrade', '--no-sync',
+		'-d',         $old_node->data_dir,
+		'-D',         $new_node->data_dir,
+		'-b',         $bindir,
+		'-B',         $bindir,
+		'-s',         $new_node->host,
+		'-p',         $old_node->port,
+		'-P',         $new_node->port,
+		$mode,        '--include-logical-replication-slots'
+	],
+	'run of pg_upgrade of old node');
+ok( !-d $new_node->data_dir . "/pg_upgrade_output.d",
+	"pg_upgrade_output.d/ removed after pg_upgrade success");
+
+$new_node->start;
+my $result = $new_node->safe_psql('postgres',
+	"SELECT slot_name, two_phase FROM pg_replication_slots");
+is($result, qq(test_slot|t), 'check the slot exists on new node');
+
+done_testing();
diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list
index 66823bc2a7..5a34e351ef 100644
--- a/src/tools/pgindent/typedefs.list
+++ b/src/tools/pgindent/typedefs.list
@@ -1500,7 +1500,10 @@ LogicalRepStreamAbortData
 LogicalRepTupleData
 LogicalRepTyp
 LogicalRepWorker
+LogicalReplicationSlotInfo
 LogicalRewriteMappingData
+LogicalSlotInfo
+LogicalSlotInfoArr
 LogicalTape
 LogicalTapeSet
 LsnReadQueue
-- 
2.27.0

From e83458e9238639da95c32d3902ccd5926ff97413 Mon Sep 17 00:00:00 2001
From: Julien Rouhaud <julien.rouhaud@free.fr>
Date: Fri, 14 Apr 2023 13:49:09 +0800
Subject: [PATCH v18 2/4] Always persist to disk logical slots during a
 shutdown checkpoint.

It's entirely possible for a logical slot to have a confirmed_flush_lsn higher
than the last value saved on disk while not being marked as dirty.  It's
currently not a problem to lose that value during a clean shutdown / restart
cycle, but a later patch adding support for pg_upgrade of publications and
logical slots will rely on that value being properly persisted to disk.

Author: Julien Rouhaud
Reviewed-by: Wang Wei
Discussion: FIXME
---
 src/backend/access/transam/xlog.c |  2 +-
 src/backend/replication/slot.c    | 26 ++++++++++++++++----------
 src/include/replication/slot.h    |  2 +-
 3 files changed, 18 insertions(+), 12 deletions(-)

diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 60c0b7ec3a..6dced61cf4 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7026,7 +7026,7 @@ static void
 CheckPointGuts(XLogRecPtr checkPointRedo, int flags)
 {
 	CheckPointRelationMap();
-	CheckPointReplicationSlots();
+	CheckPointReplicationSlots(flags & CHECKPOINT_IS_SHUTDOWN);
 	CheckPointSnapBuild();
 	CheckPointLogicalRewriteHeap();
 	CheckPointReplicationOrigin();
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index 1dc27264f6..5aed7cd190 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -109,7 +109,8 @@ static void ReplicationSlotDropPtr(ReplicationSlot *slot);
 /* internal persistency functions */
 static void RestoreSlotFromDisk(const char *name);
 static void CreateSlotOnDisk(ReplicationSlot *slot);
-static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel);
+static void SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+						   bool is_shutdown);
 
 /*
  * Report shared-memory space needed by ReplicationSlotsShmemInit.
@@ -783,7 +784,7 @@ ReplicationSlotSave(void)
 	Assert(MyReplicationSlot != NULL);
 
 	sprintf(path, "pg_replslot/%s", NameStr(MyReplicationSlot->data.name));
-	SaveSlotToPath(MyReplicationSlot, path, ERROR);
+	SaveSlotToPath(MyReplicationSlot, path, ERROR, false);
 }
 
 /*
@@ -1565,11 +1566,10 @@ restart:
 /*
  * Flush all replication slots to disk.
  *
- * This needn't actually be part of a checkpoint, but it's a convenient
- * location.
+ * is_shutdown is true in case of a shutdown checkpoint.
  */
 void
-CheckPointReplicationSlots(void)
+CheckPointReplicationSlots(bool is_shutdown)
 {
 	int			i;
 
@@ -1594,7 +1594,7 @@ CheckPointReplicationSlots(void)
 
 		/* save the slot to disk, locking is handled in SaveSlotToPath() */
 		sprintf(path, "pg_replslot/%s", NameStr(s->data.name));
-		SaveSlotToPath(s, path, LOG);
+		SaveSlotToPath(s, path, LOG, is_shutdown);
 	}
 	LWLockRelease(ReplicationSlotAllocationLock);
 }
@@ -1700,7 +1700,7 @@ CreateSlotOnDisk(ReplicationSlot *slot)
 
 	/* Write the actual state file. */
 	slot->dirty = true;			/* signal that we really need to write */
-	SaveSlotToPath(slot, tmppath, ERROR);
+	SaveSlotToPath(slot, tmppath, ERROR, false);
 
 	/* Rename the directory into place. */
 	if (rename(tmppath, path) != 0)
@@ -1726,7 +1726,8 @@ CreateSlotOnDisk(ReplicationSlot *slot)
  * Shared functionality between saving and creating a replication slot.
  */
 static void
-SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
+SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel,
+			   bool is_shutdown)
 {
 	char		tmppath[MAXPGPATH];
 	char		path[MAXPGPATH];
@@ -1740,8 +1741,13 @@ SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
 	slot->just_dirtied = false;
 	SpinLockRelease(&slot->mutex);
 
-	/* and don't do anything if there's nothing to write */
-	if (!was_dirty)
+	/*
+	 * and don't do anything if there's nothing to write, unless it's this is
+	 * called for a logical slot during a shutdown checkpoint, as we want to
+	 * persist the confirmed_flush_lsn in that case, even if that's the only
+	 * modification.
+	 */
+	if (!was_dirty && (SlotIsPhysical(slot) || !is_shutdown))
 		return;
 
