unique indexes on partitioned tables
Started by Alvaro Herreraabout 8 years ago11 messages
This is the patch series for UNIQUE / PRIMARY KEY indexes on partitioned
tables. This is on top of the patch in
/messages/by-id/20171229175930.3aew7lzwd5w6m2x6@alvherre.pgsql
but I included it here as 0001 for simplicity. (Don't review that patch
in this thread please). This is essentially the same patch I posted
elsewhere in that thread.
I included Amit's support for ON CONFLICT DO UPDATE, but as I mentioned
in the other thread, it has a small bug. In principle we could push
0002 together with 0003, but I'd rather fix 0004 first and push it all
as one commit.
This serves as basis to build foreign keys on top; I'll post that
separately.
--
�lvaro Herrera https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
Attachments:
v1-0001-Local-partitioned-indexes.patchtext/plain; charset=us-asciiDownload
From 72a7e75e39161cd5da50d8543dd4de314a25cc03 Mon Sep 17 00:00:00 2001
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
Date: Mon, 23 Oct 2017 10:18:38 +0200
Subject: [PATCH v1 1/4] Local partitioned indexes
When CREATE INDEX is run on a partitioned table, create catalog entries
for an index on the partitioned table (which is just a placeholder since
the table proper has no data of its own), and recurse to create actual
indexes on the existing partitions; create them in future partitions
also.
As a convenience gadget, if the new index definition matches some
existing index in partitions, these are picked up and used instead of
creating new ones. Whichever way these indexes come about, they become
attached to the index on the parent table and are dropped alongside it,
and cannot be dropped on isolation unless they are detached first.
To support pg_dump'ing these indexes, add commands
CREATE INDEX ON ONLY <table>
(which creates the index on the parent partitioned table, without
recursing) and
ALTER INDEX ATTACH PARTITION
(which is used after the indexes have been created individually on each
partition, to attach them to the parent index). These reconstruct prior
database state exactly.
Reviewed-by: Robert Haas, Amit Langote, Jesper Pedersen, Simon Riggs, David Rowley
Discussion: https://postgr.es/m/20171113170646.gzweigyrgg6pwsg4@alvherre.pgsql
---
doc/src/sgml/catalogs.sgml | 18 ++
doc/src/sgml/ref/alter_index.sgml | 14 +
doc/src/sgml/ref/alter_table.sgml | 8 +-
doc/src/sgml/ref/create_index.sgml | 33 +-
doc/src/sgml/ref/reindex.sgml | 5 +
src/backend/access/common/reloptions.c | 1 +
src/backend/access/heap/heapam.c | 9 +-
src/backend/access/index/indexam.c | 3 +-
src/backend/bootstrap/bootparse.y | 2 +
src/backend/catalog/aclchk.c | 9 +-
src/backend/catalog/dependency.c | 9 +-
src/backend/catalog/heap.c | 1 +
src/backend/catalog/index.c | 133 +++++++-
src/backend/catalog/objectaddress.c | 5 +-
src/backend/catalog/pg_depend.c | 13 +-
src/backend/catalog/toasting.c | 2 +
src/backend/commands/indexcmds.c | 263 +++++++++++++++-
src/backend/commands/tablecmds.c | 487 +++++++++++++++++++++++++++++-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/nodes/outfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 10 +-
src/backend/parser/gram.y | 32 +-
src/backend/parser/parse_utilcmd.c | 65 ++--
src/backend/tcop/utility.c | 9 +
src/backend/utils/adt/amutils.c | 3 +-
src/backend/utils/adt/ruleutils.c | 17 +-
src/backend/utils/cache/relcache.c | 30 +-
src/bin/pg_dump/common.c | 104 +++++++
src/bin/pg_dump/pg_dump.c | 88 +++++-
src/bin/pg_dump/pg_dump.h | 11 +
src/bin/pg_dump/pg_dump_sort.c | 56 +++-
src/bin/psql/describe.c | 20 +-
src/bin/psql/tab-complete.c | 34 ++-
src/include/catalog/catversion.h | 2 +-
src/include/catalog/dependency.h | 8 +
src/include/catalog/index.h | 7 +
src/include/catalog/indexing.h | 2 +
src/include/catalog/pg_class.h | 1 +
src/include/catalog/pg_index.h | 38 +--
src/include/commands/defrem.h | 3 +-
src/include/nodes/execnodes.h | 1 +
src/include/nodes/parsenodes.h | 7 +-
src/include/parser/parse_utilcmd.h | 3 +
src/test/regress/expected/alter_table.out | 65 +++-
src/test/regress/expected/indexing.out | 471 +++++++++++++++++++++++++++++
src/test/regress/parallel_schedule | 2 +-
src/test/regress/serial_schedule | 1 +
src/test/regress/sql/alter_table.sql | 16 +
src/test/regress/sql/indexing.sql | 199 ++++++++++++
50 files changed, 2181 insertions(+), 142 deletions(-)
create mode 100644 src/test/regress/expected/indexing.out
create mode 100644 src/test/regress/sql/indexing.sql
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3f02202caf..5db63985b2 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -2996,6 +2996,24 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
</varlistentry>
<varlistentry>
+ <term><symbol>DEPENDENCY_INTERNAL_AUTO</symbol> (<literal>I</literal>)</term>
+ <listitem>
+ <para>
+ The dependent object was created as part of creation of the
+ referenced object, and is really just a part of its internal
+ implementation. A <command>DROP</command> of the dependent object
+ will be disallowed outright (we'll tell the user to issue a
+ <command>DROP</command> against the referenced object, instead).
+ As opposed to <literal>DEPENDENCY_INTERNAL</literal>, [... explain the
+ difference ].
+ Example: a index
+ on a partition is made internal-auto-dependent on both the partition
+ itself as well as on the index on the partitioned table.
+ </para>
+ </listitem>
+ </varlistentry>
+
+ <varlistentry>
<term><symbol>DEPENDENCY_EXTENSION</symbol> (<literal>e</literal>)</term>
<listitem>
<para>
diff --git a/doc/src/sgml/ref/alter_index.sgml b/doc/src/sgml/ref/alter_index.sgml
index e54237272c..3984686d67 100644
--- a/doc/src/sgml/ref/alter_index.sgml
+++ b/doc/src/sgml/ref/alter_index.sgml
@@ -23,6 +23,7 @@ PostgreSQL documentation
<synopsis>
ALTER INDEX [ IF EXISTS ] <replaceable class="parameter">name</replaceable> RENAME TO <replaceable class="parameter">new_name</replaceable>
ALTER INDEX [ IF EXISTS ] <replaceable class="parameter">name</replaceable> SET TABLESPACE <replaceable class="parameter">tablespace_name</replaceable>
+ALTER INDEX <replaceable class="parameter">name</replaceable> ATTACH PARTITION <replaceable class="parameter">index_name</replaceable>
ALTER INDEX <replaceable class="parameter">name</replaceable> DEPENDS ON EXTENSION <replaceable class="parameter">extension_name</replaceable>
ALTER INDEX [ IF EXISTS ] <replaceable class="parameter">name</replaceable> SET ( <replaceable class="parameter">storage_parameter</replaceable> = <replaceable class="parameter">value</replaceable> [, ... ] )
ALTER INDEX [ IF EXISTS ] <replaceable class="parameter">name</replaceable> RESET ( <replaceable class="parameter">storage_parameter</replaceable> [, ... ] )
@@ -76,6 +77,19 @@ ALTER INDEX ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
</varlistentry>
<varlistentry>
+ <term><literal>ATTACH</literal></term>
+ <listitem>
+ <para>
+ Causes the named index to become attached to the altered index.
+ The named index must be on a partition of the table containing the
+ index being altered, and have an equivalent definition. An attached
+ index cannot be dropped by itself, and will automatically be dropped
+ if its parent index is dropped.
+ </para>
+ </listitem>
+ </varlistentry>
+
+ <varlistentry>
<term><literal>DEPENDS ON EXTENSION</literal></term>
<listitem>
<para>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index 7bcf242846..0a2f3e3646 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -783,7 +783,10 @@ ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
as a partition of the target table. The table can be attached
as a partition for specific values using <literal>FOR VALUES
</literal> or as a default partition by using <literal>DEFAULT
- </literal>.
+ </literal>. For each index in the target table, a corresponding
+ one will be created in the attached table; or, if an equivalent
+ index already exists, will be attached to the target table's index,
+ as if <command>ALTER INDEX ATTACH</command> had been executed.
</para>
<para>
@@ -844,7 +847,8 @@ ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
<para>
This form detaches specified partition of the target table. The detached
partition continues to exist as a standalone table, but no longer has any
- ties to the table from which it was detached.
+ ties to the table from which it was detached. Any indexes that were
+ attached to the target table's indexes are detached.
</para>
</listitem>
</varlistentry>
diff --git a/doc/src/sgml/ref/create_index.sgml b/doc/src/sgml/ref/create_index.sgml
index 025537575b..5137fe6383 100644
--- a/doc/src/sgml/ref/create_index.sgml
+++ b/doc/src/sgml/ref/create_index.sgml
@@ -21,7 +21,7 @@ PostgreSQL documentation
<refsynopsisdiv>
<synopsis>
-CREATE [ UNIQUE ] INDEX [ CONCURRENTLY ] [ [ IF NOT EXISTS ] <replaceable class="parameter">name</replaceable> ] ON <replaceable class="parameter">table_name</replaceable> [ USING <replaceable class="parameter">method</replaceable> ]
+CREATE [ UNIQUE ] INDEX [ CONCURRENTLY ] [ [ IF NOT EXISTS ] <replaceable class="parameter">name</replaceable> ] ON [ ONLY ] <replaceable class="parameter">table_name</replaceable> [ USING <replaceable class="parameter">method</replaceable> ]
( { <replaceable class="parameter">column_name</replaceable> | ( <replaceable class="parameter">expression</replaceable> ) } [ COLLATE <replaceable class="parameter">collation</replaceable> ] [ <replaceable class="parameter">opclass</replaceable> ] [ ASC | DESC ] [ NULLS { FIRST | LAST } ] [, ...] )
[ WITH ( <replaceable class="parameter">storage_parameter</replaceable> = <replaceable class="parameter">value</replaceable> [, ... ] ) ]
[ TABLESPACE <replaceable class="parameter">tablespace_name</replaceable> ]
@@ -152,6 +152,16 @@ CREATE [ UNIQUE ] INDEX [ CONCURRENTLY ] [ [ IF NOT EXISTS ] <replaceable class=
</varlistentry>
<varlistentry>
+ <term><literal>ONLY</literal></term>
+ <listitem>
+ <para>
+ Indicates not to recurse creating indexes on partitions, if the
+ table is partitioned. The default is to recurse.
+ </para>
+ </listitem>
+ </varlistentry>
+
+ <varlistentry>
<term><replaceable class="parameter">table_name</replaceable></term>
<listitem>
<para>
@@ -546,6 +556,27 @@ Indexes:
</para>
<para>
+ When <literal>CREATE INDEX</literal> is invoked on a partitioned
+ table, the default behavior is to recurse to all partitions to ensure
+ they all have matching indexes.
+ Each partition is first checked to determine whether an equivalent
+ index already exists, and if so, that index will become attached as a
+ partition index to the index being created, which will become its
+ parent index.
+ If no matching index exists, a new index will be created and
+ automatically attached; the name of the new index in each partition
+ will be determined as if no index name had been specified in the
+ command.
+ If the <literal>ONLY</literal> option is specified, no recursion
+ is done, and the index is marked invalid
+ (<command>ALTER INDEX ... ATTACH PARTITION</command> turns the index
+ valid, once all partitions acquire the index.) Note, however, that
+ any partition that is created in the future using
+ <command>CREATE TABLE ... PARTITION OF</command> will automatically
+ contain the index regardless of whether this option was specified.
+ </para>
+
+ <para>
For index methods that support ordered scans (currently, only B-tree),
the optional clauses <literal>ASC</literal>, <literal>DESC</literal>, <literal>NULLS
FIRST</literal>, and/or <literal>NULLS LAST</literal> can be specified to modify
diff --git a/doc/src/sgml/ref/reindex.sgml b/doc/src/sgml/ref/reindex.sgml
index 79f6931c6a..1c21fafb80 100644
--- a/doc/src/sgml/ref/reindex.sgml
+++ b/doc/src/sgml/ref/reindex.sgml
@@ -231,6 +231,11 @@ REINDEX [ ( VERBOSE ) ] { INDEX | TABLE | SCHEMA | DATABASE | SYSTEM } <replacea
reindex anything.
</para>
+ <para>
+ Reindexing partitioned tables or partitioned indexes is not supported.
+ Each individual partition can be reindexed separately instead.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/access/common/reloptions.c b/src/backend/access/common/reloptions.c
index aa9c0f1bb9..039c91ab5a 100644
--- a/src/backend/access/common/reloptions.c
+++ b/src/backend/access/common/reloptions.c
@@ -993,6 +993,7 @@ extractRelOptions(HeapTuple tuple, TupleDesc tupdesc,
options = view_reloptions(datum, false);
break;
case RELKIND_INDEX:
+ case RELKIND_PARTITIONED_INDEX:
options = index_reloptions(amoptions, datum, false);
break;
case RELKIND_FOREIGN_TABLE:
diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c
index 54f1100ffd..b9fd7875fc 100644
--- a/src/backend/access/heap/heapam.c
+++ b/src/backend/access/heap/heapam.c
@@ -1293,7 +1293,8 @@ heap_open(Oid relationId, LOCKMODE lockmode)
r = relation_open(relationId, lockmode);
- if (r->rd_rel->relkind == RELKIND_INDEX)
+ if (r->rd_rel->relkind == RELKIND_INDEX ||
+ r->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is an index",
@@ -1321,7 +1322,8 @@ heap_openrv(const RangeVar *relation, LOCKMODE lockmode)
r = relation_openrv(relation, lockmode);
- if (r->rd_rel->relkind == RELKIND_INDEX)
+ if (r->rd_rel->relkind == RELKIND_INDEX ||
+ r->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is an index",
@@ -1353,7 +1355,8 @@ heap_openrv_extended(const RangeVar *relation, LOCKMODE lockmode,
if (r)
{
- if (r->rd_rel->relkind == RELKIND_INDEX)
+ if (r->rd_rel->relkind == RELKIND_INDEX ||
+ r->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is an index",
diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c
index edf4172eb2..c60db1eef5 100644
--- a/src/backend/access/index/indexam.c
+++ b/src/backend/access/index/indexam.c
@@ -154,7 +154,8 @@ index_open(Oid relationId, LOCKMODE lockmode)
r = relation_open(relationId, lockmode);
- if (r->rd_rel->relkind != RELKIND_INDEX)
+ if (r->rd_rel->relkind != RELKIND_INDEX &&
+ r->rd_rel->relkind != RELKIND_PARTITIONED_INDEX)
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is not an index",
diff --git a/src/backend/bootstrap/bootparse.y b/src/backend/bootstrap/bootparse.y
index 2e1fef0350..95835ac1e7 100644
--- a/src/backend/bootstrap/bootparse.y
+++ b/src/backend/bootstrap/bootparse.y
@@ -321,6 +321,7 @@ Boot_DeclareIndexStmt:
DefineIndex(relationId,
stmt,
$4,
+ InvalidOid,
false,
false,
false,
@@ -365,6 +366,7 @@ Boot_DeclareUniqueIndexStmt:
DefineIndex(relationId,
stmt,
$5,
+ InvalidOid,
false,
false,
false,
diff --git a/src/backend/catalog/aclchk.c b/src/backend/catalog/aclchk.c
index e481cf3d11..b09dba5dbf 100644
--- a/src/backend/catalog/aclchk.c
+++ b/src/backend/catalog/aclchk.c
@@ -1824,7 +1824,8 @@ ExecGrant_Relation(InternalGrant *istmt)
pg_class_tuple = (Form_pg_class) GETSTRUCT(tuple);
/* Not sensible to grant on an index */
- if (pg_class_tuple->relkind == RELKIND_INDEX)
+ if (pg_class_tuple->relkind == RELKIND_INDEX ||
+ pg_class_tuple->relkind == RELKIND_PARTITIONED_INDEX)
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is an index",
@@ -5405,7 +5406,8 @@ recordExtObjInitPriv(Oid objoid, Oid classoid)
pg_class_tuple = (Form_pg_class) GETSTRUCT(tuple);
/* Indexes don't have permissions */
- if (pg_class_tuple->relkind == RELKIND_INDEX)
+ if (pg_class_tuple->relkind == RELKIND_INDEX ||
+ pg_class_tuple->relkind == RELKIND_PARTITIONED_INDEX)
return;
/* Composite types don't have permissions either */
@@ -5690,7 +5692,8 @@ removeExtObjInitPriv(Oid objoid, Oid classoid)
pg_class_tuple = (Form_pg_class) GETSTRUCT(tuple);
/* Indexes don't have permissions */
- if (pg_class_tuple->relkind == RELKIND_INDEX)
+ if (pg_class_tuple->relkind == RELKIND_INDEX ||
+ pg_class_tuple->relkind == RELKIND_PARTITIONED_INDEX)
return;
/* Composite types don't have permissions either */
diff --git a/src/backend/catalog/dependency.c b/src/backend/catalog/dependency.c
index 033c4358ea..e412f7649b 100644
--- a/src/backend/catalog/dependency.c
+++ b/src/backend/catalog/dependency.c
@@ -582,6 +582,7 @@ findDependentObjects(const ObjectAddress *object,
/* FALL THRU */
case DEPENDENCY_INTERNAL:
+ case DEPENDENCY_INTERNAL_AUTO:
/*
* This object is part of the internal implementation of
@@ -628,6 +629,9 @@ findDependentObjects(const ObjectAddress *object,
if (stack_address_present_add_flags(&otherObject, 0, stack))
break;
+ if (foundDep->deptype == DEPENDENCY_INTERNAL_AUTO)
+ break;
+
/*
* 3. Not all the owning objects have been visited, so
* transform this deletion request into a delete of this
@@ -675,6 +679,7 @@ findDependentObjects(const ObjectAddress *object,
/* And we're done here. */
systable_endscan(scan);
return;
+
case DEPENDENCY_PIN:
/*
@@ -762,6 +767,7 @@ findDependentObjects(const ObjectAddress *object,
case DEPENDENCY_AUTO_EXTENSION:
subflags = DEPFLAG_AUTO;
break;
+ case DEPENDENCY_INTERNAL_AUTO:
case DEPENDENCY_INTERNAL:
subflags = DEPFLAG_INTERNAL;
break;
@@ -1109,7 +1115,8 @@ doDeletion(const ObjectAddress *object, int flags)
{
char relKind = get_rel_relkind(object->objectId);
- if (relKind == RELKIND_INDEX)
+ if (relKind == RELKIND_INDEX ||
+ relKind == RELKIND_PARTITIONED_INDEX)
{
bool concurrent = ((flags & PERFORM_DELETION_CONCURRENTLY) != 0);
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 4319fc6b8c..5f5961b246 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -294,6 +294,7 @@ heap_create(const char *relname,
case RELKIND_COMPOSITE_TYPE:
case RELKIND_FOREIGN_TABLE:
case RELKIND_PARTITIONED_TABLE:
+ case RELKIND_PARTITIONED_INDEX:
create_storage = false;
/*
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 0125c18bc1..a4787c3ad6 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -41,6 +41,7 @@
#include "catalog/pg_collation.h"
#include "catalog/pg_constraint.h"
#include "catalog/pg_constraint_fn.h"
+#include "catalog/pg_depend.h"
#include "catalog/pg_operator.h"
#include "catalog/pg_opclass.h"
#include "catalog/pg_tablespace.h"
@@ -98,6 +99,7 @@ static void InitializeAttributeOids(Relation indexRelation,
int numatts, Oid indexoid);
static void AppendAttributeTuples(Relation indexRelation, int numatts);
static void UpdateIndexRelation(Oid indexoid, Oid heapoid,
+ Oid parentIndexId,
IndexInfo *indexInfo,
Oid *collationOids,
Oid *classOids,
@@ -105,7 +107,8 @@ static void UpdateIndexRelation(Oid indexoid, Oid heapoid,
bool primary,
bool isexclusion,
bool immediate,
- bool isvalid);
+ bool isvalid,
+ bool isready);
static void index_update_stats(Relation rel,
bool hasindex, bool isprimary,
double reltuples);
@@ -551,6 +554,7 @@ AppendAttributeTuples(Relation indexRelation, int numatts)
static void
UpdateIndexRelation(Oid indexoid,
Oid heapoid,
+ Oid parentIndexOid,
IndexInfo *indexInfo,
Oid *collationOids,
Oid *classOids,
@@ -558,7 +562,8 @@ UpdateIndexRelation(Oid indexoid,
bool primary,
bool isexclusion,
bool immediate,
- bool isvalid)
+ bool isvalid,
+ bool isready)
{
int2vector *indkey;
oidvector *indcollation;
@@ -624,6 +629,7 @@ UpdateIndexRelation(Oid indexoid,
values[Anum_pg_index_indexrelid - 1] = ObjectIdGetDatum(indexoid);
values[Anum_pg_index_indrelid - 1] = ObjectIdGetDatum(heapoid);
+ values[Anum_pg_index_indparentidx - 1] = ObjectIdGetDatum(parentIndexOid);
values[Anum_pg_index_indnatts - 1] = Int16GetDatum(indexInfo->ii_NumIndexAttrs);
values[Anum_pg_index_indisunique - 1] = BoolGetDatum(indexInfo->ii_Unique);
values[Anum_pg_index_indisprimary - 1] = BoolGetDatum(primary);
@@ -632,8 +638,7 @@ UpdateIndexRelation(Oid indexoid,
values[Anum_pg_index_indisclustered - 1] = BoolGetDatum(false);
values[Anum_pg_index_indisvalid - 1] = BoolGetDatum(isvalid);
values[Anum_pg_index_indcheckxmin - 1] = BoolGetDatum(false);
- /* we set isvalid and isready the same way */
- values[Anum_pg_index_indisready - 1] = BoolGetDatum(isvalid);
+ values[Anum_pg_index_indisready - 1] = BoolGetDatum(isready);
values[Anum_pg_index_indislive - 1] = BoolGetDatum(true);
values[Anum_pg_index_indisreplident - 1] = BoolGetDatum(false);
values[Anum_pg_index_indkey - 1] = PointerGetDatum(indkey);
@@ -670,6 +675,8 @@ UpdateIndexRelation(Oid indexoid,
* indexRelationId: normally, pass InvalidOid to let this routine
* generate an OID for the index. During bootstrap this may be
* nonzero to specify a preselected OID.
+ * parentIndexRelid: if creating an index partition, the OID of the
+ * parent index; otherwise InvalidOid.
* relFileNode: normally, pass InvalidOid to get new storage. May be
* nonzero to attach an existing valid build.
* indexInfo: same info executor uses to insert into the index
@@ -695,6 +702,8 @@ UpdateIndexRelation(Oid indexoid,
* INDEX_CREATE_IF_NOT_EXISTS:
* do not throw an error if a relation with the same name
* already exists.
+ * INDEX_CREATE_PARTITIONED:
+ * create a partitioned index (table must be partitioned)
* constr_flags: flags passed to index_constraint_create
* (only if INDEX_CREATE_ADD_CONSTRAINT is set)
* allow_system_table_mods: allow table to be a system catalog
@@ -706,6 +715,7 @@ Oid
index_create(Relation heapRelation,
const char *indexRelationName,
Oid indexRelationId,
+ Oid parentIndexRelid,
Oid relFileNode,
IndexInfo *indexInfo,
List *indexColNames,
@@ -731,12 +741,18 @@ index_create(Relation heapRelation,
int i;
char relpersistence;
bool isprimary = (flags & INDEX_CREATE_IS_PRIMARY) != 0;
+ bool invalid = (flags & INDEX_CREATE_INVALID) != 0;
bool concurrent = (flags & INDEX_CREATE_CONCURRENT) != 0;
+ bool partitioned = (flags & INDEX_CREATE_PARTITIONED) != 0;
+ char relkind;
/* constraint flags can only be set when a constraint is requested */
Assert((constr_flags == 0) ||
((flags & INDEX_CREATE_ADD_CONSTRAINT) != 0));
+ /* partitioned indexes must never be "built" by themselves */
+ Assert(!partitioned || (flags & INDEX_CREATE_SKIP_BUILD));
+ relkind = partitioned ? RELKIND_PARTITIONED_INDEX : RELKIND_INDEX;
is_exclusion = (indexInfo->ii_ExclusionOps != NULL);
pg_class = heap_open(RelationRelationId, RowExclusiveLock);
@@ -864,7 +880,7 @@ index_create(Relation heapRelation,
indexRelationId,
relFileNode,
indexTupDesc,
- RELKIND_INDEX,
+ relkind,
relpersistence,
shared_relation,
mapped_relation,
@@ -921,10 +937,12 @@ index_create(Relation heapRelation,
* (Or, could define a rule to maintain the predicate) --Nels, Feb '92
* ----------------
*/
- UpdateIndexRelation(indexRelationId, heapRelationId, indexInfo,
+ UpdateIndexRelation(indexRelationId, heapRelationId, parentIndexRelid,
+ indexInfo,
collationObjectId, classObjectId, coloptions,
isprimary, is_exclusion,
(constr_flags & INDEX_CONSTR_CREATE_DEFERRABLE) == 0,
+ !concurrent && !invalid,
!concurrent);
/*
@@ -978,6 +996,9 @@ index_create(Relation heapRelation,
else
{
bool have_simple_col = false;
+ DependencyType deptype;
+
+ deptype = OidIsValid(parentIndexRelid) ? DEPENDENCY_INTERNAL_AUTO : DEPENDENCY_AUTO;
/* Create auto dependencies on simply-referenced columns */
for (i = 0; i < indexInfo->ii_NumIndexAttrs; i++)
@@ -988,7 +1009,7 @@ index_create(Relation heapRelation,
referenced.objectId = heapRelationId;
referenced.objectSubId = indexInfo->ii_KeyAttrNumbers[i];
- recordDependencyOn(&myself, &referenced, DEPENDENCY_AUTO);
+ recordDependencyOn(&myself, &referenced, deptype);
have_simple_col = true;
}
@@ -1006,10 +1027,21 @@ index_create(Relation heapRelation,
referenced.objectId = heapRelationId;
referenced.objectSubId = 0;
- recordDependencyOn(&myself, &referenced, DEPENDENCY_AUTO);
+ recordDependencyOn(&myself, &referenced, deptype);
}
}
+ /* Store dependency on parent index, if any */
+ if (OidIsValid(parentIndexRelid))
+ {
+ referenced.classId = RelationRelationId;
+ referenced.objectId = parentIndexRelid;
+ referenced.objectSubId = 0;
+
+ recordDependencyOn(&myself, &referenced, DEPENDENCY_INTERNAL_AUTO);
+ }
+
+
/* Store dependency on collations */
/* The default collation is pinned, so don't bother recording it */
for (i = 0; i < indexInfo->ii_NumIndexAttrs; i++)
@@ -1555,9 +1587,10 @@ index_drop(Oid indexId, bool concurrent)
}
/*
- * Schedule physical removal of the files
+ * Schedule physical removal of the files (if any)
*/
- RelationDropStorage(userIndexRelation);
+ if (userIndexRelation->rd_rel->relkind != RELKIND_PARTITIONED_INDEX)
+ RelationDropStorage(userIndexRelation);
/*
* Close and flush the index's relcache entry, to ensure relcache doesn't
@@ -1694,12 +1727,65 @@ BuildIndexInfo(Relation index)
ii->ii_BrokenHotChain = false;
/* set up for possible use by index AM */
+ ii->ii_Am = index->rd_rel->relam;
ii->ii_AmCache = NULL;
ii->ii_Context = CurrentMemoryContext;
return ii;
}
+/*
+ * CompareIndexInfo
+ * Compare two IndexInfos, and return true if they are similar enough that
+ * an index built with one can pass as an index built with the other.
+ *
+ * attmap is an attribute map where info2 is input and info1 is output.
+ */
+bool
+CompareIndexInfo(IndexInfo *info1, IndexInfo *info2, AttrNumber *attmap)
+{
+ int i;
+
+ /* indexes are only equivalent if they have the same access method */
+ if (info1->ii_Am != info2->ii_Am)
+ return false;
+
+ /* and same number of attributes */
+ if (info1->ii_NumIndexAttrs != info2->ii_NumIndexAttrs)
+ return false;
+
+ /*
+ * and columns match through the attribute map (actual attribute numbers
+ * might differ!)
+ */
+ for (i = 0; i < info1->ii_NumIndexAttrs; i++)
+ {
+ if (attmap[info1->ii_KeyAttrNumbers[i] - 1] !=
+ info2->ii_KeyAttrNumbers[i])
+ return false;
+ }
+
+ /*
+ * Expression indexes are currently not considered equal. Not needed for
+ * current callers.
+ */
+ if (info1->ii_Expressions != NIL || info2->ii_Expressions != NIL)
+ return false;
+
+ /* Index predicates must be identical */
+ if (!equal(info1->ii_Predicate, info2->ii_Predicate))
+ return false;
+
+ /* No support currently for comparing exclusion indexes. */
+ if (info1->ii_ExclusionOps != NULL || info2->ii_ExclusionOps != NULL)
+ return false;
+
+ if (info1->ii_Unique != info2->ii_Unique)
+ return false;
+
+ return true;
+}
+
/* ----------------
* BuildSpeculativeIndexInfo
* Add extra state to IndexInfo record
@@ -1922,6 +2008,9 @@ index_update_stats(Relation rel,
elog(ERROR, "could not find tuple for relation %u", relid);
rd_rel = (Form_pg_class) GETSTRUCT(tuple);
+ /* Should this be a more comprehensive test? */
+ Assert(rd_rel->relkind != RELKIND_PARTITIONED_INDEX);
+
/* Apply required updates, if any, to copied tuple */
dirty = false;
@@ -3332,6 +3421,14 @@ reindex_index(Oid indexId, bool skip_constraint_checks, char persistence,
iRel = index_open(indexId, AccessExclusiveLock);
/*
+ * The case of reindexing partitioned tables and indexes is handled
+ * differently by upper layers, so this case shouldn't arise.
+ */
+ if (iRel->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
+ elog(ERROR, "unsupported relation kind for index \"%s\"",
+ RelationGetRelationName(iRel));
+
+ /*
* Don't allow reindex on temp tables of other backends ... their local
* buffer manager is not going to cope.
*/
@@ -3530,6 +3627,22 @@ reindex_relation(Oid relid, int flags, int options)
*/
rel = heap_open(relid, ShareLock);
+ /*
+ * This may be useful when implemented someday; but that day is not today.
+ * For now, avoid erroring out when called in a multi-table context
+ * (REINDEX SCHEMA) and happen to come across a partitioned table. The
+ * partitions may be reindexed on their own anyway.
+ */
+ if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
+ {
+ ereport(WARNING,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("REINDEX of partitioned tables is not yet implemented, skipping \"%s\"",
+ RelationGetRelationName(rel))));
+ heap_close(rel, ShareLock);
+ return false;
+ }
+
toast_relid = rel->rd_rel->reltoastrelid;
/*
diff --git a/src/backend/catalog/objectaddress.c b/src/backend/catalog/objectaddress.c
index 9553675975..90714e4838 100644
--- a/src/backend/catalog/objectaddress.c
+++ b/src/backend/catalog/objectaddress.c
@@ -1217,7 +1217,8 @@ get_relation_by_qualified_name(ObjectType objtype, List *object,
switch (objtype)
{
case OBJECT_INDEX:
- if (relation->rd_rel->relkind != RELKIND_INDEX)
+ if (relation->rd_rel->relkind != RELKIND_INDEX &&
+ relation->rd_rel->relkind != RELKIND_PARTITIONED_INDEX)
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is not an index",
@@ -3483,6 +3484,7 @@ getRelationDescription(StringInfo buffer, Oid relid)
relname);
break;
case RELKIND_INDEX:
+ case RELKIND_PARTITIONED_INDEX:
appendStringInfo(buffer, _("index %s"),
relname);
break;
@@ -3957,6 +3959,7 @@ getRelationTypeDescription(StringInfo buffer, Oid relid, int32 objectSubId)
appendStringInfoString(buffer, "table");
break;
case RELKIND_INDEX:
+ case RELKIND_PARTITIONED_INDEX:
appendStringInfoString(buffer, "index");
break;
case RELKIND_SEQUENCE:
diff --git a/src/backend/catalog/pg_depend.c b/src/backend/catalog/pg_depend.c
index cf0086b9bd..53db972dec 100644
--- a/src/backend/catalog/pg_depend.c
+++ b/src/backend/catalog/pg_depend.c
@@ -656,14 +656,19 @@ get_constraint_index(Oid constraintId)
/*
* We assume any internal dependency of an index on the constraint
- * must be what we are looking for. (The relkind test is just
- * paranoia; there shouldn't be any such dependencies otherwise.)
+ * must be what we are looking for.
*/
if (deprec->classid == RelationRelationId &&
deprec->objsubid == 0 &&
- deprec->deptype == DEPENDENCY_INTERNAL &&
- get_rel_relkind(deprec->objid) == RELKIND_INDEX)
+ deprec->deptype == DEPENDENCY_INTERNAL)
{
+ char relkind = get_rel_relkind(deprec->objid);
+
+ /* This is pure paranoia; there shouldn't be any such */
+ if (relkind != RELKIND_INDEX &&
+ relkind != RELKIND_PARTITIONED_INDEX)
+ break;
+
indexId = deprec->objid;
break;
}
diff --git a/src/backend/catalog/toasting.c b/src/backend/catalog/toasting.c
index 539ca79ad3..09582a8d52 100644
--- a/src/backend/catalog/toasting.c
+++ b/src/backend/catalog/toasting.c
@@ -315,6 +315,7 @@ create_toast_table(Relation rel, Oid toastOid, Oid toastIndexOid,
indexInfo->ii_ReadyForInserts = true;
indexInfo->ii_Concurrent = false;
indexInfo->ii_BrokenHotChain = false;
+ indexInfo->ii_Am = BTREE_AM_OID;
indexInfo->ii_AmCache = NULL;
indexInfo->ii_Context = CurrentMemoryContext;
@@ -328,6 +329,7 @@ create_toast_table(Relation rel, Oid toastOid, Oid toastIndexOid,
coloptions[1] = 0;
index_create(toast_rel, toast_idxname, toastIndexOid, InvalidOid,
+ InvalidOid,
indexInfo,
list_make2("chunk_id", "chunk_seq"),
BTREE_AM_OID,
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 97091dd9fb..e925351056 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -23,7 +23,9 @@
#include "catalog/catalog.h"
#include "catalog/index.h"
#include "catalog/indexing.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits_fn.h"
#include "catalog/pg_opclass.h"
#include "catalog/pg_opfamily.h"
#include "catalog/pg_tablespace.h"
@@ -35,6 +37,7 @@
#include "commands/tablespace.h"
#include "mb/pg_wchar.h"
#include "miscadmin.h"
+#include "nodes/makefuncs.h"
#include "nodes/nodeFuncs.h"
#include "optimizer/clauses.h"
#include "optimizer/planner.h"
@@ -77,6 +80,7 @@ static char *ChooseIndexNameAddition(List *colnames);
static List *ChooseIndexColumnNames(List *indexElems);
static void RangeVarCallbackForReindexIndex(const RangeVar *relation,
Oid relId, Oid oldRelId, void *arg);
+static void ReindexPartitionedIndex(Relation parentIdx);
/*
* CheckIndexCompatible
@@ -183,6 +187,7 @@ CheckIndexCompatible(Oid oldId,
indexInfo->ii_ExclusionOps = NULL;
indexInfo->ii_ExclusionProcs = NULL;
indexInfo->ii_ExclusionStrats = NULL;
+ indexInfo->ii_Am = accessMethodId;
indexInfo->ii_AmCache = NULL;
indexInfo->ii_Context = CurrentMemoryContext;
typeObjectId = (Oid *) palloc(numberOfAttributes * sizeof(Oid));
@@ -292,14 +297,15 @@ CheckIndexCompatible(Oid oldId,
* 'stmt': IndexStmt describing the properties of the new index.
* 'indexRelationId': normally InvalidOid, but during bootstrap can be
* nonzero to specify a preselected OID for the index.
+ * 'parentIndexId': the OID of the parent index; InvalidOid if not the child
+ * of a partitioned index.
* 'is_alter_table': this is due to an ALTER rather than a CREATE operation.
* 'check_rights': check for CREATE rights in namespace and tablespace. (This
* should be true except when ALTER is deleting/recreating an index.)
* 'check_not_in_use': check for table not already in use in current session.
* This should be true unless caller is holding the table open, in which
* case the caller had better have checked it earlier.
- * 'skip_build': make the catalog entries but leave the index file empty;
- * it will be filled later.
+ * 'skip_build': make the catalog entries but don't create the index files
* 'quiet': suppress the NOTICE chatter ordinarily provided for constraints.
*
* Returns the object address of the created index.
@@ -308,6 +314,7 @@ ObjectAddress
DefineIndex(Oid relationId,
IndexStmt *stmt,
Oid indexRelationId,
+ Oid parentIndexId,
bool is_alter_table,
bool check_rights,
bool check_not_in_use,
@@ -330,6 +337,7 @@ DefineIndex(Oid relationId,
IndexAmRoutine *amRoutine;
bool amcanorder;
amoptions_function amoptions;
+ bool partitioned;
Datum reloptions;
int16 *coloptions;
IndexInfo *indexInfo;
@@ -382,23 +390,56 @@ DefineIndex(Oid relationId,
{
case RELKIND_RELATION:
case RELKIND_MATVIEW:
+ case RELKIND_PARTITIONED_TABLE:
/* OK */
break;
case RELKIND_FOREIGN_TABLE:
+ /*
+ * Custom error message for FOREIGN TABLE since the term is close
+ * to a regular table and can confuse the user.
+ */
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("cannot create index on foreign table \"%s\"",
RelationGetRelationName(rel))));
- case RELKIND_PARTITIONED_TABLE:
- ereport(ERROR,
- (errcode(ERRCODE_WRONG_OBJECT_TYPE),
- errmsg("cannot create index on partitioned table \"%s\"",
- RelationGetRelationName(rel))));
default:
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is not a table or materialized view",
RelationGetRelationName(rel))));
+ break;
+ }
+
+ /*
+ * Establish behavior for partitioned tables, and verify sanity of
+ * parameters.
+ *
+ * We do not build an actual index in this case; we only create a few
+ * catalog entries. The actual indexes are built by recursing for each
+ * partition.