 	LWLockAcquire(&slot->io_in_progress_lock, LW_EXCLUSIVE);
diff --git a/src/include/replication/slot.h b/src/include/replication/slot.h
index a8a89dc784..7ca37c9f70 100644
--- a/src/include/replication/slot.h
+++ b/src/include/replication/slot.h
@@ -241,7 +241,7 @@ extern void ReplicationSlotNameForTablesync(Oid suboid, Oid relid, char *syncslo
 extern void ReplicationSlotDropAtPubNode(WalReceiverConn *wrconn, char *slotname, bool missing_ok);
 
 extern void StartupReplicationSlots(void);
-extern void CheckPointReplicationSlots(void);
+extern void CheckPointReplicationSlots(bool is_shutdown);
 
 extern void CheckSlotRequirements(void);
 extern void CheckSlotPermissions(void);
-- 
2.27.0

From e1cbde65108376c61377f12089bdc3990b53330e Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Tue, 1 Aug 2023 14:04:04 +0000
Subject: [PATCH v18 3/4] Move pg_get_wal_record_info functionality from
 pg_walinspect to core

Upgrade of publications required pg_get_wal_record_info to check that there are
no WAL records other than CHECKPOINT_SHUTDOWN WAL record to be consumed. Hence
moved pg_get_wal_record_info functionality as pg_get_wal_record_content
to core so that it can be called from pg_upgrade.
---
 contrib/pg_walinspect/Makefile                |   2 +-
 contrib/pg_walinspect/meson.build             |   1 +
 .../pg_walinspect/pg_walinspect--1.1--1.2.sql |  31 +++
 contrib/pg_walinspect/pg_walinspect.c         | 180 +-----------------
 contrib/pg_walinspect/pg_walinspect.control   |   2 +-
 src/backend/access/transam/xlogfuncs.c        |  50 +++++
 src/backend/access/transam/xlogutils.c        | 173 +++++++++++++++++
 src/backend/catalog/system_functions.sql      |   4 +
 src/include/access/xlogutils.h                |   6 +
 src/include/catalog/pg_proc.dat               |   9 +
 src/test/regress/expected/misc_functions.out  |  18 ++
 src/test/regress/sql/misc_functions.sql       |  16 ++
 12 files changed, 313 insertions(+), 179 deletions(-)
 create mode 100644 contrib/pg_walinspect/pg_walinspect--1.1--1.2.sql

diff --git a/contrib/pg_walinspect/Makefile b/contrib/pg_walinspect/Makefile
index 22090f7716..5cc7d81b42 100644
--- a/contrib/pg_walinspect/Makefile
+++ b/contrib/pg_walinspect/Makefile
@@ -7,7 +7,7 @@ OBJS = \
 PGFILEDESC = "pg_walinspect - functions to inspect contents of PostgreSQL Write-Ahead Log"
 
 EXTENSION = pg_walinspect
-DATA = pg_walinspect--1.0.sql pg_walinspect--1.0--1.1.sql
+DATA = pg_walinspect--1.0.sql pg_walinspect--1.0--1.1.sql pg_walinspect--1.1--1.2.sql
 
 REGRESS = pg_walinspect oldextversions
 
diff --git a/contrib/pg_walinspect/meson.build b/contrib/pg_walinspect/meson.build
index 80059f6119..8f7a99a493 100644
--- a/contrib/pg_walinspect/meson.build
+++ b/contrib/pg_walinspect/meson.build
@@ -20,6 +20,7 @@ install_data(
   'pg_walinspect.control',
   'pg_walinspect--1.0.sql',
   'pg_walinspect--1.0--1.1.sql',
+  'pg_walinspect--1.1--1.2.sql',
   kwargs: contrib_data_args,
 )
 
diff --git a/contrib/pg_walinspect/pg_walinspect--1.1--1.2.sql b/contrib/pg_walinspect/pg_walinspect--1.1--1.2.sql
new file mode 100644
index 0000000000..48e2da3034
--- /dev/null
+++ b/contrib/pg_walinspect/pg_walinspect--1.1--1.2.sql
@@ -0,0 +1,31 @@
+/* contrib/pg_walinspect/pg_walinspect--1.1--1.2.sql */
+
+-- complain if script is sourced in psql, rather than via ALTER EXTENSION
+\echo Use "ALTER EXTENSION pg_walinspect UPDATE TO '1.2'" to load this file. \quit
+
+-- The function is now in the backend and callers should update to use those.
+
+ALTER EXTENSION pg_walinspect DROP FUNCTION pg_get_wal_record_info;
+DROP FUNCTION pg_get_wal_record_info(pg_lsn);
+
+--
+-- pg_get_wal_record_info()
+--
+CREATE FUNCTION pg_get_wal_record_info(IN in_lsn pg_lsn,
+    OUT start_lsn pg_lsn,
+    OUT end_lsn pg_lsn,
+    OUT prev_lsn pg_lsn,
+    OUT xid xid,
+    OUT resource_manager text,
+    OUT record_type text,
+    OUT record_length int4,
+    OUT main_data_length int4,
+    OUT fpi_length int4,
+    OUT description text,
+    OUT block_ref text
+)
+AS 'pg_get_wal_record_content'
+LANGUAGE INTERNAL STRICT PARALLEL SAFE;
+
+REVOKE EXECUTE ON FUNCTION pg_get_wal_record_info(pg_lsn) FROM PUBLIC;
+GRANT EXECUTE ON FUNCTION pg_get_wal_record_info(pg_lsn) TO pg_read_server_files;
diff --git a/contrib/pg_walinspect/pg_walinspect.c b/contrib/pg_walinspect/pg_walinspect.c
index 796a74f322..8cba14e789 100644
--- a/contrib/pg_walinspect/pg_walinspect.c
+++ b/contrib/pg_walinspect/pg_walinspect.c
@@ -39,11 +39,6 @@ PG_FUNCTION_INFO_V1(pg_get_wal_stats);
 PG_FUNCTION_INFO_V1(pg_get_wal_stats_till_end_of_wal);
 
 static void ValidateInputLSNs(XLogRecPtr start_lsn, XLogRecPtr *end_lsn);
-static XLogRecPtr GetCurrentLSN(void);
-static XLogReaderState *InitXLogReaderState(XLogRecPtr lsn);
-static XLogRecord *ReadNextXLogRecord(XLogReaderState *xlogreader);
-static void GetWALRecordInfo(XLogReaderState *record, Datum *values,
-							 bool *nulls, uint32 ncols);
 static void GetWALRecordsInfo(FunctionCallInfo fcinfo,
 							  XLogRecPtr start_lsn,
 							  XLogRecPtr end_lsn);
@@ -62,178 +57,6 @@ static void GetWalStats(FunctionCallInfo fcinfo,
 static void GetWALBlockInfo(FunctionCallInfo fcinfo, XLogReaderState *record,
 							bool show_data);
 