+ */
+ partitioned = rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE;
+ if (partitioned)
+ {
+ if (stmt->concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("cannot create index on partitioned table \"%s\" concurrently",
+ RelationGetRelationName(rel))));
+ if (stmt->unique)
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("cannot create unique index on partitioned table \"%s\"",
+ RelationGetRelationName(rel))));
+ if (stmt->excludeOpNames)
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("cannot create exclusion constraints on partitioned table \"%s\"",
+ RelationGetRelationName(rel))));
+ if (stmt->primary || stmt->isconstraint)
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("cannot create constraints on partitioned tables")));
}
/*
@@ -574,6 +615,7 @@ DefineIndex(Oid relationId,
indexInfo->ii_ReadyForInserts = !stmt->concurrent;
indexInfo->ii_Concurrent = stmt->concurrent;
indexInfo->ii_BrokenHotChain = false;
+ indexInfo->ii_Am = accessMethodId;
indexInfo->ii_AmCache = NULL;
indexInfo->ii_Context = CurrentMemoryContext;
@@ -665,19 +707,24 @@ DefineIndex(Oid relationId,
/*
* Make the catalog entries for the index, including constraints. This
* step also actually builds the index, except if caller requested not to
- * or in concurrent mode, in which case it'll be done later.
+ * or in concurrent mode, in which case it'll be done later, or
+ * doing a partitioned index (because those don't have storage).
*/
flags = constr_flags = 0;
if (stmt->isconstraint)
flags |= INDEX_CREATE_ADD_CONSTRAINT;
- if (skip_build || stmt->concurrent)
+ if (skip_build || stmt->concurrent || partitioned)
flags |= INDEX_CREATE_SKIP_BUILD;
if (stmt->if_not_exists)
flags |= INDEX_CREATE_IF_NOT_EXISTS;
if (stmt->concurrent)
flags |= INDEX_CREATE_CONCURRENT;
+ if (partitioned)
+ flags |= INDEX_CREATE_PARTITIONED;
if (stmt->primary)
flags |= INDEX_CREATE_IS_PRIMARY;
+ if (partitioned && stmt->relation && !stmt->relation->inh)
+ flags |= INDEX_CREATE_INVALID;
if (stmt->deferrable)
constr_flags |= INDEX_CONSTR_CREATE_DEFERRABLE;
@@ -685,8 +732,8 @@ DefineIndex(Oid relationId,
constr_flags |= INDEX_CONSTR_CREATE_INIT_DEFERRED;
indexRelationId =
- index_create(rel, indexRelationName, indexRelationId, stmt->oldNode,
- indexInfo, indexColNames,
+ index_create(rel, indexRelationName, indexRelationId, parentIndexId,
+ stmt->oldNode, indexInfo, indexColNames,
accessMethodId, tablespaceId,
collationObjectId, classObjectId,
coloptions, reloptions,
@@ -706,6 +753,118 @@ DefineIndex(Oid relationId,
CreateComments(indexRelationId, RelationRelationId, 0,
stmt->idxcomment);
+ if (partitioned)
+ {
+ /*
+ * Unless caller specified to skip this step (via ONLY), process
+ * each partition to make sure they all contain a corresponding index.
+ *
+ * If we're called internally (no stmt->relation), recurse always.
+ */
+ if (!stmt->relation || stmt->relation->inh)
+ {
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ int nparts = partdesc->nparts;
+ Oid *part_oids;
+ TupleDesc parentDesc;
+
+ nparts = partdesc->nparts;
+ part_oids = palloc(sizeof(Oid) * nparts);
+ memcpy(part_oids, partdesc->oids, sizeof(Oid) * nparts);
+
+ parentDesc = CreateTupleDescCopy(RelationGetDescr(rel));
+
+ heap_close(rel, NoLock);
+
+ /*
+ * For each partition, scan all existing indexes; if one matches
+ * our index definition and is not already attached to some other
+ * parent index, attach it to the one we just created.
+ *
+ * If none matches, build a new index by calling ourselves
+ * recursively with the same options (except for the index name).
+ */
+ for (i = 0; i < nparts; i++)
+ {
+ Oid childRelid = part_oids[i];
+ Relation childrel;
+ List *childidxs;
+ ListCell *cell;
+ AttrNumber *attmap = NULL;
+ bool found = false;
+
+ childrel = heap_open(childRelid, lockmode);
+ childidxs = RelationGetIndexList(childrel);
+
+ foreach(cell, childidxs)
+ {
+ Oid cldidxid = lfirst_oid(cell);
+ Relation cldidx;
+ IndexInfo *cldIdxInfo;
+
+ cldidx = index_open(cldidxid, lockmode);
+
+ /* this index is already partition of another one */
+ if (cldidx->rd_index->indparentidx != 0)
+ {
+ index_close(cldidx, lockmode);
+ continue;
+ }
+
+ cldIdxInfo = BuildIndexInfo(cldidx);
+ if (attmap == NULL)
+ attmap =
+ convert_tuples_by_name_map(RelationGetDescr(childrel),
+ parentDesc,
+ gettext_noop("could not convert row type"));
+
+ if (CompareIndexInfo(cldIdxInfo, indexInfo, attmap))
+ {
+ /*
+ * Found a match. Attach index to parent and we're
+ * done, but keep lock till commit.
+ */
+ IndexSetParentIndex(cldidx, indexRelationId);
+ found = true;
+ index_close(cldidx, NoLock);
+ break;
+ }
+
+ index_close(cldidx, lockmode);
+ }
+
+ list_free(childidxs);
+ heap_close(childrel, NoLock);
+ if (attmap)
+ pfree(attmap);
+
+ /*
+ * If no matching index was found, create our own.
+ */
+ if (!found)
+ {
+ IndexStmt *childStmt = copyObject(stmt);
+
+ childStmt->idxname = NULL;
+ childStmt->relationId = childRelid;
+ DefineIndex(childRelid, childStmt,
+ InvalidOid, /* no predefined OID */
+ indexRelationId, /* this is our child */
+ false, check_rights, check_not_in_use,
+ false, quiet);
+ }
+ }
+ }
+ else
+ heap_close(rel, NoLock);
+
+ /*
+ * Indexes on partitioned tables are not themselves built, so we're
+ * done here.
+ */
+ return address;
+ }
+
if (!stmt->concurrent)
{
/* Close the heap and we're done, in the non-concurrent case */
@@ -1762,7 +1921,7 @@ ChooseIndexColumnNames(List *indexElems)
* ReindexIndex
* Recreate a specific index.
*/
-Oid
+void
ReindexIndex(RangeVar *indexRelation, int options)
{
Oid indOid;
@@ -1785,12 +1944,17 @@ ReindexIndex(RangeVar *indexRelation, int options)
* lock on the index.
*/
irel = index_open(indOid, NoLock);
+
+ if (irel->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
+ {
+ ReindexPartitionedIndex(irel);
+ return;
+ }
+
persistence = irel->rd_rel->relpersistence;
index_close(irel, NoLock);
reindex_index(indOid, false, persistence, options);
-
- return indOid;
}
/*
@@ -1829,7 +1993,8 @@ RangeVarCallbackForReindexIndex(const RangeVar *relation,
relkind = get_rel_relkind(relId);
if (!relkind)
return;
- if (relkind != RELKIND_INDEX)
+ if (relkind != RELKIND_INDEX &&
+ relkind != RELKIND_PARTITIONED_INDEX)
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is not an index", relation->relname)));
@@ -1973,6 +2138,12 @@ ReindexMultipleTables(const char *objectName, ReindexObjectType objectKind,
/*
* Only regular tables and matviews can have indexes, so ignore any
* other kind of relation.
+ *
+ * It is tempting to also consider partitioned tables here, but that
+ * has the problem that if the children are in the same schema, they
+ * would be processed twice. Maybe we could have a separate list of
+ * partitioned tables, and expand that afterwards into relids,
+ * ignoring any duplicates.
*/
if (classtuple->relkind != RELKIND_RELATION &&
classtuple->relkind != RELKIND_MATVIEW)
@@ -2035,3 +2206,65 @@ ReindexMultipleTables(const char *objectName, ReindexObjectType objectKind,
MemoryContextDelete(private_context);
}
+
+/*
+ * Reindex each child of a partitioned index.
+ *
+ * The parent index is given, locked in AccessExclusive mode; this routine
+ * obtains the list of children and releases the lock on parent before
+ * applying reindex on each child.
+ */
+static void
+ReindexPartitionedIndex(Relation parentIdx)
+{
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("REINDEX is not yet implemented for partitioned indexes")));
+}
+
+/*
+ * Update the pg_index tuple corresponding to the given index on a partition
+ * to indicate that the given index OID is now its parent partitioned index.
+ *
+ * (De-)register the dependency from/in pg_depend.
+ */
+void
+IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
+{
+ Relation pgindex;
+ HeapTuple indTup;
+ Form_pg_index indForm;
+
+ /* Make sure this is an index */
+ Assert(partitionIdx->rd_rel->relkind == RELKIND_INDEX ||
+ partitionIdx->rd_rel->relkind == RELKIND_PARTITIONED_INDEX);
+
+ pgindex = heap_open(IndexRelationId, RowExclusiveLock);
+ indTup = partitionIdx->rd_indextuple;
+ indForm = (Form_pg_index) GETSTRUCT(indTup);
+ indForm->indparentidx = parentOid;
+
+ CatalogTupleUpdate(pgindex, &(indTup->t_self), indTup);
+
+ heap_close(pgindex, RowExclusiveLock);
+
+ /*
+ * If setting a parent, add a pg_depend row; if making standalone, remove
+ * all existing rows.
+ */
+ if (OidIsValid(parentOid))
+ {
+ ObjectAddress parent;
+ ObjectAddress partition;
+
+ ObjectAddressSet(parent, RelationRelationId, parentOid);
+ ObjectAddressSet(partition,
+ RelationRelationId, RelationGetRelid(partitionIdx));
+ recordDependencyOn(&partition, &parent, DEPENDENCY_INTERNAL);
+ }
+ else
+ deleteDependencyRecordsForClass(RelationRelationId,
+ RelationGetRelid(partitionIdx),
+ RelationRelationId,
+ DEPENDENCY_INTERNAL);
+}
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index d979ce266d..26cec0b480 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -266,6 +266,12 @@ static const struct dropmsgstrings dropmsgstringarray[] = {
gettext_noop("table \"%s\" does not exist, skipping"),
gettext_noop("\"%s\" is not a table"),
gettext_noop("Use DROP TABLE to remove a table.")},
+ {RELKIND_PARTITIONED_INDEX,
+ ERRCODE_UNDEFINED_OBJECT,
+ gettext_noop("index \"%s\" does not exist"),
+ gettext_noop("index \"%s\" does not exist, skipping"),
+ gettext_noop("\"%s\" is not an index"),
+ gettext_noop("Use DROP INDEX to remove an index.")},
{'\0', 0, NULL, NULL, NULL, NULL}
};
@@ -481,6 +487,11 @@ static void ValidatePartitionConstraints(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
+ RangeVar *name);
+static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
+static void refuseDupeIndexAttach(Relation parentIdx, Relation partIdx,
+ Relation partitionTbl);
/* ----------------------------------------------------------------
@@ -898,6 +909,53 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
StorePartitionKey(rel, strategy, partnatts, partattrs, partexprs,
partopclass, partcollation);
+
+ /* make it all visible */
+ CommandCounterIncrement();
+ }
+
+ /*
+ * If we're creating a partition, create now all the indexes defined in
+ * the parent. We can't do it earlier, because DefineIndex wants to know
+ * the partition key which we just stored.
+ */
+ if (stmt->partbound)
+ {
+ Oid parentId = linitial_oid(inheritOids);
+ Relation parent;
+ List *idxlist;
+ ListCell *cell;
+
+ /* Already have strong enough lock on the parent */
+ parent = heap_open(parentId, NoLock);
+ idxlist = RelationGetIndexList(parent);
+
+ /*
+ * For each index in the parent table, create one in the partition
+ */
+ foreach(cell, idxlist)
+ {
+ Relation idxRel = index_open(lfirst_oid(cell), AccessShareLock);
+ AttrNumber *attmap;
+ IndexStmt *idxstmt;
+
+ attmap = convert_tuples_by_name_map(RelationGetDescr(rel),
+ RelationGetDescr(parent),
+ gettext_noop("could not convert row type"));
+ idxstmt =
+ generateClonedIndexStmt(NULL, RelationGetRelid(rel), idxRel,
+ attmap, RelationGetDescr(rel)->natts);
+ DefineIndex(RelationGetRelid(rel),
+ idxstmt,
+ InvalidOid,
+ RelationGetRelid(idxRel),
+ false, false, false, false, false);
+
+ index_close(idxRel, AccessShareLock);
+ }
+
+ list_free(idxlist);
+ heap_close(parent, NoLock);
}
/*
@@ -1180,10 +1238,13 @@ RangeVarCallbackForDropRelation(const RangeVar *rel, Oid relOid, Oid oldRelOid,
* but RemoveRelations() can only pass one relkind for a given relation.
* It chooses RELKIND_RELATION for both regular and partitioned tables.
* That means we must be careful before giving the wrong type error when
- * the relation is RELKIND_PARTITIONED_TABLE.
+ * the relation is RELKIND_PARTITIONED_TABLE. An equivalent problem
+ * exists with indexes.
*/
if (classform->relkind == RELKIND_PARTITIONED_TABLE)
expected_relkind = RELKIND_RELATION;
+ else if (classform->relkind == RELKIND_PARTITIONED_INDEX)
+ expected_relkind = RELKIND_INDEX;
else
expected_relkind = classform->relkind;
@@ -1211,7 +1272,8 @@ RangeVarCallbackForDropRelation(const RangeVar *rel, Oid relOid, Oid oldRelOid,
* we do it the other way around. No error if we don't find a pg_index
* entry, though --- the relation may have been dropped.
*/
- if (relkind == RELKIND_INDEX && relOid != oldRelOid)
+ if ((relkind == RELKIND_INDEX || relkind == RELKIND_PARTITIONED_INDEX) &&
+ relOid != oldRelOid)
{
state->heapOid = IndexGetRelation(relOid, true);
if (OidIsValid(state->heapOid))
@@ -2541,6 +2603,7 @@ renameatt_check(Oid myrelid, Form_pg_class classform, bool recursing)
relkind != RELKIND_MATVIEW &&
relkind != RELKIND_COMPOSITE_TYPE &&
relkind != RELKIND_INDEX &&
+ relkind != RELKIND_PARTITIONED_INDEX &&
relkind != RELKIND_FOREIGN_TABLE &&
relkind != RELKIND_PARTITIONED_TABLE)
ereport(ERROR,
@@ -3020,7 +3083,8 @@ RenameRelationInternal(Oid myrelid, const char *newrelname, bool is_internal)
/*
* Also rename the associated constraint, if any.
*/
- if (targetrelation->rd_rel->relkind == RELKIND_INDEX)
+ if (targetrelation->rd_rel->relkind == RELKIND_INDEX ||
+ targetrelation->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
{
Oid constraintId = get_index_constraint(myrelid);
@@ -3074,6 +3138,7 @@ CheckTableNotInUse(Relation rel, const char *stmt)
stmt, RelationGetRelationName(rel))));
if (rel->rd_rel->relkind != RELKIND_INDEX &&
+ rel->rd_rel->relkind != RELKIND_PARTITIONED_INDEX &&
AfterTriggerPendingOnRel(RelationGetRelid(rel)))
ereport(ERROR,
(errcode(ERRCODE_OBJECT_IN_USE),
@@ -3765,6 +3830,10 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
pass = AT_PASS_MISC;
break;
case AT_AttachPartition:
+ ATSimplePermissions(rel, ATT_TABLE | ATT_INDEX);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
case AT_DetachPartition:
ATSimplePermissions(rel, ATT_TABLE);
/* No command-specific prep needed */
@@ -4113,9 +4182,14 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab, Relation rel,
ATExecGenericOptions(rel, (List *) cmd->def);
break;
case AT_AttachPartition:
- ATExecAttachPartition(wqueue, rel, (PartitionCmd *) cmd->def);
+ if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
+ ATExecAttachPartition(wqueue, rel, (PartitionCmd *) cmd->def);
+ else
+ ATExecAttachPartitionIdx(wqueue, rel,
+ ((PartitionCmd *) cmd->def)->name);
break;
case AT_DetachPartition:
+ Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
@@ -4751,6 +4825,7 @@ ATSimplePermissions(Relation rel, int allowed_targets)
actual_target = ATT_MATVIEW;
break;
case RELKIND_INDEX:
+ case RELKIND_PARTITIONED_INDEX:
actual_target = ATT_INDEX;
break;
case RELKIND_COMPOSITE_TYPE:
@@ -6195,6 +6270,7 @@ ATPrepSetStatistics(Relation rel, const char *colName, int16 colNum, Node *newVa
if (rel->rd_rel->relkind != RELKIND_RELATION &&
rel->rd_rel->relkind != RELKIND_MATVIEW &&
rel->rd_rel->relkind != RELKIND_INDEX &&
+ rel->rd_rel->relkind != RELKIND_PARTITIONED_INDEX &&
rel->rd_rel->relkind != RELKIND_FOREIGN_TABLE &&
rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
ereport(ERROR,
@@ -6206,7 +6282,9 @@ ATPrepSetStatistics(Relation rel, const char *colName, int16 colNum, Node *newVa
* We allow referencing columns by numbers only for indexes, since table
* column numbers could contain gaps if columns are later dropped.
*/
- if (rel->rd_rel->relkind != RELKIND_INDEX && !colName)
+ if (rel->rd_rel->relkind != RELKIND_INDEX &&
+ rel->rd_rel->relkind != RELKIND_PARTITIONED_INDEX &&
+ !colName)
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot refer to non-index column by number")));
@@ -6284,7 +6362,8 @@ ATExecSetStatistics(Relation rel, const char *colName, int16 colNum, Node *newVa
errmsg("cannot alter system column \"%s\"",
colName)));
- if (rel->rd_rel->relkind == RELKIND_INDEX &&
+ if ((rel->rd_rel->relkind == RELKIND_INDEX ||
+ rel->rd_rel->relkind == RELKIND_PARTITIONED_INDEX) &&
rel->rd_index->indkey.values[attnum - 1] != 0)
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
@@ -6797,6 +6876,7 @@ ATExecAddIndex(AlteredTableInfo *tab, Relation rel,
address = DefineIndex(RelationGetRelid(rel),
stmt,
InvalidOid, /* no predefined OID */
+ InvalidOid, /* no parent index */
true, /* is_alter_table */
check_rights,
false, /* check_not_in_use - we did it already */
@@ -9198,7 +9278,8 @@ ATExecAlterColumnType(AlteredTableInfo *tab, Relation rel,
{
char relKind = get_rel_relkind(foundObject.objectId);
- if (relKind == RELKIND_INDEX)
+ if (relKind == RELKIND_INDEX ||
+ relKind == RELKIND_PARTITIONED_INDEX)
{
Assert(foundObject.objectSubId == 0);
if (!list_member_oid(tab->changedIndexOids, foundObject.objectId))
@@ -10041,6 +10122,15 @@ ATExecChangeOwner(Oid relationOid, Oid newOwnerId, bool recursing, LOCKMODE lock
newOwnerId = tuple_class->relowner;
}
break;
+ case RELKIND_PARTITIONED_INDEX:
+ if (recursing)
+ break;
+ ereport(ERROR,
+ (errcode(ERRCODE_WRONG_OBJECT_TYPE),
+ errmsg("cannot change owner of index \"%s\"",
+ NameStr(tuple_class->relname)),
+ errhint("Change the ownership of the index's table, instead.")));
+ break;
case RELKIND_SEQUENCE:
if (!recursing &&
tuple_class->relowner != newOwnerId)
@@ -10162,6 +10252,7 @@ ATExecChangeOwner(Oid relationOid, Oid newOwnerId, bool recursing, LOCKMODE lock
*/
if (tuple_class->relkind != RELKIND_COMPOSITE_TYPE &&
tuple_class->relkind != RELKIND_INDEX &&
+ tuple_class->relkind != RELKIND_PARTITIONED_INDEX &&
tuple_class->relkind != RELKIND_TOASTVALUE)
changeDependencyOnOwner(RelationRelationId, relationOid,
newOwnerId);
@@ -10169,7 +10260,8 @@ ATExecChangeOwner(Oid relationOid, Oid newOwnerId, bool recursing, LOCKMODE lock
/*
* Also change the ownership of the table's row type, if it has one
*/
- if (tuple_class->relkind != RELKIND_INDEX)
+ if (tuple_class->relkind != RELKIND_INDEX &&
+ tuple_class->relkind != RELKIND_PARTITIONED_INDEX)
AlterTypeOwnerInternal(tuple_class->reltype, newOwnerId);
/*
@@ -10178,6 +10270,7 @@ ATExecChangeOwner(Oid relationOid, Oid newOwnerId, bool recursing, LOCKMODE lock
* relation, as well as its toast table (if it has one).