-/*
- * Return the LSN up to which the server has WAL.
- */
-static XLogRecPtr
-GetCurrentLSN(void)
-{
-	XLogRecPtr	curr_lsn;
-
-	/*
-	 * We determine the current LSN of the server similar to how page_read
-	 * callback read_local_xlog_page_no_wait does.
-	 */
-	if (!RecoveryInProgress())
-		curr_lsn = GetFlushRecPtr(NULL);
-	else
-		curr_lsn = GetXLogReplayRecPtr(NULL);
-
-	Assert(!XLogRecPtrIsInvalid(curr_lsn));
-
-	return curr_lsn;
-}
-
-/*
- * Initialize WAL reader and identify first valid LSN.
- */
-static XLogReaderState *
-InitXLogReaderState(XLogRecPtr lsn)
-{
-	XLogReaderState *xlogreader;
-	ReadLocalXLogPageNoWaitPrivate *private_data;
-	XLogRecPtr	first_valid_record;
-
-	/*
-	 * Reading WAL below the first page of the first segments isn't allowed.
-	 * This is a bootstrap WAL page and the page_read callback fails to read
-	 * it.
-	 */
-	if (lsn < XLOG_BLCKSZ)
-		ereport(ERROR,
-				(errmsg("could not read WAL at LSN %X/%X",
-						LSN_FORMAT_ARGS(lsn))));
-
-	private_data = (ReadLocalXLogPageNoWaitPrivate *)
-		palloc0(sizeof(ReadLocalXLogPageNoWaitPrivate));
-
-	xlogreader = XLogReaderAllocate(wal_segment_size, NULL,
-									XL_ROUTINE(.page_read = &read_local_xlog_page_no_wait,
-											   .segment_open = &wal_segment_open,
-											   .segment_close = &wal_segment_close),
-									private_data);
-
-	if (xlogreader == NULL)
-		ereport(ERROR,
-				(errcode(ERRCODE_OUT_OF_MEMORY),
-				 errmsg("out of memory"),
-				 errdetail("Failed while allocating a WAL reading processor.")));
-
-	/* first find a valid recptr to start from */
-	first_valid_record = XLogFindNextRecord(xlogreader, lsn);
-
-	if (XLogRecPtrIsInvalid(first_valid_record))
-		ereport(ERROR,
-				(errmsg("could not find a valid record after %X/%X",
-						LSN_FORMAT_ARGS(lsn))));
-
-	return xlogreader;
-}
-
-/*
- * Read next WAL record.
- *
- * By design, to be less intrusive in a running system, no slot is allocated
- * to reserve the WAL we're about to read. Therefore this function can
- * encounter read errors for historical WAL.
- *
- * We guard against ordinary errors trying to read WAL that hasn't been
- * written yet by limiting end_lsn to the flushed WAL, but that can also
- * encounter errors if the flush pointer falls in the middle of a record. In
- * that case we'll return NULL.
- */
-static XLogRecord *
-ReadNextXLogRecord(XLogReaderState *xlogreader)
-{
-	XLogRecord *record;
-	char	   *errormsg;
-
-	record = XLogReadRecord(xlogreader, &errormsg);
-
-	if (record == NULL)
-	{
-		ReadLocalXLogPageNoWaitPrivate *private_data;
-
-		/* return NULL, if end of WAL is reached */
-		private_data = (ReadLocalXLogPageNoWaitPrivate *)
-			xlogreader->private_data;
-
-		if (private_data->end_of_wal)
-			return NULL;
-
-		if (errormsg)
-			ereport(ERROR,
-					(errcode_for_file_access(),
-					 errmsg("could not read WAL at %X/%X: %s",
-							LSN_FORMAT_ARGS(xlogreader->EndRecPtr), errormsg)));
-		else
-			ereport(ERROR,
-					(errcode_for_file_access(),
-					 errmsg("could not read WAL at %X/%X",
-							LSN_FORMAT_ARGS(xlogreader->EndRecPtr))));
-	}
-
-	return record;
-}
-
-/*
- * Output values that make up a row describing caller's WAL record.
- *
- * This function leaks memory.  Caller may need to use its own custom memory
- * context.
- *
- * Keep this in sync with GetWALBlockInfo.
- */
-static void
-GetWALRecordInfo(XLogReaderState *record, Datum *values,
-				 bool *nulls, uint32 ncols)
-{
-	const char *record_type;
-	RmgrData	desc;
-	uint32		fpi_len = 0;
-	StringInfoData rec_desc;
-	StringInfoData rec_blk_ref;
-	int			i = 0;
-
-	desc = GetRmgr(XLogRecGetRmid(record));
-	record_type = desc.rm_identify(XLogRecGetInfo(record));
-
-	if (record_type == NULL)
-		record_type = psprintf("UNKNOWN (%x)", XLogRecGetInfo(record) & ~XLR_INFO_MASK);
-
-	initStringInfo(&rec_desc);
-	desc.rm_desc(&rec_desc, record);
-
-	if (XLogRecHasAnyBlockRefs(record))
-	{
-		initStringInfo(&rec_blk_ref);
-		XLogRecGetBlockRefInfo(record, false, true, &rec_blk_ref, &fpi_len);
-	}
-
-	values[i++] = LSNGetDatum(record->ReadRecPtr);
-	values[i++] = LSNGetDatum(record->EndRecPtr);
-	values[i++] = LSNGetDatum(XLogRecGetPrev(record));
-	values[i++] = TransactionIdGetDatum(XLogRecGetXid(record));
-	values[i++] = CStringGetTextDatum(desc.rm_name);
-	values[i++] = CStringGetTextDatum(record_type);
-	values[i++] = UInt32GetDatum(XLogRecGetTotalLen(record));
-	values[i++] = UInt32GetDatum(XLogRecGetDataLen(record));
-	values[i++] = UInt32GetDatum(fpi_len);
-
-	if (rec_desc.len > 0)
-		values[i++] = CStringGetTextDatum(rec_desc.data);
-	else
-		nulls[i++] = true;
-
-	if (XLogRecHasAnyBlockRefs(record))
-		values[i++] = CStringGetTextDatum(rec_blk_ref.data);
-	else
-		nulls[i++] = true;
-
-	Assert(i == ncols);
-}
-
-
 /*
  * Output one or more rows in rsinfo tuple store, each describing a single
  * block reference from caller's WAL record. (Should only be called with
@@ -454,6 +277,9 @@ pg_get_wal_block_info(PG_FUNCTION_ARGS)
 