*/
if (tuple_class->relkind == RELKIND_RELATION ||
+ tuple_class->relkind == RELKIND_PARTITIONED_TABLE ||
tuple_class->relkind == RELKIND_MATVIEW ||
tuple_class->relkind == RELKIND_TOASTVALUE)
{
@@ -10486,6 +10579,7 @@ ATExecSetRelOptions(Relation rel, List *defList, AlterTableType operation,
(void) view_reloptions(newOptions, true);
break;
case RELKIND_INDEX:
+ case RELKIND_PARTITIONED_INDEX:
(void) index_reloptions(rel->rd_amroutine->amoptions, newOptions, true);
break;
default:
@@ -10898,7 +10992,8 @@ AlterTableMoveAll(AlterTableMoveAllStmt *stmt)
relForm->relkind != RELKIND_RELATION &&
relForm->relkind != RELKIND_PARTITIONED_TABLE) ||
(stmt->objtype == OBJECT_INDEX &&
- relForm->relkind != RELKIND_INDEX) ||
+ relForm->relkind != RELKIND_INDEX &&
+ relForm->relkind != RELKIND_PARTITIONED_INDEX) ||
(stmt->objtype == OBJECT_MATVIEW &&
relForm->relkind != RELKIND_MATVIEW))
continue;
@@ -13285,7 +13380,8 @@ RangeVarCallbackForAlterRelation(const RangeVar *rv, Oid relid, Oid oldrelid,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is not a composite type", rv->relname)));
- if (reltype == OBJECT_INDEX && relkind != RELKIND_INDEX
+ if (reltype == OBJECT_INDEX && relkind != RELKIND_INDEX &&
+ relkind != RELKIND_PARTITIONED_INDEX
&& !IsA(stmt, RenameStmt))
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
@@ -14006,6 +14102,103 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
StorePartitionBound(attachrel, rel, cmd->bound);
/*
+ * Ensure a correct set of indexes in the partition. This either creates
+ * a new index in the table being attached, or re-parents an existing one.
+ */
+ {
+ AttrNumber *attmap = NULL;
+ List *idxes;
+ List *attachRelIdxs;
+ Relation *attachrelIdxRels;
+ IndexInfo **attachInfos;
+ int i;
+ ListCell *cell;
+
+ idxes = RelationGetIndexList(rel);
+ attachRelIdxs = RelationGetIndexList(attachrel);
+ attachrelIdxRels = palloc(sizeof(Relation) * list_length(attachRelIdxs));
+ attachInfos = palloc(sizeof(IndexInfo *) * list_length(attachRelIdxs));
+
+ /* Build arrays of all existing indexes and their IndexInfos */
+ i = 0;
+ foreach(cell, attachRelIdxs)
+ {
+ Oid cldIdxId = lfirst_oid(cell);
+
+ attachrelIdxRels[i] = index_open(cldIdxId, AccessShareLock);
+ attachInfos[i] = BuildIndexInfo(attachrelIdxRels[i]);
+ i++;
+ }
+
+ /*
+ * For each index on the partitioned table, find a matching one in the
+ * partition-to-be; if one is not found, create one.
+ */
+ foreach(cell, idxes)
+ {
+ Oid idx = lfirst_oid(cell);
+ Relation idxRel = index_open(idx, AccessShareLock);
+ IndexInfo *info;
+ bool found = false;
+
+ if (idxRel->rd_rel->relkind != RELKIND_PARTITIONED_INDEX)
+ {
+ index_close(idxRel, AccessShareLock);
+ continue;
+ }
+ info = BuildIndexInfo(idxRel);
+ if (attmap == NULL)
+ attmap =
+ convert_tuples_by_name_map(RelationGetDescr(attachrel),
+ RelationGetDescr(rel),
+ gettext_noop("could not convert row type"));
+
+ for (i = 0; i < list_length(attachRelIdxs); i++)
+ {
+ /* already used it */
+ if (attachrelIdxRels[i]->rd_index->indparentidx != 0)
+ continue;
+
+ if (CompareIndexInfo(info, attachInfos[i], attmap))
+ {
+ /* bingo. */
+ IndexSetParentIndex(attachrelIdxRels[i], idx);
+ found = true;
+ break;
+ }
+ }
+ if (!found)
+ {
+ IndexStmt *stmt;
+
+ stmt = generateClonedIndexStmt(NULL, RelationGetRelid(attachrel),
+ idxRel, attmap,
+ RelationGetDescr(rel)->natts);
+ DefineIndex(RelationGetRelid(attachrel), stmt, InvalidOid,
+ RelationGetRelid(idxRel),
+ false, false, false, false, false);
+ }
+
+ index_close(idxRel, AccessShareLock);
+ }
+
+ /* Clean up. */
+ if (attmap)
+ pfree(attmap);
+
+ for (i = 0; i < list_length(attachRelIdxs); i++)
+ {
+ pfree(attachInfos[i]);
+ index_close(attachrelIdxRels[i], AccessShareLock);
+ }
+
+ if (idxes)
+ pfree(idxes);
+ if (attachRelIdxs)
+ pfree(attachRelIdxs);
+ }
+
+ /*
* Generate partition constraint from the partition bound specification.
* If the parent itself is a partition, make sure to include its
* constraint as well.
@@ -14092,6 +14285,8 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
new_repl[Natts_pg_class];
ObjectAddress address;
Oid defaultPartOid;
+ List *indexes;
+ ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
@@ -14153,6 +14348,25 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
}
}
+ /* detach indexes too */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx = index_open(idxid, AccessExclusiveLock);
+
+ if (idx->rd_index->indparentidx != InvalidOid)
+ {
+ Assert(IndexGetRelation(idx->rd_index->indparentidx, false) ==
+ RelationGetRelid(rel));
+
+ IndexSetParentIndex(idx, InvalidOid);
+ }
+
+ relation_close(idx, AccessExclusiveLock);
+ }
+
+
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
@@ -14166,3 +14380,256 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
return address;
}
+
+/*
+ * Before acquiring lock on an index, acquire the same lock on the owning
+ * table.
+ */
+struct AttachIndexCallbackState
+{
+ Oid partitionOid;
+ Oid parentTblOid;
+ bool lockedParentTbl;
+};
+
+static void
+RangeVarCallbackForAttachIndex(const RangeVar *rv, Oid relOid, Oid oldRelOid,
+ void *arg)
+{
+ struct AttachIndexCallbackState *state;
+ Form_pg_class classform;
+ HeapTuple tuple;
+
+ state = (struct AttachIndexCallbackState *) arg;
+
+ if (!state->lockedParentTbl)
+ {
+ LockRelationOid(state->parentTblOid, AccessShareLock);
+ state->lockedParentTbl = true;
+ }
+
+ /*
+ * If we previously locked some other heap, and the name we're looking up
+ * no longer refers to an index on that relation, release the now-useless
+ * lock. XXX maybe we should do *after* we verify whether the index does
+ * not actually belong to the same relation ...
+ */
+ if (relOid != oldRelOid && OidIsValid(state->partitionOid))
+ {
+ UnlockRelationOid(state->partitionOid, AccessShareLock);
+ state->partitionOid = InvalidOid;
+ }
+
+ /* Didn't find a relation, so no need for locking or permission checks. */
+ if (!OidIsValid(relOid))
+ return;
+
+ tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(relOid));
+ if (!HeapTupleIsValid(tuple))
+ return; /* concurrently dropped, so nothing to do */
+ classform = (Form_pg_class) GETSTRUCT(tuple);
+ if (classform->relkind != RELKIND_PARTITIONED_INDEX &&
+ classform->relkind != RELKIND_INDEX)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_OBJECT_DEFINITION),
+ errmsg("\"%s\" is not an index", rv->relname)));
+ ReleaseSysCache(tuple);
+
+ /*
+ * Since we need only examine the heap's tupledesc, an access share lock
+ * on it (preventing any DDL) is sufficient.
+ */
+ state->partitionOid = IndexGetRelation(relOid, false);
+ LockRelationOid(state->partitionOid, AccessShareLock);
+}
+
+/*
+ * ALTER INDEX i1 ATTACH PARTITION i2
+ */
+static ObjectAddress
+ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
+{
+ Relation partIdx;
+ Relation partTbl;
+ Relation parentTbl;
+ ObjectAddress address;
+ Oid partIdxId;
+ struct AttachIndexCallbackState state;
+
+ /*
+ * We need to obtain lock on the index 'name' to modify it, but we also
+ * need to read its owning table's tuple descriptor -- so we need to lock
+ * both. To avoid deadlocks, obtain lock on the table before doing so on
+ * the index. Furthermore, we need to examine the parent table of the
+ * partition, so lock that one too.
+ */
+ state.partitionOid = InvalidOid;
+ state.parentTblOid = parentIdx->rd_index->indrelid;
+ state.lockedParentTbl = false;
+ partIdxId =
+ RangeVarGetRelidExtended(name, AccessExclusiveLock, false, false,
+ RangeVarCallbackForAttachIndex,
+ (void *) &state);
+ /* Not there? */
+ if (!OidIsValid(partIdxId))
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_OBJECT),
+ errmsg("index \"%s\" does not exist", name->relname)));
+
+ /* no deadlock risk: our callback above already acquired the lock */
+ partIdx = relation_open(partIdxId, AccessExclusiveLock);
+
+ /* we already hold lock on both tables, so this is safe: */
+ parentTbl = relation_open(parentIdx->rd_index->indrelid, AccessShareLock);
+ partTbl = relation_open(partIdx->rd_index->indrelid, NoLock);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partIdx));
+
+ /* Silently do nothing if already the right state */
+ if (partIdx->rd_index->indparentidx != RelationGetRelid(parentIdx))
+ {
+ IndexInfo *childInfo;
+ IndexInfo *parentInfo;
+ AttrNumber *attmap;
+ bool found;
+ int i;
+ PartitionDesc partDesc;
+
+ /*
+ * If this partition already has an index attached, refuse the operation.
+ */
+ refuseDupeIndexAttach(parentIdx, partIdx, partTbl);
+
+ if (OidIsValid(partIdx->rd_index->indparentidx))
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot attach index \"%s\" as a partition of index \"%s\"",
+ RelationGetRelationName(partIdx),
+ RelationGetRelationName(parentIdx)),
+ errdetail("Index \"%s\" is already attached to another index.",
+ RelationGetRelationName(partIdx))));
+
+ /* Make sure it indexes a partition of the other index's table */
+ partDesc = RelationGetPartitionDesc(parentTbl);
+ found = false;
+ for (i = 0; i < partDesc->nparts; i++)
+ {
+ if (partDesc->oids[i] == state.partitionOid)
+ {
+ found = true;
+ break;
+ }
+ }
+ if (!found)
+ ereport(ERROR,
+ (errmsg("cannot attach index \"%s\" as a partition of index \"%s\"",
+ RelationGetRelationName(partIdx),
+ RelationGetRelationName(parentIdx)),
+ errdetail("Index \"%s\" is not on a partition of table \"%s\".",
+ RelationGetRelationName(partIdx),
+ RelationGetRelationName(parentTbl))));
+
+ /* Ensure the indexes are compatible */
+ childInfo = BuildIndexInfo(partIdx);
+ parentInfo = BuildIndexInfo(parentIdx);
+ attmap = convert_tuples_by_name_map(RelationGetDescr(parentTbl),
+ RelationGetDescr(partTbl),
+ gettext_noop("could not convert row type"));
+ if (!CompareIndexInfo(parentInfo, childInfo, attmap))
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_OBJECT_DEFINITION),
+ errmsg("cannot attach index \"%s\" as a partition of index \"%s\"",
+ RelationGetRelationName(partIdx),
+ RelationGetRelationName(parentIdx)),
+ errdetail("The index definitions do not match.")));
+
+ /* All good -- do it */
+ IndexSetParentIndex(partIdx, RelationGetRelid(parentIdx));
+ pfree(attmap);
+
+ CommandCounterIncrement();
+
+ validatePartitionedIndex(parentIdx, parentTbl);
+ }
+
+ relation_close(parentTbl, AccessShareLock);
+ /* keep these locks till commit */
+ relation_close(partTbl, NoLock);
+ relation_close(partIdx, NoLock);
+
+ return address;
+}
+
+/*
+ * Verify whether the given partition already contains an index attached
+ * to the given partitioned index. If so, raise an error.
+ */
+static void
+refuseDupeIndexAttach(Relation parentIdx, Relation partIdx, Relation partitionTbl)
+{
+ Relation idxRel;
+ SysScanDesc scan;
+ ScanKeyData key[2];
+
+ idxRel = heap_open(IndexRelationId, AccessShareLock);
+ ScanKeyInit(&key[0], Anum_pg_index_indparentidx,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(parentIdx)));
+ ScanKeyInit(&key[1], Anum_pg_index_indrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(partitionTbl)));
+ scan = systable_beginscan(idxRel, IndexParentidxRelidIndexId, true,
+ NULL, 2, key);
+ if (systable_getnext(scan))
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot attach index \"%s\" as a partition of index \"%s\"",
+ RelationGetRelationName(partIdx),
+ RelationGetRelationName(parentIdx)),
+ errdetail("Another index is already attached for partition \"%s\".",
+ RelationGetRelationName(partitionTbl))));
+
+ systable_endscan(scan);
+ heap_close(idxRel, AccessShareLock);
+}
+
+/*
+ * Every time a partitioned index is attached a partition, verify whether the
+ * set is complete. If it is, mark it valid.
+ */
+static void
+validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
+{
+ Relation idxRel;
+ SysScanDesc scan;
+ ScanKeyData key;
+ int tuples = 0;
+ HeapTuple tuple;
+ PartitionDesc partDesc;
+
+ Assert(partedIdx->rd_rel->relkind == RELKIND_PARTITIONED_INDEX);
+
+ idxRel = heap_open(IndexRelationId, RowExclusiveLock);
+ ScanKeyInit(&key, Anum_pg_index_indparentidx,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(partedIdx)));
+ scan = systable_beginscan(idxRel, IndexParentidxRelidIndexId, true,
+ NULL, 1, &key);
+ while ((tuple = systable_getnext(scan)) != NULL)
+ tuples += 1;
+
+ partDesc = RelationGetPartitionDesc(partedTbl);
+ if (tuples == partDesc->nparts)
+ {
+ HeapTuple newtup;
+
+ newtup = heap_copytuple(partedIdx->rd_indextuple);
+ ((Form_pg_index) GETSTRUCT(newtup))->indisvalid = true;
+
+ CatalogTupleUpdate(idxRel, &partedIdx->rd_indextuple->t_self, newtup);
+ }
+
+ systable_endscan(scan);
+
+ heap_close(idxRel, RowExclusiveLock);
+}
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 84d717102d..cdf7005d8d 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -3379,6 +3379,7 @@ _copyIndexStmt(const IndexStmt *from)
COPY_STRING_FIELD(idxname);
COPY_NODE_FIELD(relation);
+ COPY_SCALAR_FIELD(relationId);
COPY_STRING_FIELD(accessMethod);
COPY_STRING_FIELD(tableSpace);
COPY_NODE_FIELD(indexParams);
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index 2e869a9d5d..42a74bbe13 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -1332,6 +1332,7 @@ _equalIndexStmt(const IndexStmt *a, const IndexStmt *b)
{
COMPARE_STRING_FIELD(idxname);
COMPARE_NODE_FIELD(relation);
+ COMPARE_SCALAR_FIELD(relationId);
COMPARE_STRING_FIELD(accessMethod);
COMPARE_STRING_FIELD(tableSpace);
COMPARE_NODE_FIELD(indexParams);
diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index e468d7cc41..5d5f6efad8 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -2650,6 +2650,7 @@ _outIndexStmt(StringInfo str, const IndexStmt *node)
WRITE_STRING_FIELD(idxname);
WRITE_NODE_FIELD(relation);
+ WRITE_OID_FIELD(relationId);
WRITE_STRING_FIELD(accessMethod);
WRITE_STRING_FIELD(tableSpace);
WRITE_NODE_FIELD(indexParams);
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index f7438714c4..7a8371aa47 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -372,7 +372,12 @@ get_relation_info(PlannerInfo *root, Oid relationObjectId, bool inhparent,
* a table, except we can be sure that the index is not larger
* than the table.
*/
- if (info->indpred == NIL)
+ if (indexRelation->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
+ {
+ info->pages = 0;
+ info->tuples = 0;
+ }
+ else if (info->indpred == NIL)
{
info->pages = RelationGetNumberOfBlocks(indexRelation);
info->tuples = rel->tuples;
@@ -387,7 +392,8 @@ get_relation_info(PlannerInfo *root, Oid relationObjectId, bool inhparent,
info->tuples = rel->tuples;
}
- if (info->relam == BTREE_AM_OID)
+ if (indexRelation->rd_rel->relkind == RELKIND_INDEX &&
+ info->relam == BTREE_AM_OID)
{
/* For btrees, get tree height while we have the index open */
info->tree_height = _bt_getrootheight(indexRelation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index ebfc94f896..5fda4c8f68 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -290,7 +290,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
%type <ival> add_drop opt_asc_desc opt_nulls_order
%type <node> alter_table_cmd alter_type_cmd opt_collate_clause
- replica_identity partition_cmd
+ replica_identity partition_cmd index_partition_cmd
%type <list> alter_table_cmds alter_type_cmds
%type <list> alter_identity_column_option_list
%type <defelt> alter_identity_column_option
@@ -1891,6 +1891,15 @@ AlterTableStmt:
n->missing_ok = true;
$$ = (Node *)n;
}
+ | ALTER INDEX qualified_name index_partition_cmd
+ {
+ AlterTableStmt *n = makeNode(AlterTableStmt);
+ n->relation = $3;
+ n->cmds = list_make1($4);
+ n->relkind = OBJECT_INDEX;
+ n->missing_ok = false;
+ $$ = (Node *)n;
+ }
| ALTER INDEX ALL IN_P TABLESPACE name SET TABLESPACE name opt_nowait
{
AlterTableMoveAllStmt *n =
@@ -2025,6 +2034,22 @@ partition_cmd:
}
;
+index_partition_cmd:
+ /* ALTER INDEX <name> ATTACH PARTITION <index_name> */
+ ATTACH PARTITION qualified_name
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_AttachPartition;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ n->def = (Node *) cmd;
+
+ $$ = (Node *) n;
+ }
+ ;
+
alter_table_cmd:
/* ALTER TABLE <name> ADD <coldef> */
ADD_P columnDef
@@ -7330,7 +7355,7 @@ defacl_privilege_target:
*****************************************************************************/
IndexStmt: CREATE opt_unique INDEX opt_concurrently opt_index_name
- ON qualified_name access_method_clause '(' index_params ')'
+ ON relation_expr access_method_clause '(' index_params ')'
opt_reloptions OptTableSpace where_clause
{
IndexStmt *n = makeNode(IndexStmt);
@@ -7338,6 +7363,7 @@ IndexStmt: CREATE opt_unique INDEX opt_concurrently opt_index_name
n->concurrent = $4;
n->idxname = $5;
n->relation = $7;
+ n->relationId = InvalidOid;
n->accessMethod = $8;
n->indexParams = $10;
n->options = $12;
@@ -7356,7 +7382,7 @@ IndexStmt: CREATE opt_unique INDEX opt_concurrently opt_index_name
$$ = (Node *)n;
}
| CREATE opt_unique INDEX opt_concurrently IF_P NOT EXISTS index_name
- ON qualified_name access_method_clause '(' index_params ')'
+ ON relation_expr access_method_clause '(' index_params ')'
opt_reloptions OptTableSpace where_clause
{
IndexStmt *n = makeNode(IndexStmt);
diff --git a/src/backend/parser/parse_utilcmd.c b/src/backend/parser/parse_utilcmd.c
index f67379f8ed..45f6ec2820 100644
--- a/src/backend/parser/parse_utilcmd.c
+++ b/src/backend/parser/parse_utilcmd.c
@@ -118,9 +118,6 @@ static void transformTableLikeClause(CreateStmtContext *cxt,
TableLikeClause *table_like_clause);
static void transformOfType(CreateStmtContext *cxt,
TypeName *ofTypename);
-static IndexStmt *generateClonedIndexStmt(CreateStmtContext *cxt,
- Relation source_idx,
- const AttrNumber *attmap, int attmap_length);
static List *get_collation(Oid collation, Oid actual_datatype);
static List *get_opclass(Oid opclass, Oid actual_datatype);
static void transformIndexConstraints(CreateStmtContext *cxt);
@@ -1185,7 +1182,8 @@ transformTableLikeClause(CreateStmtContext *cxt, TableLikeClause *table_like_cla
parent_index = index_open(parent_index_oid, AccessShareLock);
/* Build CREATE INDEX statement to recreate the parent_index */
- index_stmt = generateClonedIndexStmt(cxt, parent_index,
+ index_stmt = generateClonedIndexStmt(cxt->relation, InvalidOid,
+ parent_index,
attmap, tupleDesc->natts);
/* Copy comment on index, if requested */
@@ -1263,10 +1261,12 @@ transformOfType(CreateStmtContext *cxt, TypeName *ofTypename)
/*
* Generate an IndexStmt node using information from an already existing index
- * "source_idx". Attribute numbers should be adjusted according to attmap.