 /*
  * Get WAL record info.
+ *
+ * Note that this function have been removed in newer versions in 1.2, but they
+ * are kept around for compatibility.
  */
 Datum
 pg_get_wal_record_info(PG_FUNCTION_ARGS)
diff --git a/contrib/pg_walinspect/pg_walinspect.control b/contrib/pg_walinspect/pg_walinspect.control
index efa3cb2cfe..5f574b865b 100644
--- a/contrib/pg_walinspect/pg_walinspect.control
+++ b/contrib/pg_walinspect/pg_walinspect.control
@@ -1,5 +1,5 @@
 # pg_walinspect extension
 comment = 'functions to inspect contents of PostgreSQL Write-Ahead Log'
-default_version = '1.1'
+default_version = '1.2'
 module_pathname = '$libdir/pg_walinspect'
 relocatable = true
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 5044ff0643..71ee1069f6 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -754,3 +754,53 @@ pg_promote(PG_FUNCTION_ARGS)
 						   wait_seconds)));
 	PG_RETURN_BOOL(false);
 }
+
+/*
+ * Get WAL record content.
+ */
+Datum
+pg_get_wal_record_content(PG_FUNCTION_ARGS)
+{
+#define PG_GET_WAL_RECORD_CONTENT_COLS 11
+	Datum		result;
+	Datum		values[PG_GET_WAL_RECORD_CONTENT_COLS] = {0};
+	bool		nulls[PG_GET_WAL_RECORD_CONTENT_COLS] = {0};
+	XLogRecPtr	lsn;
+	XLogRecPtr	curr_lsn;
+	XLogReaderState *xlogreader;
+	TupleDesc	tupdesc;
+	HeapTuple	tuple;
+
+	lsn = PG_GETARG_LSN(0);
+	curr_lsn = GetCurrentLSN();
+
+	if (lsn > curr_lsn)
+		ereport(ERROR,
+				(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
+				 errmsg("WAL input LSN must be less than current LSN"),
+				 errdetail("Current WAL LSN on the database system is at %X/%X.",
+						   LSN_FORMAT_ARGS(curr_lsn))));
+
+	/* Build a tuple descriptor for our result type. */
+	if (get_call_result_type(fcinfo, NULL, &tupdesc) != TYPEFUNC_COMPOSITE)
+		elog(ERROR, "return type must be a row type");
+
+	xlogreader = InitXLogReaderState(lsn);
+
+	if (!ReadNextXLogRecord(xlogreader))
+		ereport(ERROR,
+				(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
+				 errmsg("could not read WAL at %X/%X",
+						LSN_FORMAT_ARGS(xlogreader->EndRecPtr))));
+
+	GetWALRecordInfo(xlogreader, values, nulls, PG_GET_WAL_RECORD_CONTENT_COLS);
+
+	pfree(xlogreader->private_data);
+	XLogReaderFree(xlogreader);
+
+	tuple = heap_form_tuple(tupdesc, values, nulls);
+	result = HeapTupleGetDatum(tuple);
+
+	PG_RETURN_DATUM(result);
+#undef PG_GET_WAL_RECORD_CONTENT_COLS
+}
diff --git a/src/backend/access/transam/xlogutils.c b/src/backend/access/transam/xlogutils.c
index e174a2a891..f637028b8e 100644
--- a/src/backend/access/transam/xlogutils.c
+++ b/src/backend/access/transam/xlogutils.c
@@ -28,8 +28,10 @@
 #include "pgstat.h"
 #include "storage/fd.h"
 #include "storage/smgr.h"
+#include "utils/builtins.h"
 #include "utils/guc.h"
 #include "utils/hsearch.h"
+#include "utils/pg_lsn.h"
 #include "utils/rel.h"
 