+ * "source_idx", for the rel identified either by heapRel or heapRelid.
+ *
+ * Attribute numbers should be adjusted according to attmap.
*/
-static IndexStmt *
-generateClonedIndexStmt(CreateStmtContext *cxt, Relation source_idx,
+IndexStmt *
+generateClonedIndexStmt(RangeVar *heapRel, Oid heapRelid, Relation source_idx,
const AttrNumber *attmap, int attmap_length)
{
Oid source_relid = RelationGetRelid(source_idx);
@@ -1287,6 +1287,9 @@ generateClonedIndexStmt(CreateStmtContext *cxt, Relation source_idx,
Datum datum;
bool isnull;
+ Assert((heapRel == NULL && OidIsValid(heapRelid)) ||
+ (heapRel != NULL && !OidIsValid(heapRelid)));
+
/*
* Fetch pg_class tuple of source index. We can't use the copy in the
* relcache entry because it doesn't include optional fields.
@@ -1322,7 +1325,8 @@ generateClonedIndexStmt(CreateStmtContext *cxt, Relation source_idx,
/* Begin building the IndexStmt */
index = makeNode(IndexStmt);
- index->relation = cxt->relation;
+ index->relation = heapRel;
+ index->relationId = heapRelid;
index->accessMethod = pstrdup(NameStr(amrec->amname));
if (OidIsValid(idxrelrec->reltablespace))
index->tableSpace = get_tablespace_name(idxrelrec->reltablespace);
@@ -3289,18 +3293,39 @@ transformPartitionCmd(CreateStmtContext *cxt, PartitionCmd *cmd)
{
Relation parentRel = cxt->rel;
- /* the table must be partitioned */
- if (parentRel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_OBJECT_DEFINITION),
- errmsg("\"%s\" is not partitioned",
- RelationGetRelationName(parentRel))));
-
- /* transform the partition bound, if any */
- Assert(RelationGetPartitionKey(parentRel) != NULL);
- if (cmd->bound != NULL)
- cxt->partbound = transformPartitionBound(cxt->pstate, parentRel,
- cmd->bound);
+ switch (parentRel->rd_rel->relkind)
+ {
+ case RELKIND_PARTITIONED_TABLE:
+ /* transform the partition bound, if any */
+ Assert(RelationGetPartitionKey(parentRel) != NULL);
+ if (cmd->bound != NULL)
+ cxt->partbound = transformPartitionBound(cxt->pstate, parentRel,
+ cmd->bound);
+ break;
+ case RELKIND_PARTITIONED_INDEX:
+ /* nothing to check */
+ Assert(cmd->bound == NULL);
+ break;
+ case RELKIND_RELATION:
+ /* the table must be partitioned */
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_OBJECT_DEFINITION),
+ errmsg("table \"%s\" is not partitioned",
+ RelationGetRelationName(parentRel))));
+ break;
+ case RELKIND_INDEX:
+ /* the index must be partitioned */
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_OBJECT_DEFINITION),
+ errmsg("index \"%s\" is not partitioned",
+ RelationGetRelationName(parentRel))));
+ break;
+ default:
+ /* parser shouldn't let this case through */
+ elog(ERROR, "\"%s\" is not a partitioned table or index",
+ RelationGetRelationName(parentRel));
+ break;
+ }
}
/*
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 4da1f8f643..8b5fd95a96 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -23,6 +23,7 @@
#include "access/xlog.h"
#include "catalog/catalog.h"
#include "catalog/namespace.h"
+#include "catalog/pg_inherits_fn.h"
#include "catalog/toasting.h"
#include "commands/alter.h"
#include "commands/async.h"
@@ -1300,6 +1301,7 @@ ProcessUtilitySlow(ParseState *pstate,
IndexStmt *stmt = (IndexStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ List *inheritors = NIL;
if (stmt->concurrent)
PreventTransactionChain(isTopLevel,
@@ -1321,6 +1323,9 @@ ProcessUtilitySlow(ParseState *pstate,
false, false,
RangeVarCallbackOwnsRelation,
NULL);
+ /* Also, lock any descendant tables if recursive */
+ if (stmt->relation->inh)
+ inheritors = find_all_inheritors(relid, lockmode, NULL);
/* Run parse analysis ... */
stmt = transformIndexStmt(relid, stmt, queryString);
@@ -1331,6 +1336,7 @@ ProcessUtilitySlow(ParseState *pstate,
DefineIndex(relid, /* OID of heap relation */
stmt,
InvalidOid, /* no predefined OID */
+ InvalidOid, /* no parent index */
false, /* is_alter_table */
true, /* check_rights */
true, /* check_not_in_use */
@@ -1346,6 +1352,9 @@ ProcessUtilitySlow(ParseState *pstate,
parsetree);
commandCollected = true;
EventTriggerAlterTableEnd();
+
+ if (inheritors)
+ list_free(inheritors);
}
break;
diff --git a/src/backend/utils/adt/amutils.c b/src/backend/utils/adt/amutils.c
index f53b251b30..6af53402ce 100644
--- a/src/backend/utils/adt/amutils.c
+++ b/src/backend/utils/adt/amutils.c
@@ -183,7 +183,8 @@ indexam_property(FunctionCallInfo fcinfo,
if (!HeapTupleIsValid(tuple))
PG_RETURN_NULL();
rd_rel = (Form_pg_class) GETSTRUCT(tuple);
- if (rd_rel->relkind != RELKIND_INDEX)
+ if (rd_rel->relkind != RELKIND_INDEX &&
+ rd_rel->relkind != RELKIND_PARTITIONED_INDEX)
{
ReleaseSysCache(tuple);
PG_RETURN_NULL();
diff --git a/src/backend/utils/adt/ruleutils.c b/src/backend/utils/adt/ruleutils.c
index 8514c21c40..a4874bdfb6 100644
--- a/src/backend/utils/adt/ruleutils.c
+++ b/src/backend/utils/adt/ruleutils.c
@@ -317,7 +317,7 @@ static void decompile_column_index_array(Datum column_index_array, Oid relId,
static char *pg_get_ruledef_worker(Oid ruleoid, int prettyFlags);
static char *pg_get_indexdef_worker(Oid indexrelid, int colno,
const Oid *excludeOps,
- bool attrsOnly, bool showTblSpc,
+ bool attrsOnly, bool showTblSpc, bool inherits,
int prettyFlags, bool missing_ok);
static char *pg_get_statisticsobj_worker(Oid statextid, bool missing_ok);
static char *pg_get_partkeydef_worker(Oid relid, int prettyFlags,
@@ -1086,7 +1086,7 @@ pg_get_indexdef(PG_FUNCTION_ARGS)
prettyFlags = PRETTYFLAG_INDENT;
- res = pg_get_indexdef_worker(indexrelid, 0, NULL, false, false,
+ res = pg_get_indexdef_worker(indexrelid, 0, NULL, false, false, false,
prettyFlags, true);
if (res == NULL)
@@ -1107,7 +1107,7 @@ pg_get_indexdef_ext(PG_FUNCTION_ARGS)
prettyFlags = pretty ? PRETTYFLAG_PAREN | PRETTYFLAG_INDENT : PRETTYFLAG_INDENT;
res = pg_get_indexdef_worker(indexrelid, colno, NULL, colno != 0, false,
- prettyFlags, true);
+ false, prettyFlags, true);
if (res == NULL)
PG_RETURN_NULL();
@@ -1123,7 +1123,7 @@ pg_get_indexdef_ext(PG_FUNCTION_ARGS)
char *
pg_get_indexdef_string(Oid indexrelid)
{
- return pg_get_indexdef_worker(indexrelid, 0, NULL, false, true, 0, false);
+ return pg_get_indexdef_worker(indexrelid, 0, NULL, false, true, true, 0, false);
}
/* Internal version that just reports the column definitions */
@@ -1133,7 +1133,7 @@ pg_get_indexdef_columns(Oid indexrelid, bool pretty)
int prettyFlags;
prettyFlags = pretty ? PRETTYFLAG_PAREN | PRETTYFLAG_INDENT : PRETTYFLAG_INDENT;
- return pg_get_indexdef_worker(indexrelid, 0, NULL, true, false,
+ return pg_get_indexdef_worker(indexrelid, 0, NULL, true, false, false,
prettyFlags, false);
}
@@ -1146,7 +1146,7 @@ pg_get_indexdef_columns(Oid indexrelid, bool pretty)
static char *
pg_get_indexdef_worker(Oid indexrelid, int colno,
const Oid *excludeOps,
- bool attrsOnly, bool showTblSpc,
+ bool attrsOnly, bool showTblSpc, bool inherits,
int prettyFlags, bool missing_ok)
{
/* might want a separate isConstraint parameter later */
@@ -1259,9 +1259,11 @@ pg_get_indexdef_worker(Oid indexrelid, int colno,
if (!attrsOnly)
{
if (!isConstraint)
- appendStringInfo(&buf, "CREATE %sINDEX %s ON %s USING %s (",
+ appendStringInfo(&buf, "CREATE %sINDEX %s ON %s%s USING %s (",
idxrec->indisunique ? "UNIQUE " : "",
quote_identifier(NameStr(idxrelrec->relname)),
+ idxrelrec->relkind == RELKIND_PARTITIONED_INDEX
+ && !inherits ? "ONLY " : "",
generate_relation_name(indrelid, NIL),
quote_identifier(NameStr(amrec->amname)));
else /* currently, must be EXCLUDE constraint */
@@ -2148,6 +2150,7 @@ pg_get_constraintdef_worker(Oid constraintId, bool fullCommand,
operators,
false,
false,
+ false,
prettyFlags,
false));
break;
diff --git a/src/backend/utils/cache/relcache.c b/src/backend/utils/cache/relcache.c
index 1d0cc6cb79..5afa1b117b 100644
--- a/src/backend/utils/cache/relcache.c
+++ b/src/backend/utils/cache/relcache.c
@@ -430,6 +430,7 @@ static void
RelationParseRelOptions(Relation relation, HeapTuple tuple)
{
bytea *options;
+ bool isindex;
relation->rd_options = NULL;
@@ -439,6 +440,7 @@ RelationParseRelOptions(Relation relation, HeapTuple tuple)
case RELKIND_RELATION:
case RELKIND_TOASTVALUE:
case RELKIND_INDEX:
+ case RELKIND_PARTITIONED_INDEX:
case RELKIND_VIEW:
case RELKIND_MATVIEW:
case RELKIND_PARTITIONED_TABLE:
@@ -452,10 +454,12 @@ RelationParseRelOptions(Relation relation, HeapTuple tuple)
* we might not have any other for pg_class yet (consider executing this
* code for pg_class itself)
*/
+ isindex = relation->rd_rel->relkind == RELKIND_INDEX ||
+ relation->rd_rel->relkind == RELKIND_PARTITIONED_INDEX;
options = extractRelOptions(tuple,
GetPgClassDescriptor(),
- relation->rd_rel->relkind == RELKIND_INDEX ?
- relation->rd_amroutine->amoptions : NULL);
+ isindex ? relation->rd_amroutine->amoptions :
+ NULL);
/*
* Copy parsed data into CacheMemoryContext. To guard against the
@@ -2053,7 +2057,8 @@ RelationIdGetRelation(Oid relationId)
* and we don't want to use the full-blown procedure because it's
* a headache for indexes that reload itself depends on.
*/
- if (rd->rd_rel->relkind == RELKIND_INDEX)
+ if (rd->rd_rel->relkind == RELKIND_INDEX ||
+ rd->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
RelationReloadIndexInfo(rd);
else
RelationClearRelation(rd, true);
@@ -2167,7 +2172,8 @@ RelationReloadIndexInfo(Relation relation)
Form_pg_class relp;
/* Should be called only for invalidated indexes */
- Assert(relation->rd_rel->relkind == RELKIND_INDEX &&
+ Assert((relation->rd_rel->relkind == RELKIND_INDEX ||
+ relation->rd_rel->relkind == RELKIND_PARTITIONED_INDEX) &&
!relation->rd_isvalid);
/* Ensure it's closed at smgr level */
@@ -2387,7 +2393,8 @@ RelationClearRelation(Relation relation, bool rebuild)
{
RelationInitPhysicalAddr(relation);
- if (relation->rd_rel->relkind == RELKIND_INDEX)
+ if (relation->rd_rel->relkind == RELKIND_INDEX ||
+ relation->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
{
relation->rd_isvalid = false; /* needs to be revalidated */
if (relation->rd_refcnt > 1 && IsTransactionState())
@@ -2403,7 +2410,8 @@ RelationClearRelation(Relation relation, bool rebuild)
* re-read the pg_class row to handle possible physical relocation of the
* index, and we check for pg_index updates too.
*/
- if (relation->rd_rel->relkind == RELKIND_INDEX &&
+ if ((relation->rd_rel->relkind == RELKIND_INDEX ||
+ relation->rd_rel->relkind == RELKIND_PARTITIONED_INDEX) &&
relation->rd_refcnt > 0 &&
relation->rd_indexcxt != NULL)
{
@@ -5461,7 +5469,10 @@ load_relcache_init_file(bool shared)
rel->rd_att->constr = constr;
}
- /* If it's an index, there's more to do */
+ /*
+ * If it's an index, there's more to do. Note we explicitly ignore
+ * partitioned indexes here.
+ */
if (rel->rd_rel->relkind == RELKIND_INDEX)
{
MemoryContext indexcxt;
@@ -5825,7 +5836,10 @@ write_relcache_init_file(bool shared)
(rel->rd_options ? VARSIZE(rel->rd_options) : 0),
fp);
- /* If it's an index, there's more to do */
+ /*
+ * If it's an index, there's more to do. Note we explicitly ignore
+ * partitioned indexes here.
+ */
if (rel->rd_rel->relkind == RELKIND_INDEX)
{
/* write the pg_index tuple */
diff --git a/src/bin/pg_dump/common.c b/src/bin/pg_dump/common.c
index 4b47951de1..f5b542c4ee 100644
--- a/src/bin/pg_dump/common.c
+++ b/src/bin/pg_dump/common.c
@@ -68,6 +68,7 @@ static int numextmembers;
static void flagInhTables(Archive *fout, TableInfo *tbinfo, int numTables,
InhInfo *inhinfo, int numInherits);
+static void flagInhIndexes(Archive *fout, TableInfo *tblinfo, int numTables);
static void flagInhAttrs(DumpOptions *dopt, TableInfo *tblinfo, int numTables);
static DumpableObject **buildIndexArray(void *objArray, int numObjs,
Size objSize);
@@ -76,6 +77,8 @@ static int ExtensionMemberIdCompare(const void *p1, const void *p2);
static void findParentsByOid(TableInfo *self,
InhInfo *inhinfo, int numInherits);
static int strInArray(const char *pattern, char **arr, int arr_size);
+static IndxInfo *findIndexByOid(Oid oid, DumpableObject **idxinfoindex,
+ int numIndexes);
/*
@@ -258,6 +261,10 @@ getSchemaData(Archive *fout, int *numTablesPtr)
getIndexes(fout, tblinfo, numTables);
if (g_verbose)
+ write_msg(NULL, "flagging indexes in partitioned tables\n");
+ flagInhIndexes(fout, tblinfo, numTables);
+
+ if (g_verbose)
write_msg(NULL, "reading extended statistics\n");
getExtendedStatistics(fout, tblinfo, numTables);
@@ -354,6 +361,91 @@ flagInhTables(Archive *fout, TableInfo *tblinfo, int numTables,
}
}
+/*
+ * flagInhIndexes -
+ * Create AttachIndexInfo objects for partitioned indexes, and add
+ * appropriate dependency links.
+ */
+static void
+flagInhIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
+{
+ int i,
+ j,
+ k;
+ DumpableObject ***parentIndexArray;
+
+ parentIndexArray = (DumpableObject ***)
+ pg_malloc0(getMaxDumpId() * sizeof(DumpableObject **));
+
+ for (i = 0; i < numTables; i++)
+ {
+ TableInfo *parenttbl;
+ IndexAttachInfo *attachinfo;
+
+ if (!tblinfo[i].ispartition || tblinfo[i].numParents == 0)
+ continue;
+
+ Assert(tblinfo[i].numParents == 1);
+ parenttbl = tblinfo[i].parents[0];
+
+ /*
+ * We need access to each parent table's index list, but there is no
+ * index to cover them outside of this function. To avoid having to
+ * sort every parent table's indexes each time we come across each of
+ * its partitions, create an indexed array for each parent the first
+ * time it is required.
+ */
+ if (parentIndexArray[parenttbl->dobj.dumpId] == NULL)
+ parentIndexArray[parenttbl->dobj.dumpId] =
+ buildIndexArray(parenttbl->indexes,
+ parenttbl->numIndexes,
+ sizeof(IndxInfo));
+
+ attachinfo = (IndexAttachInfo *)
+ pg_malloc0(tblinfo[i].numIndexes * sizeof(IndexAttachInfo));
+ for (j = 0, k = 0; j < tblinfo[i].numIndexes; j++)
+ {
+ IndxInfo *index = &(tblinfo[i].indexes[j]);
+ IndxInfo *parentidx;
+
+ if (index->indparentidx == 0)
+ continue;
+
+ parentidx = findIndexByOid(index->indparentidx,
+ parentIndexArray[parenttbl->dobj.dumpId],
+ parenttbl->numIndexes);
+ if (parentidx == NULL)
+ continue;
+
+ attachinfo[k].dobj.objType = DO_INDEX_ATTACH;
+ attachinfo[k].dobj.catId.tableoid = 0;
+ attachinfo[k].dobj.catId.oid = 0;
+ AssignDumpId(&attachinfo[k].dobj);
+ attachinfo[k].dobj.name = pg_strdup(index->dobj.name);
+ attachinfo[k].parentIdx = parentidx;
+ attachinfo[k].partitionIdx = index;
+
+ /*
+ * Normal dependency goes from child idx to parent idx; remove it
+ * to avoid creating a cycle only to resolve it later. We want
+ * one from parent to partition (so that the partition index is
+ * created first), and another one from attach object to parent
+ * (so that the partition index is attached once the parent index
+ * has been created).
+ */
+ addObjectDependency(&parentidx->dobj, index->dobj.dumpId);
+ addObjectDependency(&attachinfo[k].dobj, parentidx->dobj.dumpId);
+
+ k++;
+ }
+ }
+
+ for (i = 0; i < numTables; i++)
+ if (parentIndexArray[i])
+ pg_free(parentIndexArray[i]);
+ pg_free(parentIndexArray);
+}
+
/* flagInhAttrs -
* for each dumpable table in tblinfo, flag its inherited attributes
*
@@ -827,6 +919,18 @@ findExtensionByOid(Oid oid)
return (ExtensionInfo *) findObjectByOid(oid, extinfoindex, numExtensions);
}
+/*
+ * findIndexByOid
+ * find the entry of the index with the given oid
+ *
+ * This one's signature is different from the previous ones because we lack a
+ * global array of all indexes, so caller must pass their array as argument.
+ */
+static IndxInfo *
+findIndexByOid(Oid oid, DumpableObject **idxinfoindex, int numIndexes)
+{
+ return (IndxInfo *) findObjectByOid(oid, idxinfoindex, numIndexes);
+}
/*
* setExtensionMembership
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index e6701aaa78..aeaa58d64f 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -193,6 +193,7 @@ static void dumpAttrDef(Archive *fout, AttrDefInfo *adinfo);
static void dumpSequence(Archive *fout, TableInfo *tbinfo);
static void dumpSequenceData(Archive *fout, TableDataInfo *tdinfo);
static void dumpIndex(Archive *fout, IndxInfo *indxinfo);
+static void dumpIndexAttach(Archive *fout, IndexAttachInfo *attachinfo);
static void dumpStatisticsExt(Archive *fout, StatsExtInfo *statsextinfo);
static void dumpConstraint(Archive *fout, ConstraintInfo *coninfo);
static void dumpTableConstraintComment(Archive *fout, ConstraintInfo *coninfo);
@@ -6509,6 +6510,7 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
int i_tableoid,
i_oid,
i_indexname,
+ i_parentidx,
i_indexdef,
i_indnkeys,
i_indkey,
@@ -6530,10 +6532,6 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
{
TableInfo *tbinfo = &tblinfo[i];
- /* Only plain tables and materialized views have indexes. */
- if (tbinfo->relkind != RELKIND_RELATION &&
- tbinfo->relkind != RELKIND_MATVIEW)
- continue;
if (!tbinfo->hasindex)
continue;
@@ -6561,7 +6559,37 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
* is not.
*/
resetPQExpBuffer(query);
- if (fout->remoteVersion >= 90400)
+ if (fout->remoteVersion >= 11000)
+ {
+ appendPQExpBuffer(query,
+ "SELECT t.tableoid, t.oid, "
+ "t.relname AS indexname, "
+ "i.indparentidx, "
+ "pg_catalog.pg_get_indexdef(i.indexrelid) AS indexdef, "
+ "t.relnatts AS indnkeys, "
+ "i.indkey, i.indisclustered, "
+ "i.indisreplident, t.relpages, "
+ "c.contype, c.conname, "
+ "c.condeferrable, c.condeferred, "
+ "c.tableoid AS contableoid, "
+ "c.oid AS conoid, "
+ "pg_catalog.pg_get_constraintdef(c.oid, false) AS condef, "
+ "(SELECT spcname FROM pg_catalog.pg_tablespace s WHERE s.oid = t.reltablespace) AS tablespace, "
+ "t.reloptions AS indreloptions "
+ "FROM pg_catalog.pg_index i "
+ "JOIN pg_catalog.pg_class t ON (t.oid = i.indexrelid) "
+ "JOIN pg_catalog.pg_class t2 ON (t2.oid = i.indrelid) "
+ "LEFT JOIN pg_catalog.pg_constraint c "
+ "ON (i.indrelid = c.conrelid AND "
+ "i.indexrelid = c.conindid AND "
+ "c.contype IN ('p','u','x')) "
+ "WHERE i.indrelid = '%u'::pg_catalog.oid "
+ "AND (i.indisvalid OR t2.relkind = 'p') "
+ "AND i.indisready "
+ "ORDER BY indexname",
+ tbinfo->dobj.catId.oid);
+ }
+ else if (fout->remoteVersion >= 90400)
{
/*
* the test on indisready is necessary in 9.2, and harmless in
@@ -6570,6 +6598,7 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
appendPQExpBuffer(query,
"SELECT t.tableoid, t.oid, "
"t.relname AS indexname, "
+ "0 AS indparentidx, "
"pg_catalog.pg_get_indexdef(i.indexrelid) AS indexdef, "
"t.relnatts AS indnkeys, "
"i.indkey, i.indisclustered, "
@@ -6601,6 +6630,7 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
appendPQExpBuffer(query,
"SELECT t.tableoid, t.oid, "
"t.relname AS indexname, "
+ "0 AS indparentidx, "
"pg_catalog.pg_get_indexdef(i.indexrelid) AS indexdef, "
"t.relnatts AS indnkeys, "
"i.indkey, i.indisclustered, "
@@ -6628,6 +6658,7 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
appendPQExpBuffer(query,
"SELECT t.tableoid, t.oid, "
"t.relname AS indexname, "
+ "0 AS indparentidx, "
"pg_catalog.pg_get_indexdef(i.indexrelid) AS indexdef, "
"t.relnatts AS indnkeys, "
"i.indkey, i.indisclustered, "
@@ -6658,6 +6689,7 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
appendPQExpBuffer(query,
"SELECT t.tableoid, t.oid, "
"t.relname AS indexname, "
+ "0 AS indparentidx, "
"pg_catalog.pg_get_indexdef(i.indexrelid) AS indexdef, "
"t.relnatts AS indnkeys, "
"i.indkey, i.indisclustered, "
@@ -6690,6 +6722,7 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
i_tableoid = PQfnumber(res, "tableoid");
i_oid = PQfnumber(res, "oid");
i_indexname = PQfnumber(res, "indexname");
+ i_parentidx = PQfnumber(res, "indparentidx");
i_indexdef = PQfnumber(res, "indexdef");
i_indnkeys = PQfnumber(res, "indnkeys");
i_indkey = PQfnumber(res, "indkey");
@@ -6706,8 +6739,10 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
i_tablespace = PQfnumber(res, "tablespace");
i_indreloptions = PQfnumber(res, "indreloptions");
- indxinfo = (IndxInfo *) pg_malloc(ntups * sizeof(IndxInfo));
+ tbinfo->indexes = indxinfo =
+ (IndxInfo *) pg_malloc(ntups * sizeof(IndxInfo));
constrinfo = (ConstraintInfo *) pg_malloc(ntups * sizeof(ConstraintInfo));
+ tbinfo->numIndexes = ntups;
for (j = 0; j < ntups; j++)
{
@@ -6729,6 +6764,7 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
indxinfo[j].indkeys, indxinfo[j].indnkeys);
indxinfo[j].indisclustered = (PQgetvalue(res, j, i_indisclustered)[0] == 't');
indxinfo[j].indisreplident = (PQgetvalue(res, j, i_indisreplident)[0] == 't');
+ indxinfo[j].indparentidx = atooid(PQgetvalue(res, j, i_parentidx));
indxinfo[j].relpages = atoi(PQgetvalue(res, j, i_relpages));
contype = *(PQgetvalue(res, j, i_contype));
@@ -9512,6 +9548,9 @@ dumpDumpableObject(Archive *fout, DumpableObject *dobj)
case DO_INDEX:
dumpIndex(fout, (IndxInfo *) dobj);
break;
+ case DO_INDEX_ATTACH:
+ dumpIndexAttach(fout, (IndexAttachInfo *) dobj);
+ break;
case DO_STATSEXT:
dumpStatisticsExt(fout, (StatsExtInfo *) dobj);
break;
@@ -16173,6 +16212,42 @@ dumpIndex(Archive *fout, IndxInfo *indxinfo)
}
/*
+ * dumpIndexAttach
+ * write out to fout a partitioned-index attachment clause
+ */
+void
+dumpIndexAttach(Archive *fout, IndexAttachInfo *attachinfo)
+{
+ if (fout->dopt->dataOnly)
+ return;
+
+ if (attachinfo->partitionIdx->dobj.dump & DUMP_COMPONENT_DEFINITION)
+ {
+ PQExpBuffer q = createPQExpBuffer();
+
+ appendPQExpBuffer(q, "\nALTER INDEX %s ",
+ fmtQualifiedId(fout->remoteVersion,
+ attachinfo->parentIdx->dobj.namespace->dobj.name,
+ attachinfo->parentIdx->dobj.name));
+ appendPQExpBuffer(q, "ATTACH PARTITION %s;\n",
+ fmtQualifiedId(fout->remoteVersion,
+ attachinfo->partitionIdx->dobj.namespace->dobj.name,
+ attachinfo->partitionIdx->dobj.name));
+
+ ArchiveEntry(fout, attachinfo->dobj.catId, attachinfo->dobj.dumpId,
+ attachinfo->dobj.name,
+ NULL, NULL,
+ "",
+ false, "INDEX ATTACH", SECTION_POST_DATA,
+ q->data, "", NULL,
+ NULL, 0,
+ NULL, NULL);
+
+ destroyPQExpBuffer(q);
+ }
+}
+
+/*
* dumpStatisticsExt
* write out to fout an extended statistics object
*/
@@ -17803,6 +17878,7 @@ addBoundaryDependencies(DumpableObject **dobjs, int numObjs,
addObjectDependency(postDataBound, dobj->dumpId);
break;
case DO_INDEX:
+ case DO_INDEX_ATTACH:
case DO_STATSEXT:
case DO_REFRESH_MATVIEW:
case DO_TRIGGER:
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index da884ffd09..f9342058aa 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -56,6 +56,7 @@ typedef enum
DO_TABLE,
DO_ATTRDEF,
DO_INDEX,
+ DO_INDEX_ATTACH,
DO_STATSEXT,
DO_RULE,
DO_TRIGGER,
@@ -328,6 +329,8 @@ typedef struct _tableInfo
*/
int numParents; /* number of (immediate) parent tables */
struct _tableInfo **parents; /* TableInfos of immediate parents */
+ int numIndexes; /* number of indexes */
+ struct _indxInfo *indexes; /* indexes */
struct _tableDataInfo *dataObj; /* TableDataInfo, if dumping its data */
int numTriggers; /* number of triggers for table */
struct _triggerInfo *triggers; /* array of TriggerInfo structs */
@@ -361,11 +364,19 @@ typedef struct _indxInfo
Oid *indkeys;
bool indisclustered;
bool indisreplident;
+ Oid indparentidx; /* if partitioned, parent index OID */
/* if there is an associated constraint object, its dumpId: */
DumpId indexconstraint;
int relpages; /* relpages of the underlying table */
} IndxInfo;
+typedef struct _indexAttachInfo
+{
+ DumpableObject dobj;
+ IndxInfo *parentIdx; /* link to index on partitioned table */
+ IndxInfo *partitionIdx; /* link to index on partition */
+} IndexAttachInfo;
+
typedef struct _statsExtInfo
{
DumpableObject dobj;
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index 48b6dd594c..a4f6db3ae2 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -35,6 +35,10 @@ static const char *modulename = gettext_noop("sorter");
* pg_dump.c; that is, PRE_DATA objects must sort before DO_PRE_DATA_BOUNDARY,
* POST_DATA objects must sort after DO_POST_DATA_BOUNDARY, and DATA objects
* must sort between them.
+ *
+ * Note: sortDataAndIndexObjectsBySize wants to have all DO_TABLE_DATA and
+ * DO_INDEX objects in contiguous chunks, so do not reuse the values for those
+ * for other object types.
*/
static const int dbObjectTypePriority[] =
{
@@ -53,11 +57,12 @@ static const int dbObjectTypePriority[] =
18, /* DO_TABLE */
20, /* DO_ATTRDEF */
28, /* DO_INDEX */
- 29, /* DO_STATSEXT */
- 30, /* DO_RULE */
- 31, /* DO_TRIGGER */
+ 29, /* DO_INDEX_ATTACH */
+ 30, /* DO_STATSEXT */
+ 31, /* DO_RULE */
+ 32, /* DO_TRIGGER */
27, /* DO_CONSTRAINT */
- 32, /* DO_FK_CONSTRAINT */
+ 33, /* DO_FK_CONSTRAINT */
2, /* DO_PROCLANG */
10, /* DO_CAST */
23, /* DO_TABLE_DATA */
@@ -69,18 +74,18 @@ static const int dbObjectTypePriority[] =
15, /* DO_TSCONFIG */
16, /* DO_FDW */
17, /* DO_FOREIGN_SERVER */
- 32, /* DO_DEFAULT_ACL */
+ 33, /* DO_DEFAULT_ACL */
3, /* DO_TRANSFORM */
21, /* DO_BLOB */
25, /* DO_BLOB_DATA */
22, /* DO_PRE_DATA_BOUNDARY */
26, /* DO_POST_DATA_BOUNDARY */
- 33, /* DO_EVENT_TRIGGER */
- 38, /* DO_REFRESH_MATVIEW */
- 34, /* DO_POLICY */
- 35, /* DO_PUBLICATION */
- 36, /* DO_PUBLICATION_REL */
- 37 /* DO_SUBSCRIPTION */
+ 34, /* DO_EVENT_TRIGGER */
+ 39, /* DO_REFRESH_MATVIEW */
+ 35, /* DO_POLICY */
+ 36, /* DO_PUBLICATION */
+ 37, /* DO_PUBLICATION_REL */
+ 38 /* DO_SUBSCRIPTION */
};
static DumpId preDataBoundId;
@@ -937,6 +942,13 @@ repairDomainConstraintMultiLoop(DumpableObject *domainobj,
addObjectDependency(constraintobj, postDataBoundId);
}
+static void
+repairIndexLoop(DumpableObject *partedindex,
+ DumpableObject *partindex)
+{
+ removeObjectDependency(partedindex, partindex->dumpId);
+}
+
/*
* Fix a dependency loop, or die trying ...