 
@@ -1048,3 +1050,174 @@ WALReadRaiseError(WALReadError *errinfo)
 						errinfo->wre_req)));
 	}
 }
+
+/*
+ * Return the LSN up to which the server has WAL.
+ */
+XLogRecPtr
+GetCurrentLSN(void)
+{
+	XLogRecPtr	curr_lsn;
+
+	/*
+	 * We determine the current LSN of the server similar to how page_read
+	 * callback read_local_xlog_page_no_wait does.
+	 */
+	if (!RecoveryInProgress())
+		curr_lsn = GetFlushRecPtr(NULL);
+	else
+		curr_lsn = GetXLogReplayRecPtr(NULL);
+
+	Assert(!XLogRecPtrIsInvalid(curr_lsn));
+
+	return curr_lsn;
+}
+
+/*
+ * Initialize WAL reader and identify first valid LSN.
+ */
+XLogReaderState *
+InitXLogReaderState(XLogRecPtr lsn)
+{
+	XLogReaderState *xlogreader;
+	ReadLocalXLogPageNoWaitPrivate *private_data;
+	XLogRecPtr	first_valid_record;
+
+	/*
+	 * Reading WAL below the first page of the first segments isn't allowed.
+	 * This is a bootstrap WAL page and the page_read callback fails to read
+	 * it.
+	 */
+	if (lsn < XLOG_BLCKSZ)
+		ereport(ERROR,
+				(errmsg("could not read WAL at LSN %X/%X",
+						LSN_FORMAT_ARGS(lsn))));
+
+	private_data = (ReadLocalXLogPageNoWaitPrivate *)
+		palloc0(sizeof(ReadLocalXLogPageNoWaitPrivate));
+
+	xlogreader = XLogReaderAllocate(wal_segment_size, NULL,
+									XL_ROUTINE(.page_read = &read_local_xlog_page_no_wait,
+											   .segment_open = &wal_segment_open,
+											   .segment_close = &wal_segment_close),
+									private_data);
+
+	if (xlogreader == NULL)
+		ereport(ERROR,
+				(errcode(ERRCODE_OUT_OF_MEMORY),
+				 errmsg("out of memory"),
+				 errdetail("Failed while allocating a WAL reading processor.")));
+
+	/* first find a valid recptr to start from */
+	first_valid_record = XLogFindNextRecord(xlogreader, lsn);
+
+	if (XLogRecPtrIsInvalid(first_valid_record))
+		ereport(ERROR,
+				(errmsg("could not find a valid record after %X/%X",
+						LSN_FORMAT_ARGS(lsn))));
+
+	return xlogreader;
+}
+
+/*
+ * Read next WAL record.
+ *
+ * By design, to be less intrusive in a running system, no slot is allocated
+ * to reserve the WAL we're about to read. Therefore this function can
+ * encounter read errors for historical WAL.
+ *
+ * We guard against ordinary errors trying to read WAL that hasn't been
+ * written yet by limiting end_lsn to the flushed WAL, but that can also
+ * encounter errors if the flush pointer falls in the middle of a record. In
+ * that case we'll return NULL.
+ */
+XLogRecord *
+ReadNextXLogRecord(XLogReaderState *xlogreader)
+{
+	XLogRecord *record;
+	char	   *errormsg;
+
+	record = XLogReadRecord(xlogreader, &errormsg);
+
+	if (record == NULL)
+	{
+		ReadLocalXLogPageNoWaitPrivate *private_data;
+
+		/* return NULL, if end of WAL is reached */
+		private_data = (ReadLocalXLogPageNoWaitPrivate *)
+			xlogreader->private_data;
+
+		if (private_data->end_of_wal)
+			return NULL;
+
+		if (errormsg)
+			ereport(ERROR,
+					(errcode_for_file_access(),
+					 errmsg("could not read WAL at %X/%X: %s",
+							LSN_FORMAT_ARGS(xlogreader->EndRecPtr), errormsg)));
+		else
+			ereport(ERROR,
+					(errcode_for_file_access(),
+					 errmsg("could not read WAL at %X/%X",
+							LSN_FORMAT_ARGS(xlogreader->EndRecPtr))));
+	}
+
+	return record;
+}
+
+/*
+ * Output values that make up a row describing caller's WAL record.
+ *
+ * This function leaks memory.  Caller may need to use its own custom memory
+ * context.
+ *
+ * Keep this in sync with GetWALBlockInfo.
+ */
+void
+GetWALRecordInfo(XLogReaderState *record, Datum *values,
+				 bool *nulls, uint32 ncols)
+{
+	const char *record_type;
+	RmgrData	desc;
+	uint32		fpi_len = 0;
+	StringInfoData rec_desc;
+	StringInfoData rec_blk_ref;
+	int			i = 0;
+
+	desc = GetRmgr(XLogRecGetRmid(record));
+	record_type = desc.rm_identify(XLogRecGetInfo(record));
+
+	if (record_type == NULL)
+		record_type = psprintf("UNKNOWN (%x)", XLogRecGetInfo(record) & ~XLR_INFO_MASK);
+
+	initStringInfo(&rec_desc);
+	desc.rm_desc(&rec_desc, record);
+
+	if (XLogRecHasAnyBlockRefs(record))
+	{
+		initStringInfo(&rec_blk_ref);
+		XLogRecGetBlockRefInfo(record, false, true, &rec_blk_ref, &fpi_len);
+	}
+
+	values[i++] = LSNGetDatum(record->ReadRecPtr);
+	values[i++] = LSNGetDatum(record->EndRecPtr);
+	values[i++] = LSNGetDatum(XLogRecGetPrev(record));
+	values[i++] = TransactionIdGetDatum(XLogRecGetXid(record));
+	values[i++] = CStringGetTextDatum(desc.rm_name);
+	values[i++] = CStringGetTextDatum(record_type);
+	values[i++] = UInt32GetDatum(XLogRecGetTotalLen(record));
+	values[i++] = UInt32GetDatum(XLogRecGetDataLen(record));
+	values[i++] = UInt32GetDatum(fpi_len);
+
+	if (rec_desc.len > 0)
+		values[i++] = CStringGetTextDatum(rec_desc.data);
+	else
+		nulls[i++] = true;
+
+	if (XLogRecHasAnyBlockRefs(record))
+		values[i++] = CStringGetTextDatum(rec_blk_ref.data);
+	else
+		nulls[i++] = true;
+
+	Assert(i == ncols);
+}
diff --git a/src/backend/catalog/system_functions.sql b/src/backend/catalog/system_functions.sql
index 07c0d89c4f..0186d594fb 100644
--- a/src/backend/catalog/system_functions.sql
+++ b/src/backend/catalog/system_functions.sql
@@ -616,6 +616,8 @@ REVOKE EXECUTE ON FUNCTION pg_backup_stop(boolean) FROM public;
 
 REVOKE EXECUTE ON FUNCTION pg_create_restore_point(text) FROM public;
 
+REVOKE EXECUTE ON FUNCTION pg_get_wal_record_content(pg_lsn) FROM public;
+
 REVOKE EXECUTE ON FUNCTION pg_switch_wal() FROM public;
 
 REVOKE EXECUTE ON FUNCTION pg_log_standby_snapshot() FROM public;
@@ -726,6 +728,8 @@ REVOKE EXECUTE ON FUNCTION pg_ls_replslotdir(text) FROM PUBLIC;
 -- We also set up some things as accessible to standard roles.
 --
 
+GRANT EXECUTE ON FUNCTION pg_get_wal_record_content(pg_lsn) TO pg_read_server_files;
+
 GRANT EXECUTE ON FUNCTION pg_ls_logdir() TO pg_monitor;
 
 GRANT EXECUTE ON FUNCTION pg_ls_waldir() TO pg_monitor;
diff --git a/src/include/access/xlogutils.h b/src/include/access/xlogutils.h
index 5b77b11f50..95ad84ac6f 100644
--- a/src/include/access/xlogutils.h
+++ b/src/include/access/xlogutils.h
@@ -115,4 +115,10 @@ extern void XLogReadDetermineTimeline(XLogReaderState *state,
 
 extern void WALReadRaiseError(WALReadError *errinfo);
 
+extern XLogRecPtr GetCurrentLSN(void);
+extern XLogReaderState *InitXLogReaderState(XLogRecPtr lsn);
+extern XLogRecord *ReadNextXLogRecord(XLogReaderState *xlogreader);
+extern void GetWALRecordInfo(XLogReaderState *record, Datum *values,
+							 bool *nulls, uint32 ncols);
+
 #endif
diff --git a/src/include/catalog/pg_proc.dat b/src/include/catalog/pg_proc.dat
index 6996073989..ffe146ba84 100644
--- a/src/include/catalog/pg_proc.dat
+++ b/src/include/catalog/pg_proc.dat
@@ -6489,6 +6489,15 @@
   proargnames => '{rm_id, rm_name, rm_builtin}',
   prosrc => 'pg_get_wal_resource_managers' },
 
+
+{ oid => '8045', descr => 'Info of the WAL conent',
+  proname => 'pg_get_wal_record_content', prorows => '1', proretset => 't',
+  provolatile => 's', prorettype => 'record', proargtypes => 'pg_lsn',
+  proallargtypes => '{pg_lsn,pg_lsn,pg_lsn,pg_lsn,xid,text,text,int4,int4,int4,text,text}',
+  proargmodes => '{i,o,o,o,o,o,o,o,o,o,o,o}',
+  proargnames => '{in_lsn,start_lsn,end_lsn,prev_lsn,xid,resource_manager,record_type,record_length,main_data_length,fpi_length,description,block_ref}',
+  prosrc => 'pg_get_wal_record_content' },
+
 { oid => '2621', descr => 'reload configuration files',
   proname => 'pg_reload_conf', provolatile => 'v', prorettype => 'bool',
   proargtypes => '', prosrc => 'pg_reload_conf' },
diff --git a/src/test/regress/expected/misc_functions.out b/src/test/regress/expected/misc_functions.out
index c669948370..bf2d1a5f9b 100644
--- a/src/test/regress/expected/misc_functions.out
+++ b/src/test/regress/expected/misc_functions.out
@@ -642,3 +642,21 @@ SELECT segment_number > 0 AS ok_segment_number, timeline_id
  t                 |  4294967295
 (1 row)
 
+-- pg_get_wal_record_content
+CREATE TABLE sample_tbl(col1 int, col2 int);
+SELECT pg_current_wal_lsn() AS wal_lsn \gset
+INSERT INTO sample_tbl SELECT * FROM generate_series(1, 2);
+-- Mask DETAIL messages as these could refer to current LSN positions.
+\set VERBOSITY terse
+-- Invalid input LSN.
+SELECT * FROM pg_get_wal_record_content('0/0');
+ERROR:  could not read WAL at LSN 0/0
+-- LSNs with the highest value possible.
+SELECT * FROM pg_get_wal_record_content('FFFFFFFF/FFFFFFFF');
+ERROR:  WAL input LSN must be less than current LSN
+SELECT COUNT(*) >= 1 AS ok FROM pg_get_wal_record_content(:'wal_lsn');
+ ok 
+----
+ t
+(1 row)
+
diff --git a/src/test/regress/sql/misc_functions.sql b/src/test/regress/sql/misc_functions.sql
index b57f01f3e9..4ff19b4927 100644
--- a/src/test/regress/sql/misc_functions.sql
+++ b/src/test/regress/sql/misc_functions.sql
@@ -237,3 +237,19 @@ SELECT segment_number > 0 AS ok_segment_number, timeline_id
   FROM pg_split_walfile_name('000000010000000100000000');
 SELECT segment_number > 0 AS ok_segment_number, timeline_id
   FROM pg_split_walfile_name('ffffffFF00000001000000af');
+
+-- pg_get_wal_record_content
+CREATE TABLE sample_tbl(col1 int, col2 int);
+SELECT pg_current_wal_lsn() AS wal_lsn \gset
+INSERT INTO sample_tbl SELECT * FROM generate_series(1, 2);
+
+-- Mask DETAIL messages as these could refer to current LSN positions.
+\set VERBOSITY terse
+
+-- Invalid input LSN.
+SELECT * FROM pg_get_wal_record_content('0/0');
+
+-- LSNs with the highest value possible.
+SELECT * FROM pg_get_wal_record_content('FFFFFFFF/FFFFFFFF');
+
+SELECT COUNT(*) >= 1 AS ok FROM pg_get_wal_record_content(:'wal_lsn');
-- 
2.27.0

From 0d6ed85edf1fbfbf9dca62a70d0a06fe6f4ccc98 Mon Sep 17 00:00:00 2001
From: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Date: Fri, 14 Apr 2023 08:59:03 +0000
Subject: [PATCH v18 4/4] pg_upgrade: Add check function for
 --include-logical-replication-slots option

XXX: Actually, this commit disallows to support slots which are created by user
backends. In the checking function we ensure that all the avtive slots have
confirmed_flush_lsn which is same as current WAL position, and they would not be
the same. For slots which are used by logical replication, logical walsenders
guarantee that at the shutdown. For individual slots, however, cannot be handled
by walsenders, so confirmed_flush_lsn is behind shutdown checkpoint record.