*
@@ -1099,6 +1111,23 @@ repairDependencyLoop(DumpableObject **loop,
return;
}
+ /* index on partitioned table and corresponding index on partition */
+ if (nLoop == 2 &&
+ loop[0]->objType == DO_INDEX &&
+ loop[1]->objType == DO_INDEX)
+ {
+ if (((IndxInfo *) loop[0])->indparentidx == loop[1]->catId.oid)
+ {
+ repairIndexLoop(loop[0], loop[1]);
+ return;
+ }
+ else if (((IndxInfo *) loop[1])->indparentidx == loop[0]->catId.oid)
+ {
+ repairIndexLoop(loop[1], loop[0]);
+ return;
+ }
+ }
+
/* Indirect loop involving table and attribute default */
if (nLoop > 2)
{
@@ -1292,6 +1321,11 @@ describeDumpableObject(DumpableObject *obj, char *buf, int bufsize)
"INDEX %s (ID %d OID %u)",
obj->name, obj->dumpId, obj->catId.oid);
return;
+ case DO_INDEX_ATTACH:
+ snprintf(buf, bufsize,
+ "INDEX ATTACH %s (ID %d)",
+ obj->name, obj->dumpId);
+ return;
case DO_STATSEXT:
snprintf(buf, bufsize,
"STATISTICS %s (ID %d OID %u)",
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 3fc69c46c0..4ab1fb2af1 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -1705,7 +1705,8 @@ describeOneTableDetails(const char *schemaname,
appendPQExpBufferStr(&buf, ",\n a.attidentity");
else
appendPQExpBufferStr(&buf, ",\n ''::pg_catalog.char AS attidentity");
- if (tableinfo.relkind == RELKIND_INDEX)
+ if (tableinfo.relkind == RELKIND_INDEX ||
+ tableinfo.relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ",\n pg_catalog.pg_get_indexdef(a.attrelid, a.attnum, TRUE) AS indexdef");
else
appendPQExpBufferStr(&buf, ",\n NULL AS indexdef");
@@ -1766,6 +1767,7 @@ describeOneTableDetails(const char *schemaname,
schemaname, relationname);
break;
case RELKIND_INDEX:
+ case RELKIND_PARTITIONED_INDEX:
if (tableinfo.relpersistence == 'u')
printfPQExpBuffer(&title, _("Unlogged index \"%s.%s\""),
schemaname, relationname);
@@ -1823,7 +1825,8 @@ describeOneTableDetails(const char *schemaname,
show_column_details = true;
}
- if (tableinfo.relkind == RELKIND_INDEX)
+ if (tableinfo.relkind == RELKIND_INDEX ||
+ tableinfo.relkind == RELKIND_PARTITIONED_INDEX)
headers[cols++] = gettext_noop("Definition");
if (tableinfo.relkind == RELKIND_FOREIGN_TABLE && pset.sversion >= 90200)
@@ -1834,6 +1837,7 @@ describeOneTableDetails(const char *schemaname,
headers[cols++] = gettext_noop("Storage");
if (tableinfo.relkind == RELKIND_RELATION ||
tableinfo.relkind == RELKIND_INDEX ||
+ tableinfo.relkind == RELKIND_PARTITIONED_INDEX ||
tableinfo.relkind == RELKIND_MATVIEW ||
tableinfo.relkind == RELKIND_FOREIGN_TABLE ||
tableinfo.relkind == RELKIND_PARTITIONED_TABLE)
@@ -1906,7 +1910,8 @@ describeOneTableDetails(const char *schemaname,
}
/* Expression for index column */
- if (tableinfo.relkind == RELKIND_INDEX)
+ if (tableinfo.relkind == RELKIND_INDEX ||
+ tableinfo.relkind == RELKIND_PARTITIONED_INDEX)
printTableAddCell(&cont, PQgetvalue(res, i, 7), false, false);
/* FDW options for foreign table column, only for 9.2 or later */
@@ -1930,6 +1935,7 @@ describeOneTableDetails(const char *schemaname,
/* Statistics target, if the relkind supports this feature */
if (tableinfo.relkind == RELKIND_RELATION ||
tableinfo.relkind == RELKIND_INDEX ||
+ tableinfo.relkind == RELKIND_PARTITIONED_INDEX ||
tableinfo.relkind == RELKIND_MATVIEW ||
tableinfo.relkind == RELKIND_FOREIGN_TABLE ||
tableinfo.relkind == RELKIND_PARTITIONED_TABLE)
@@ -2021,7 +2027,8 @@ describeOneTableDetails(const char *schemaname,
PQclear(result);
}
- if (tableinfo.relkind == RELKIND_INDEX)
+ if (tableinfo.relkind == RELKIND_INDEX ||
+ tableinfo.relkind == RELKIND_PARTITIONED_INDEX)
{
/* Footer information about an index */
PGresult *result;
@@ -3397,6 +3404,7 @@ listTables(const char *tabtypes, const char *pattern, bool verbose, bool showSys
" WHEN 's' THEN '%s'"
" WHEN " CppAsString2(RELKIND_FOREIGN_TABLE) " THEN '%s'"
" WHEN " CppAsString2(RELKIND_PARTITIONED_TABLE) " THEN '%s'"
+ " WHEN " CppAsString2(RELKIND_PARTITIONED_INDEX) " THEN '%s'"
" END as \"%s\",\n"
" pg_catalog.pg_get_userbyid(c.relowner) as \"%s\"",
gettext_noop("Schema"),
@@ -3409,6 +3417,7 @@ listTables(const char *tabtypes, const char *pattern, bool verbose, bool showSys
gettext_noop("special"),
gettext_noop("foreign table"),
gettext_noop("table"), /* partitioned table */
+ gettext_noop("index"), /* partitioned index */
gettext_noop("Type"),
gettext_noop("Owner"));
@@ -3454,7 +3463,8 @@ listTables(const char *tabtypes, const char *pattern, bool verbose, bool showSys
if (showMatViews)
appendPQExpBufferStr(&buf, CppAsString2(RELKIND_MATVIEW) ",");
if (showIndexes)
- appendPQExpBufferStr(&buf, CppAsString2(RELKIND_INDEX) ",");
+ appendPQExpBufferStr(&buf, CppAsString2(RELKIND_INDEX) ","
+ CppAsString2(RELKIND_PARTITIONED_INDEX) ",");
if (showSeq)
appendPQExpBufferStr(&buf, CppAsString2(RELKIND_SEQUENCE) ",");
if (showSystem || pattern)
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 468e50aa31..4d7cb1680f 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -412,7 +412,8 @@ static const SchemaQuery Query_for_list_of_indexes = {
/* catname */
"pg_catalog.pg_class c",
/* selcondition */
- "c.relkind IN (" CppAsString2(RELKIND_INDEX) ")",
+ "c.relkind IN (" CppAsString2(RELKIND_INDEX) ", "
+ CppAsString2(RELKIND_PARTITIONED_INDEX) ")",
/* viscondition */
"pg_catalog.pg_table_is_visible(c.oid)",
/* namespace */
@@ -600,6 +601,23 @@ static const SchemaQuery Query_for_list_of_tmf = {
NULL
};
+static const SchemaQuery Query_for_list_of_tpm = {
+ /* catname */
+ "pg_catalog.pg_class c",
+ /* selcondition */
+ "c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
+ CppAsString2(RELKIND_MATVIEW) ")",
+ /* viscondition */
+ "pg_catalog.pg_table_is_visible(c.oid)",
+ /* namespace */
+ "c.relnamespace",
+ /* result */
+ "pg_catalog.quote_ident(c.relname)",
+ /* qualresult */
+ NULL
+};
+
static const SchemaQuery Query_for_list_of_tm = {
/* catname */
"pg_catalog.pg_class c",
@@ -1676,7 +1694,12 @@ psql_completion(const char *text, int start, int end)
"UNION SELECT 'ALL IN TABLESPACE'");
/* ALTER INDEX <name> */
else if (Matches3("ALTER", "INDEX", MatchAny))
- COMPLETE_WITH_LIST5("ALTER COLUMN", "OWNER TO", "RENAME TO", "SET", "RESET");
+ COMPLETE_WITH_LIST6("ALTER COLUMN", "OWNER TO", "RENAME TO", "SET",
+ "RESET", "ATTACH PARTITION");
+ else if (Matches4("ALTER", "INDEX", MatchAny, "ATTACH"))
+ COMPLETE_WITH_CONST("PARTITION");
+ else if (Matches5("ALTER", "INDEX", MatchAny, "ATTACH", "PARTITION"))
+ COMPLETE_WITH_SCHEMA_QUERY(Query_for_list_of_indexes, NULL);
/* ALTER INDEX <name> ALTER COLUMN <colnum> */
else if (Matches6("ALTER", "INDEX", MatchAny, "ALTER", "COLUMN", MatchAny))
COMPLETE_WITH_CONST("SET STATISTICS");
@@ -2338,10 +2361,13 @@ psql_completion(const char *text, int start, int end)
COMPLETE_WITH_SCHEMA_QUERY(Query_for_list_of_indexes,
" UNION SELECT 'ON'"
" UNION SELECT 'CONCURRENTLY'");
- /* Complete ... INDEX|CONCURRENTLY [<name>] ON with a list of tables */
+ /*
+ * Complete ... INDEX|CONCURRENTLY [<name>] ON with a list of relations
+ * that can indexes can be created on
+ */
else if (TailMatches3("INDEX|CONCURRENTLY", MatchAny, "ON") ||
TailMatches2("INDEX|CONCURRENTLY", "ON"))
- COMPLETE_WITH_SCHEMA_QUERY(Query_for_list_of_tm, NULL);
+ COMPLETE_WITH_SCHEMA_QUERY(Query_for_list_of_tpm, NULL);
/*
* Complete CREATE|UNIQUE INDEX CONCURRENTLY with "ON" and existing
diff --git a/src/include/catalog/catversion.h b/src/include/catalog/catversion.h
index 3934582efc..d8d4f11be0 100644
--- a/src/include/catalog/catversion.h
+++ b/src/include/catalog/catversion.h
@@ -53,6 +53,6 @@
*/
/* yyyymmddN */
-#define CATALOG_VERSION_NO 201712251
+#define CATALOG_VERSION_NO 201712291
#endif
diff --git a/src/include/catalog/dependency.h b/src/include/catalog/dependency.h
index b9f98423cc..df7763cb50 100644
--- a/src/include/catalog/dependency.h
+++ b/src/include/catalog/dependency.h
@@ -49,6 +49,13 @@
* Example: a trigger that's created to enforce a foreign-key constraint
* is made internally dependent on the constraint's pg_constraint entry.
*
+ * DEPENDENCY_INTERNAL_AUTO ('I'): the dependent object was created as
+ * part of creation of the referenced object, and is really just a part
+ * of its internal implementation. A DROP of the dependent object will
+ * be disallowed outright (we'll tell the user to issue a DROP against the
+ * referenced object, instead). As opposed to DROP_INTERNAL [... explain
+ * the difference ... ]
+ *
* DEPENDENCY_EXTENSION ('e'): the dependent object is a member of the
* extension that is the referenced object. The dependent object can be
* dropped only via DROP EXTENSION on the referenced object. Functionally
@@ -75,6 +82,7 @@ typedef enum DependencyType
DEPENDENCY_NORMAL = 'n',
DEPENDENCY_AUTO = 'a',
DEPENDENCY_INTERNAL = 'i',
+ DEPENDENCY_INTERNAL_AUTO = 'I',
DEPENDENCY_EXTENSION = 'e',
DEPENDENCY_AUTO_EXTENSION = 'x',
DEPENDENCY_PIN = 'p'
diff --git a/src/include/catalog/index.h b/src/include/catalog/index.h
index ceaa91f1b2..112d69debc 100644
--- a/src/include/catalog/index.h
+++ b/src/include/catalog/index.h
@@ -47,10 +47,13 @@ extern void index_check_primary_key(Relation heapRel,
#define INDEX_CREATE_SKIP_BUILD (1 << 2)
#define INDEX_CREATE_CONCURRENT (1 << 3)
#define INDEX_CREATE_IF_NOT_EXISTS (1 << 4)
+#define INDEX_CREATE_PARTITIONED (1 << 5)
+#define INDEX_CREATE_INVALID (1 << 6)
extern Oid index_create(Relation heapRelation,
const char *indexRelationName,
Oid indexRelationId,
+ Oid parentIndexRelid,
Oid relFileNode,
IndexInfo *indexInfo,
List *indexColNames,
@@ -84,6 +87,8 @@ extern void index_drop(Oid indexId, bool concurrent);
extern IndexInfo *BuildIndexInfo(Relation index);
+extern bool CompareIndexInfo(IndexInfo *info1, IndexInfo *info2, AttrNumber *attmap);
+
extern void BuildSpeculativeIndexInfo(Relation index, IndexInfo *ii);
extern void FormIndexDatum(IndexInfo *indexInfo,
@@ -134,4 +139,6 @@ extern bool ReindexIsProcessingHeap(Oid heapOid);
extern bool ReindexIsProcessingIndex(Oid indexOid);
extern Oid IndexGetRelation(Oid indexId, bool missing_ok);
+extern void IndexSetParentIndex(Relation idx, Oid parentOid);
+
#endif /* INDEX_H */
diff --git a/src/include/catalog/indexing.h b/src/include/catalog/indexing.h
index ef8493674c..185ba379d4 100644
--- a/src/include/catalog/indexing.h
+++ b/src/include/catalog/indexing.h
@@ -162,6 +162,8 @@ DECLARE_INDEX(pg_index_indrelid_index, 2678, on pg_index using btree(indrelid oi
#define IndexIndrelidIndexId 2678
DECLARE_UNIQUE_INDEX(pg_index_indexrelid_index, 2679, on pg_index using btree(indexrelid oid_ops));
#define IndexRelidIndexId 2679
+DECLARE_INDEX(pg_index_parentidx_relid_index, 2579, on pg_index using btree(indparentidx oid_ops, indrelid oid_ops));
+#define IndexParentidxRelidIndexId 2579
DECLARE_UNIQUE_INDEX(pg_inherits_relid_seqno_index, 2680, on pg_inherits using btree(inhrelid oid_ops, inhseqno int4_ops));
#define InheritsRelidSeqnoIndexId 2680
diff --git a/src/include/catalog/pg_class.h b/src/include/catalog/pg_class.h
index b256657bda..dd8e7ea2b5 100644
--- a/src/include/catalog/pg_class.h
+++ b/src/include/catalog/pg_class.h
@@ -166,6 +166,7 @@ DESCR("");
#define RELKIND_COMPOSITE_TYPE 'c' /* composite type */
#define RELKIND_FOREIGN_TABLE 'f' /* foreign table */
#define RELKIND_PARTITIONED_TABLE 'p' /* partitioned table */
+#define RELKIND_PARTITIONED_INDEX 'I' /* partitioned index */
#define RELPERSISTENCE_PERMANENT 'p' /* regular table */
#define RELPERSISTENCE_UNLOGGED 'u' /* unlogged permanent table */
diff --git a/src/include/catalog/pg_index.h b/src/include/catalog/pg_index.h
index 8505c3be5f..e7afb0b921 100644
--- a/src/include/catalog/pg_index.h
+++ b/src/include/catalog/pg_index.h
@@ -32,6 +32,7 @@ CATALOG(pg_index,2610) BKI_WITHOUT_OIDS BKI_SCHEMA_MACRO
{
Oid indexrelid; /* OID of the index */
Oid indrelid; /* OID of the relation it indexes */
+ Oid indparentidx; /* OID of parent index, 0 if not partitioned */
int16 indnatts; /* number of columns in index */
bool indisunique; /* is this a unique index? */
bool indisprimary; /* is this index for primary key? */
@@ -70,26 +71,27 @@ typedef FormData_pg_index *Form_pg_index;
* compiler constants for pg_index
* ----------------
*/
-#define Natts_pg_index 19
+#define Natts_pg_index 20
#define Anum_pg_index_indexrelid 1
#define Anum_pg_index_indrelid 2
-#define Anum_pg_index_indnatts 3
-#define Anum_pg_index_indisunique 4
-#define Anum_pg_index_indisprimary 5
-#define Anum_pg_index_indisexclusion 6
-#define Anum_pg_index_indimmediate 7
-#define Anum_pg_index_indisclustered 8
-#define Anum_pg_index_indisvalid 9
-#define Anum_pg_index_indcheckxmin 10
-#define Anum_pg_index_indisready 11
-#define Anum_pg_index_indislive 12
-#define Anum_pg_index_indisreplident 13
-#define Anum_pg_index_indkey 14
-#define Anum_pg_index_indcollation 15
-#define Anum_pg_index_indclass 16
-#define Anum_pg_index_indoption 17
-#define Anum_pg_index_indexprs 18
-#define Anum_pg_index_indpred 19
+#define Anum_pg_index_indparentidx 3
+#define Anum_pg_index_indnatts 4
+#define Anum_pg_index_indisunique 5
+#define Anum_pg_index_indisprimary 6
+#define Anum_pg_index_indisexclusion 7
+#define Anum_pg_index_indimmediate 8
+#define Anum_pg_index_indisclustered 9
+#define Anum_pg_index_indisvalid 10
+#define Anum_pg_index_indcheckxmin 11
+#define Anum_pg_index_indisready 12
+#define Anum_pg_index_indislive 13
+#define Anum_pg_index_indisreplident 14
+#define Anum_pg_index_indkey 15
+#define Anum_pg_index_indcollation 16
+#define Anum_pg_index_indclass 17
+#define Anum_pg_index_indoption 18
+#define Anum_pg_index_indexprs 19
+#define Anum_pg_index_indpred 20
/*
* Index AMs that support ordered scans must support these two indoption
diff --git a/src/include/commands/defrem.h b/src/include/commands/defrem.h
index 52cbf61ccb..35f50c1175 100644
--- a/src/include/commands/defrem.h
+++ b/src/include/commands/defrem.h
@@ -25,12 +25,13 @@ extern void RemoveObjects(DropStmt *stmt);
extern ObjectAddress DefineIndex(Oid relationId,
IndexStmt *stmt,
Oid indexRelationId,
+ Oid parentIndexId,
bool is_alter_table,
bool check_rights,
bool check_not_in_use,
bool skip_build,
bool quiet);
-extern Oid ReindexIndex(RangeVar *indexRelation, int options);
+extern void ReindexIndex(RangeVar *indexRelation, int options);
extern Oid ReindexTable(RangeVar *relation, int options);
extern void ReindexMultipleTables(const char *objectName, ReindexObjectType objectKind,
int options);
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 94351eafad..a03742bc5d 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -159,6 +159,7 @@ typedef struct IndexInfo
bool ii_ReadyForInserts;
bool ii_Concurrent;
bool ii_BrokenHotChain;
+ Oid ii_Am;
void *ii_AmCache;
MemoryContext ii_Context;
} IndexInfo;
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 2eaa6b2774..990049247f 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -839,7 +839,7 @@ typedef struct PartitionRangeDatum
} PartitionRangeDatum;
/*
- * PartitionCmd - info for ALTER TABLE ATTACH/DETACH PARTITION commands
+ * PartitionCmd - info for ALTER TABLE/INDEX ATTACH/DETACH PARTITION commands
*/
typedef struct PartitionCmd
{
@@ -2702,6 +2702,10 @@ typedef struct FetchStmt
* index, just a UNIQUE/PKEY constraint using an existing index. isconstraint
* must always be true in this case, and the fields describing the index
* properties are empty.
+ *
+ * The relation to build the index on can be represented either by name
+ * (in which case the RangeVar indicates whether to recurse or not) or by OID
+ * (in which case the command is always recursive).
* ----------------------
*/
typedef struct IndexStmt
@@ -2709,6 +2713,7 @@ typedef struct IndexStmt
NodeTag type;
char *idxname; /* name of new index, or NULL for default */
RangeVar *relation; /* relation to build index on */
+ Oid relationId; /* OID of relation to build index on */
char *accessMethod; /* name of access method (eg. btree) */
char *tableSpace; /* tablespace, or NULL for default */
List *indexParams; /* columns to index: a list of IndexElem */
diff --git a/src/include/parser/parse_utilcmd.h b/src/include/parser/parse_utilcmd.h
index e749432ef0..599f0e8e29 100644
--- a/src/include/parser/parse_utilcmd.h
+++ b/src/include/parser/parse_utilcmd.h
@@ -27,5 +27,8 @@ extern void transformRuleStmt(RuleStmt *stmt, const char *queryString,
extern List *transformCreateSchemaStmt(CreateSchemaStmt *stmt);
extern PartitionBoundSpec *transformPartitionBound(ParseState *pstate, Relation parent,
PartitionBoundSpec *spec);
+extern IndexStmt *generateClonedIndexStmt(RangeVar *heapRel, Oid heapOid,
+ Relation source_idx,
+ const AttrNumber *attmap, int attmap_length);
#endif /* PARSE_UTILCMD_H */
diff --git a/src/test/regress/expected/alter_table.out b/src/test/regress/expected/alter_table.out
index 11f0baa11b..517fb080bd 100644
--- a/src/test/regress/expected/alter_table.out
+++ b/src/test/regress/expected/alter_table.out
@@ -1965,6 +1965,67 @@ create table tab1 (a int, b text);
create table tab2 (x int, y tab1);
alter table tab1 alter column b type varchar; -- fails
ERROR: cannot alter table "tab1" because column "tab2.y" uses its row type
+-- Alter column type that's part of a partitioned index
+create table at_partitioned (a int, b text) partition by range (a);
+create table at_part_1 partition of at_partitioned for values from (0) to (1000);
+insert into at_partitioned values (512, '0.123');
+create table at_part_2 (b text, a int);
+insert into at_part_2 values ('1.234', 1024);
+create index on at_partitioned (b);
+create index on at_partitioned (a);
+\d at_part_1
+ Table "public.at_part_1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | text | | |
+Partition of: at_partitioned FOR VALUES FROM (0) TO (1000)
+Indexes:
+ "at_part_1_a_idx" btree (a)
+ "at_part_1_b_idx" btree (b)
+
+\d at_part_2
+ Table "public.at_part_2"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ b | text | | |
+ a | integer | | |
+
+alter table at_partitioned attach partition at_part_2 for values from (1000) to (2000);
+\d at_part_2
+ Table "public.at_part_2"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ b | text | | |
+ a | integer | | |
+Partition of: at_partitioned FOR VALUES FROM (1000) TO (2000)
+Indexes:
+ "at_part_2_a_idx" btree (a)
+ "at_part_2_b_idx" btree (b)
+
+alter table at_partitioned alter column b type numeric using b::numeric;
+\d at_part_1
+ Table "public.at_part_1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | numeric | | |
+Partition of: at_partitioned FOR VALUES FROM (0) TO (1000)
+Indexes:
+ "at_part_1_a_idx" btree (a)
+ "at_part_1_b_idx" btree (b)
+
+\d at_part_2
+ Table "public.at_part_2"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ b | numeric | | |
+ a | integer | | |
+Partition of: at_partitioned FOR VALUES FROM (1000) TO (2000)
+Indexes:
+ "at_part_2_a_idx" btree (a)
+ "at_part_2_b_idx" btree (b)
+
-- disallow recursive containment of row types
create temp table recur1 (f1 int);
alter table recur1 add column f2 recur1; -- fails
@@ -3276,7 +3337,7 @@ CREATE TABLE unparted (
);
CREATE TABLE fail_part (like unparted);
ALTER TABLE unparted ATTACH PARTITION fail_part FOR VALUES IN ('a');
-ERROR: "unparted" is not partitioned
+ERROR: table "unparted" is not partitioned
DROP TABLE unparted, fail_part;
-- check that partition bound is compatible
CREATE TABLE list_parted (
@@ -3656,7 +3717,7 @@ DROP TABLE fail_part;
-- check that the table is partitioned at all
CREATE TABLE regular_table (a int);
ALTER TABLE regular_table DETACH PARTITION any_name;
-ERROR: "regular_table" is not partitioned
+ERROR: table "regular_table" is not partitioned
DROP TABLE regular_table;
-- check that the partition being detached exists at all
ALTER TABLE list_parted2 DETACH PARTITION part_4;
diff --git a/src/test/regress/expected/indexing.out b/src/test/regress/expected/indexing.out
new file mode 100644
index 0000000000..7d463cde15
--- /dev/null
+++ b/src/test/regress/expected/indexing.out
@@ -0,0 +1,471 @@
+-- Creating an index on a partitioned table makes the partitions
+-- automatically get the index
+create table idxpart (a int, b int, c text) partition by range (a);
+create table idxpart1 partition of idxpart for values from (0) to (10);
+create table idxpart2 partition of idxpart for values from (10) to (100)
+ partition by range (b);
+create table idxpart21 partition of idxpart2 for values from (0) to (100);
+create index on idxpart (a);
+select relname, relkind, indparentidx::regclass
+ from pg_class left join pg_index on (indexrelid = oid)
+ where relname like 'idxpart%' order by relname;
+ relname | relkind | indparentidx
+-----------------+---------+----------------
+ idxpart | p |
+ idxpart1 | r |
+ idxpart1_a_idx | i | idxpart_a_idx
+ idxpart2 | p |
+ idxpart21 | r |
+ idxpart21_a_idx | i | idxpart2_a_idx
+ idxpart2_a_idx | I | idxpart_a_idx
+ idxpart_a_idx | I | -
+(8 rows)
+
+drop table idxpart;
+-- Some unsupported features
+create table idxpart (a int, b int, c text) partition by range (a);
+create table idxpart1 partition of idxpart for values from (0) to (10);
+create unique index on idxpart (a);
+ERROR: cannot create unique index on partitioned table "idxpart"
+create index concurrently on idxpart (a);
+ERROR: cannot create index on partitioned table "idxpart" concurrently
+drop table idxpart;
+-- If a table without index is attached as partition to a table with
+-- an index, the index is automatically created
+create table idxpart (a int, b int, c text) partition by range (a);
+create index idxparti on idxpart (a);
+create index idxparti2 on idxpart (b, c);
+create table idxpart1 (like idxpart);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | integer | | |
+ c | text | | |
+
+alter table idxpart attach partition idxpart1 for values from (0) to (10);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | integer | | |
+ c | text | | |
+Partition of: idxpart FOR VALUES FROM (0) TO (10)
+Indexes:
+ "idxpart1_a_idx" btree (a)
+ "idxpart1_b_c_idx" btree (b, c)
+
+drop table idxpart;
+-- If a partition already has an index, don't create a duplicative one
+create table idxpart (a int, b int) partition by range (a, b);
+create table idxpart1 partition of idxpart for values from (0, 0) to (10, 10);
+create index on idxpart1 (a, b);
+create index on idxpart (a, b);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | integer | | |
+Partition of: idxpart FOR VALUES FROM (0, 0) TO (10, 10)
+Indexes:
+ "idxpart1_a_b_idx" btree (a, b)
+
+select indexrelid::regclass, indrelid::regclass, indparentidx::regclass
+ from pg_index where indexrelid::regclass::text like 'idxpart%'
+ order by indexrelid::regclass;
+ indexrelid | indrelid | indparentidx
+------------------+----------+-----------------
+ idxpart1_a_b_idx | idxpart1 | idxpart_a_b_idx
+ idxpart_a_b_idx | idxpart | -
+(2 rows)
+
+drop table idxpart;
+-- DROP behavior for partitioned indexes
+create table idxpart (a int) partition by range (a);
+create index on idxpart (a);
+create table idxpart1 partition of idxpart for values from (0) to (10);
+drop index idxpart1_a_idx; -- no way
+ERROR: cannot drop index idxpart1_a_idx because index idxpart_a_idx requires it
+HINT: You can drop index idxpart_a_idx instead.