Author: Hayato Kuroda
Reviewed-by: Wang Wei, Vignesh C
---
 src/bin/pg_upgrade/check.c                    |  59 ++++++++
 .../t/003_logical_replication_slots.pl        | 134 +++++++++++-------
 2 files changed, 140 insertions(+), 53 deletions(-)

diff --git a/src/bin/pg_upgrade/check.c b/src/bin/pg_upgrade/check.c
index e7a42620f3..b5eb1f8ccf 100644
--- a/src/bin/pg_upgrade/check.c
+++ b/src/bin/pg_upgrade/check.c
@@ -31,6 +31,7 @@ static void check_for_pg_role_prefix(ClusterInfo *cluster);
 static void check_for_new_tablespace_dir(ClusterInfo *new_cluster);
 static void check_for_user_defined_encoding_conversions(ClusterInfo *cluster);
 static void check_for_logical_replication_slots(ClusterInfo *new_cluster);
+static void check_for_confirmed_flush_lsn(ClusterInfo *cluster);
 
 static int	num_slots_on_old_cluster;
 
@@ -109,6 +110,8 @@ check_and_dump_old_cluster(bool live_check)
 	check_for_composite_data_type_usage(&old_cluster);
 	check_for_reg_data_type_usage(&old_cluster);
 	check_for_isn_and_int8_passing_mismatch(&old_cluster);
+	if (user_opts.include_logical_slots)
+		check_for_confirmed_flush_lsn(&old_cluster);
 
 	/*
 	 * PG 16 increased the size of the 'aclitem' type, which breaks the
@@ -1479,3 +1482,59 @@ check_for_logical_replication_slots(ClusterInfo *new_cluster)
 
 	check_ok();
 }
+
+/*
+ * Verify that all logical replication slots consumed all WALs, except a
+ * CHECKPOINT_SHUTDOWN record.
+ */
+static void
+check_for_confirmed_flush_lsn(ClusterInfo *cluster)
+{
+	int			i,
+				ntups,
+				i_slotname;
+	bool		is_error = false;
+	PGresult   *res;
+	DbInfo	   *active_db = &cluster->dbarr.dbs[0];
+	PGconn	   *conn = connectToServer(cluster, active_db->db_name);
+
+	Assert(user_opts.include_logical_slots);
+
+	/* --include-logical-replication-slots can be used since PG17. */
+	if (GET_MAJOR_VERSION(cluster->major_version) <= 1600)
+		return;
+
+	prep_status("Checking confirmed_flush_lsn for logical replication slots");
+
+	/*
+	 * Check that all logical replication slots have reached the current WAL
+	 * position.
+	 */
+	res = executeQueryOrDie(conn,
+							"SELECT slot_name FROM pg_catalog.pg_replication_slots "
+							"WHERE (SELECT count(record_type) "
+							"		FROM pg_catalog.pg_get_wal_record_content(confirmed_flush_lsn) "
+							"		WHERE record_type != 'CHECKPOINT_SHUTDOWN') <> 0 "
+							"AND temporary = false AND wal_status IN ('reserved', 'extended', 'unreserved');");
+
+	ntups = PQntuples(res);
+	i_slotname = PQfnumber(res, "slot_name");
+
+	for (i = 0; i < ntups; i++)
+	{
+		is_error = true;
+
+		pg_log(PG_WARNING,
+			   "\nWARNING: logical replication slot \"%s\" has not consumed WALs yet",
+			   PQgetvalue(res, i, i_slotname));
+	}
+
+	PQclear(res);
+	PQfinish(conn);
+
+	if (is_error)
+		pg_fatal("--include-logical-replication-slots requires that all "
+				 "logical replication slots consumed all the WALs");
+
+	check_ok();
+}
diff --git a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
index 9ca266f6b2..a6a20f61e0 100644
--- a/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
+++ b/src/bin/pg_upgrade/t/003_logical_replication_slots.pl
@@ -15,132 +15,160 @@ use Test::More;
 my $mode = $ENV{PG_TEST_PG_UPGRADE_MODE} || '--copy';
 
 # Initialize old node
-my $old_node = PostgreSQL::Test::Cluster->new('old_node');
-$old_node->init(allows_streaming => 'logical');
-$old_node->start;
+my $old_publisher = PostgreSQL::Test::Cluster->new('old_publisher');
+$old_publisher->init(allows_streaming => 'logical');
 
 # Initialize new node
-my $new_node = PostgreSQL::Test::Cluster->new('new_node');
-$new_node->init(allows_streaming => 1);
+my $new_publisher = PostgreSQL::Test::Cluster->new('new_publisher');
+$new_publisher->init(allows_streaming => 1);
 
-my $bindir = $new_node->config_data('--bindir');
+# Initialize subscriber node
+my $subscriber = PostgreSQL::Test::Cluster->new('subscriber');
+$subscriber->init(allows_streaming => 'logical');
 
-$old_node->stop;
+my $bindir = $new_publisher->config_data('--bindir');
 
 # Cause a failure at the start of pg_upgrade because wal_level is replica
 command_fails(
 	[
 		'pg_upgrade', '--no-sync',
-		'-d',         $old_node->data_dir,
-		'-D',         $new_node->data_dir,
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
 		'-b',         $bindir,
 		'-B',         $bindir,
-		'-s',         $new_node->host,
-		'-p',         $old_node->port,
-		'-P',         $new_node->port,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
 		$mode,        '--include-logical-replication-slots',
 	],
 	'run of pg_upgrade of old node with wrong wal_level');
-ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+ok( -d $new_publisher->data_dir . "/pg_upgrade_output.d",
 	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
 
 # Clean up
-rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+rmtree($new_publisher->data_dir . "/pg_upgrade_output.d");
 
 # Preparations for the subsequent test. The case max_replication_slots is set
 # to 0 is prohibited.
-$new_node->append_conf('postgresql.conf', "wal_level = 'logical'");
-$new_node->append_conf('postgresql.conf', "max_replication_slots = 0");
+$new_publisher->append_conf('postgresql.conf', "wal_level = 'logical'");
+$new_publisher->append_conf('postgresql.conf', "max_replication_slots = 0");
 
 # Cause a failure at the start of pg_upgrade because max_replication_slots is 0
 command_fails(
 	[
 		'pg_upgrade', '--no-sync',
-		'-d',         $old_node->data_dir,
-		'-D',         $new_node->data_dir,
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
 		'-b',         $bindir,
 		'-B',         $bindir,
-		'-s',         $new_node->host,
-		'-p',         $old_node->port,
-		'-P',         $new_node->port,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
 		$mode,        '--include-logical-replication-slots',
 	],
 	'run of pg_upgrade of old node with wrong max_replication_slots');
-ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+ok( -d $new_publisher->data_dir . "/pg_upgrade_output.d",
 	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
 
 # Clean up
-rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+rmtree($new_publisher->data_dir . "/pg_upgrade_output.d");
 
 # Preparations for the subsequent test. max_replication_slots is set to
 # non-zero value
-$new_node->append_conf('postgresql.conf', "max_replication_slots = 1");
+$new_publisher->append_conf('postgresql.conf', "max_replication_slots = 1");
 
-# Create a slot on old node, and generate WALs
-$old_node->start;
-$old_node->safe_psql(
+# Create a slot on old node
+$old_publisher->start;
+$old_publisher->safe_psql(
 	'postgres', qq[
-	SELECT pg_create_logical_replication_slot('test_slot', 'test_decoding', false, true);
-	SELECT pg_create_logical_replication_slot('to_be_dropped', 'test_decoding', false, true);
+	SELECT pg_create_logical_replication_slot('test_slot1', 'test_decoding', false, true);
+	SELECT pg_create_logical_replication_slot('test_slot2', 'test_decoding', false, true);
 	CREATE TABLE tbl AS SELECT generate_series(1, 10) AS a;
 ]);
 
-$old_node->stop;
+$old_publisher->stop;
 
 # Cause a failure at the start of pg_upgrade because max_replication_slots is
 # smaller than existing slots on old node
 command_fails(
 	[
 		'pg_upgrade', '--no-sync',
-		'-d',         $old_node->data_dir,
-		'-D',         $new_node->data_dir,
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
 		'-b',         $bindir,
 		'-B',         $bindir,
-		'-s',         $new_node->host,
-		'-p',         $old_node->port,
-		'-P',         $new_node->port,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
 		$mode,        '--include-logical-replication-slots',
 	],
 	'run of pg_upgrade of old node with small max_replication_slots');
-ok( -d $new_node->data_dir . "/pg_upgrade_output.d",
+ok( -d $new_publisher->data_dir . "/pg_upgrade_output.d",
 	"pg_upgrade_output.d/ not removed after pg_upgrade failure");
 
 # Clean up
-rmtree($new_node->data_dir . "/pg_upgrade_output.d");
+rmtree($new_publisher->data_dir . "/pg_upgrade_output.d");
 
 # Preparations for the subsequent test. max_replication_slots is set to
 # appropriate value
-$new_node->append_conf('postgresql.conf', "max_replication_slots = 10");
+$new_publisher->append_conf('postgresql.conf', "max_replication_slots = 10");
 
-# Remove an unnecessary slot and consume WALs
-$old_node->start;
-$old_node->safe_psql(
+$old_publisher->start;
+$old_publisher->safe_psql(
 	'postgres', qq[
-	SELECT pg_drop_replication_slot('to_be_dropped');
-	SELECT count(*) FROM pg_logical_slot_get_changes('test_slot', NULL, NULL)
+	SELECT pg_drop_replication_slot('test_slot1');
+	SELECT pg_drop_replication_slot('test_slot2');
 ]);
-$old_node->stop;
+
+# Setup logical replication
+my $old_connstr = $old_publisher->connstr . ' dbname=postgres';
+$old_publisher->safe_psql('postgres', "CREATE PUBLICATION pub FOR ALL TABLES");
+$subscriber->start;
+$subscriber->safe_psql(
+	'postgres', qq[
+	CREATE TABLE tbl (a int);
+	CREATE SUBSCRIPTION sub CONNECTION '$old_connstr' PUBLICATION pub WITH (two_phase = 'true')
+]);
+
+# Wait for initial table sync to finish
+$subscriber->wait_for_subscription_sync($old_publisher, 'sub');
+
+$subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub DISABLE");
+$old_publisher->stop;
 
 # Actual run, pg_upgrade_output.d is removed at the end
 command_ok(
 	[
 		'pg_upgrade', '--no-sync',
-		'-d',         $old_node->data_dir,
-		'-D',         $new_node->data_dir,
+		'-d',         $old_publisher->data_dir,
+		'-D',         $new_publisher->data_dir,
 		'-b',         $bindir,
 		'-B',         $bindir,
-		'-s',         $new_node->host,
-		'-p',         $old_node->port,
-		'-P',         $new_node->port,
+		'-s',         $new_publisher->host,
+		'-p',         $old_publisher->port,
+		'-P',         $new_publisher->port,
 		$mode,        '--include-logical-replication-slots'
 	],
 	'run of pg_upgrade of old node');
-ok( !-d $new_node->data_dir . "/pg_upgrade_output.d",
+ok( !-d $new_publisher->data_dir . "/pg_upgrade_output.d",
 	"pg_upgrade_output.d/ removed after pg_upgrade success");
 
-$new_node->start;
-my $result = $new_node->safe_psql('postgres',
+$new_publisher->start;
+my $result = $new_publisher->safe_psql('postgres',
 	"SELECT slot_name, two_phase FROM pg_replication_slots");
-is($result, qq(test_slot|t), 'check the slot exists on new node');
+is($result, qq(sub|t), 'check the slot exists on new node');
+
+# Change the connection
+my $new_connstr = $new_publisher->connstr . ' dbname=postgres';
+$subscriber->safe_psql('postgres',
+	"ALTER SUBSCRIPTION sub CONNECTION '$new_connstr'");
+$subscriber->safe_psql('postgres', "ALTER SUBSCRIPTION sub ENABLE");
+
+# Check whether changes on the new publisher get replicated to the subscriber
+$new_publisher->safe_psql('postgres',
+	"INSERT INTO tbl VALUES (generate_series(11, 20))");
+$new_publisher->wait_for_catchup('sub');
+$result = $subscriber->safe_psql('postgres', "SELECT count(*) FROM tbl");
+is($result, qq(20), 'check changes are shipped to subscriber');
 
 done_testing();
-- 
2.27.0