+drop index idxpart_a_idx; -- both indexes go away
+\di idxpart*
+ List of relations
+ Schema | Name | Type | Owner | Table
+--------+------+------+-------+-------
+(0 rows)
+
+\d idxpart*
+ Table "public.idxpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Number of partitions: 1 (Use \d+ to list them.)
+
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition of: idxpart FOR VALUES FROM (0) TO (10)
+
+create index on idxpart (a);
+drop table idxpart1; -- the index on partition goes away too
+\di idxpart*
+ List of relations
+ Schema | Name | Type | Owner | Table
+--------+---------------+-------+----------+---------
+ public | idxpart_a_idx | index | alvherre | idxpart
+(1 row)
+
+\d idxpart*
+ Table "public.idxpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "idxpart_a_idx" btree (a)
+Number of partitions: 0
+
+ Index "public.idxpart_a_idx"
+ Column | Type | Definition
+--------+---------+------------
+ a | integer | a
+btree, for table "public.idxpart"
+
+drop table idxpart;
+-- ALTER INDEX .. ATTACH, error cases
+create table idxpart (a int, b int) partition by range (a, b);
+create table idxpart1 partition of idxpart for values from (0, 0) to (10, 10);
+create index idxpart_a_b_idx on only idxpart (a, b);
+create index idxpart1_a_b_idx on idxpart1 (a, b);
+create index idxpart1_tst1 on idxpart1 (b, a);
+create index idxpart1_tst2 on idxpart1 using hash (a);
+create index idxpart1_tst3 on idxpart1 (a, b) where a > 10;
+alter index idxpart attach partition idxpart1;
+ERROR: "idxpart" is not an index
+alter index idxpart_a_b_idx attach partition idxpart1;
+ERROR: "idxpart1" is not an index
+alter index idxpart_a_b_idx attach partition idxpart_a_b_idx;
+ERROR: cannot attach index "idxpart_a_b_idx" as a partition of index "idxpart_a_b_idx"
+DETAIL: Index "idxpart_a_b_idx" is not on a partition of table "idxpart".
+alter index idxpart_a_b_idx attach partition idxpart1_b_idx;
+ERROR: relation "idxpart1_b_idx" does not exist
+alter index idxpart_a_b_idx attach partition idxpart1_tst1;
+ERROR: cannot attach index "idxpart1_tst1" as a partition of index "idxpart_a_b_idx"
+DETAIL: The index definitions do not match.
+alter index idxpart_a_b_idx attach partition idxpart1_tst2;
+ERROR: cannot attach index "idxpart1_tst2" as a partition of index "idxpart_a_b_idx"
+DETAIL: The index definitions do not match.
+alter index idxpart_a_b_idx attach partition idxpart1_tst3;
+ERROR: cannot attach index "idxpart1_tst3" as a partition of index "idxpart_a_b_idx"
+DETAIL: The index definitions do not match.
+-- OK
+alter index idxpart_a_b_idx attach partition idxpart1_a_b_idx;
+alter index idxpart_a_b_idx attach partition idxpart1_a_b_idx; -- quiet
+-- reject dupe
+create index idxpart1_2_a_b on idxpart1 (a, b);
+alter index idxpart_a_b_idx attach partition idxpart1_2_a_b;
+ERROR: cannot attach index "idxpart1_2_a_b" as a partition of index "idxpart_a_b_idx"
+DETAIL: Another index is already attached for partition "idxpart1".
+drop table idxpart;
+-- make sure everything's gone
+select indexrelid::regclass, indrelid::regclass, indparentidx::regclass
+ from pg_index where indexrelid::regclass::text like 'idxpart%';
+ indexrelid | indrelid | indparentidx
+------------+----------+--------------
+(0 rows)
+
+-- Don't auto-attach incompatible indexes
+create table idxpart (a int, b int) partition by range (a);
+create table idxpart1 (a int, b int);
+create index on idxpart1 using hash (a);
+create index on idxpart1 (a) where b > 1;
+create index on idxpart1 ((a + 0));
+create index on idxpart1 (a, a);
+create index on idxpart (a);
+alter table idxpart attach partition idxpart1 for values from (0) to (1000);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | integer | | |
+Partition of: idxpart FOR VALUES FROM (0) TO (1000)
+Indexes:
+ "idxpart1_a_a1_idx" btree (a, a)
+ "idxpart1_a_idx" hash (a)
+ "idxpart1_a_idx1" btree (a) WHERE b > 1
+ "idxpart1_a_idx2" btree (a)
+ "idxpart1_expr_idx" btree ((a + 0))
+
+drop table idxpart;
+-- If CREATE INDEX ONLY, don't create indexes on partitions; and existing
+-- indexes on partitions don't change parent. ALTER INDEX ATTACH can change
+-- the parent after the fact.
+create table idxpart (a int) partition by range (a);
+create table idxpart1 partition of idxpart for values from (0) to (100);
+create table idxpart2 partition of idxpart for values from (100) to (1000)
+ partition by range (a);
+create table idxpart21 partition of idxpart2 for values from (100) to (200);
+create table idxpart22 partition of idxpart2 for values from (200) to (300);
+create index on idxpart22 (a);
+create index on only idxpart2 (a);
+create index on idxpart (a);
+-- Here we expect that idxpart1 and idxpart2 have a new index, but idxpart21
+-- does not; also, idxpart22 is not attached.
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition of: idxpart FOR VALUES FROM (0) TO (100)
+Indexes:
+ "idxpart1_a_idx" btree (a)
+
+\d idxpart2
+ Table "public.idxpart2"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition of: idxpart FOR VALUES FROM (100) TO (1000)
+Partition key: RANGE (a)
+Indexes:
+ "idxpart2_a_idx" btree (a) INVALID
+Number of partitions: 2 (Use \d+ to list them.)
+
+\d idxpart21
+ Table "public.idxpart21"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition of: idxpart2 FOR VALUES FROM (100) TO (200)
+
+select indexrelid::regclass, indrelid::regclass, indparentidx::regclass
+ from pg_index where indexrelid::regclass::text like 'idxpart%'
+ order by indrelid::regclass;
+ indexrelid | indrelid | indparentidx
+-----------------+-----------+---------------
+ idxpart_a_idx | idxpart | -
+ idxpart1_a_idx | idxpart1 | idxpart_a_idx
+ idxpart2_a_idx | idxpart2 | idxpart_a_idx
+ idxpart22_a_idx | idxpart22 | -
+(4 rows)
+
+alter index idxpart2_a_idx attach partition idxpart22_a_idx;
+select indexrelid::regclass, indrelid::regclass, indparentidx::regclass
+ from pg_index where indexrelid::regclass::text like 'idxpart%'
+ order by indrelid::regclass;
+ indexrelid | indrelid | indparentidx
+-----------------+-----------+----------------
+ idxpart_a_idx | idxpart | -
+ idxpart1_a_idx | idxpart1 | idxpart_a_idx
+ idxpart2_a_idx | idxpart2 | idxpart_a_idx
+ idxpart22_a_idx | idxpart22 | idxpart2_a_idx
+(4 rows)
+
+-- attaching idxpart22 is not enough to set idxpart22_a_idx valid ...
+alter index idxpart2_a_idx attach partition idxpart22_a_idx;
+\d idxpart2
+ Table "public.idxpart2"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition of: idxpart FOR VALUES FROM (100) TO (1000)
+Partition key: RANGE (a)
+Indexes:
+ "idxpart2_a_idx" btree (a) INVALID
+Number of partitions: 2 (Use \d+ to list them.)
+
+-- ... but this one is.
+create index on idxpart21 (a);
+alter index idxpart2_a_idx attach partition idxpart21_a_idx;
+\d idxpart2
+ Table "public.idxpart2"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition of: idxpart FOR VALUES FROM (100) TO (1000)
+Partition key: RANGE (a)
+Indexes:
+ "idxpart2_a_idx" btree (a)
+Number of partitions: 2 (Use \d+ to list them.)
+
+drop table idxpart;
+-- When a table is attached a partition and it already has an index, a
+-- duplicate index should not get created, but rather the index becomes
+-- attached to the parent's index.
+create table idxpart (a int, b int, c text) partition by range (a);
+create index idxparti on idxpart (a);
+create index idxparti2 on idxpart (b, c);
+create table idxpart1 (like idxpart including indexes);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | integer | | |
+ c | text | | |
+Indexes:
+ "idxpart1_a_idx" btree (a)
+ "idxpart1_b_c_idx" btree (b, c)
+
+select relname, relkind, indparentidx::regclass
+ from pg_class left join pg_index on (indexrelid = oid)
+ where relname like 'idxpart%' order by relname;
+ relname | relkind | indparentidx
+------------------+---------+--------------
+ idxpart | p |
+ idxpart1 | r |
+ idxpart1_a_idx | i | -
+ idxpart1_b_c_idx | i | -
+ idxparti | I | -
+ idxparti2 | I | -
+(6 rows)
+
+alter table idxpart attach partition idxpart1 for values from (0) to (10);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | integer | | |
+ c | text | | |
+Partition of: idxpart FOR VALUES FROM (0) TO (10)
+Indexes:
+ "idxpart1_a_idx" btree (a)
+ "idxpart1_b_c_idx" btree (b, c)
+
+select relname, relkind, indparentidx::regclass
+ from pg_class left join pg_index on (indexrelid = oid)
+ where relname like 'idxpart%' order by relname;
+ relname | relkind | indparentidx
+------------------+---------+--------------
+ idxpart | p |
+ idxpart1 | r |
+ idxpart1_a_idx | i | idxparti
+ idxpart1_b_c_idx | i | idxparti2
+ idxparti | I | -
+ idxparti2 | I | -
+(6 rows)
+
+drop table idxpart;
+-- Make sure the partition columns are mapped correctly
+create table idxpart (a int, b int, c text) partition by range (a);
+create index idxparti on idxpart (a);
+create index idxparti2 on idxpart (c, b);
+create table idxpart1 (c text, a int, b int);
+alter table idxpart attach partition idxpart1 for values from (0) to (10);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ c | text | | |
+ a | integer | | |
+ b | integer | | |
+Partition of: idxpart FOR VALUES FROM (0) TO (10)
+Indexes:
+ "idxpart1_a_idx" btree (a)
+ "idxpart1_c_b_idx" btree (c, b)
+
+drop table idxpart;
+-- Column number mapping: dropped columns in the partition
+create table idxpart1 (drop_1 int, drop_2 int, col_keep int, drop_3 int);
+alter table idxpart1 drop column drop_1;
+alter table idxpart1 drop column drop_2;
+alter table idxpart1 drop column drop_3;
+create index on idxpart1 (col_keep);
+create table idxpart (col_keep int) partition by range (col_keep);
+create index on idxpart (col_keep);
+alter table idxpart attach partition idxpart1 for values from (0) to (1000);
+\d idxpart
+ Table "public.idxpart"
+ Column | Type | Collation | Nullable | Default
+----------+---------+-----------+----------+---------
+ col_keep | integer | | |
+Partition key: RANGE (col_keep)
+Indexes:
+ "idxpart_col_keep_idx" btree (col_keep)
+Number of partitions: 1 (Use \d+ to list them.)
+
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+----------+---------+-----------+----------+---------
+ col_keep | integer | | |
+Partition of: idxpart FOR VALUES FROM (0) TO (1000)
+Indexes:
+ "idxpart1_col_keep_idx" btree (col_keep)
+
+select attrelid::regclass, attname, attnum from pg_attribute
+ where attrelid::regclass::text like 'idxpart%' and attnum > 0
+ order by attrelid::regclass, attnum;
+ attrelid | attname | attnum
+-----------------------+------------------------------+--------
+ idxpart1 | ........pg.dropped.1........ | 1
+ idxpart1 | ........pg.dropped.2........ | 2
+ idxpart1 | col_keep | 3
+ idxpart1 | ........pg.dropped.4........ | 4
+ idxpart1_col_keep_idx | col_keep | 1
+ idxpart | col_keep | 1
+ idxpart_col_keep_idx | col_keep | 1
+(7 rows)
+
+drop table idxpart;
+-- Column number mapping: dropped columns in the parent table
+create table idxpart(drop_1 int, drop_2 int, col_keep int, drop_3 int) partition by range (col_keep);
+alter table idxpart drop column drop_1;
+alter table idxpart drop column drop_2;
+alter table idxpart drop column drop_3;
+create table idxpart1 (col_keep int);
+create index on idxpart1 (col_keep);
+create index on idxpart (col_keep);
+alter table idxpart attach partition idxpart1 for values from (0) to (1000);
+\d idxpart
+ Table "public.idxpart"
+ Column | Type | Collation | Nullable | Default
+----------+---------+-----------+----------+---------
+ col_keep | integer | | |
+Partition key: RANGE (col_keep)
+Indexes:
+ "idxpart_col_keep_idx" btree (col_keep)
+Number of partitions: 1 (Use \d+ to list them.)
+
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+----------+---------+-----------+----------+---------
+ col_keep | integer | | |
+Partition of: idxpart FOR VALUES FROM (0) TO (1000)
+Indexes:
+ "idxpart1_col_keep_idx" btree (col_keep)
+
+select attrelid::regclass, attname, attnum from pg_attribute
+ where attrelid::regclass::text like 'idxpart%' and attnum > 0
+ order by attrelid::regclass, attnum;
+ attrelid | attname | attnum
+-----------------------+------------------------------+--------
+ idxpart | ........pg.dropped.1........ | 1
+ idxpart | ........pg.dropped.2........ | 2
+ idxpart | col_keep | 3
+ idxpart | ........pg.dropped.4........ | 4
+ idxpart1 | col_keep | 1
+ idxpart1_col_keep_idx | col_keep | 1
+ idxpart_col_keep_idx | col_keep | 1
+(7 rows)
+
+drop table idxpart;
+-- intentionally leave some objects around
+create table idxpart (a int) partition by range (a);
+create table idxpart1 partition of idxpart for values from (0) to (100);
+create table idxpart2 partition of idxpart for values from (100) to (1000)
+ partition by range (a);
+create table idxpart21 partition of idxpart2 for values from (100) to (200);
+create table idxpart22 partition of idxpart2 for values from (200) to (300);
+create index on idxpart22 (a);
+create index on only idxpart2 (a);
+alter index idxpart2_a_idx attach partition idxpart22_a_idx;
+create index on idxpart (a);
diff --git a/src/test/regress/parallel_schedule b/src/test/regress/parallel_schedule
index e224977791..ad9434fb87 100644
--- a/src/test/regress/parallel_schedule
+++ b/src/test/regress/parallel_schedule
@@ -116,7 +116,7 @@ test: plancache limit plpgsql copy2 temp domain rangefuncs prepare without_oid c
# ----------
# Another group of parallel tests
# ----------
-test: identity partition_join partition_prune reloptions hash_part
+test: identity partition_join partition_prune reloptions hash_part indexing
# event triggers cannot run concurrently with any test that runs DDL
test: event_trigger
diff --git a/src/test/regress/serial_schedule b/src/test/regress/serial_schedule
index 9fc5f1a268..27cd49845e 100644
--- a/src/test/regress/serial_schedule
+++ b/src/test/regress/serial_schedule
@@ -184,5 +184,6 @@ test: partition_join
test: partition_prune
test: reloptions
test: hash_part
+test: indexing
test: event_trigger
test: stats
diff --git a/src/test/regress/sql/alter_table.sql b/src/test/regress/sql/alter_table.sql
index 02a33ca7c4..af25ee9e77 100644
--- a/src/test/regress/sql/alter_table.sql
+++ b/src/test/regress/sql/alter_table.sql
@@ -1330,6 +1330,22 @@ create table tab1 (a int, b text);
create table tab2 (x int, y tab1);
alter table tab1 alter column b type varchar; -- fails
+-- Alter column type that's part of a partitioned index
+create table at_partitioned (a int, b text) partition by range (a);
+create table at_part_1 partition of at_partitioned for values from (0) to (1000);
+insert into at_partitioned values (512, '0.123');
+create table at_part_2 (b text, a int);
+insert into at_part_2 values ('1.234', 1024);
+create index on at_partitioned (b);
+create index on at_partitioned (a);
+\d at_part_1
+\d at_part_2
+alter table at_partitioned attach partition at_part_2 for values from (1000) to (2000);
+\d at_part_2
+alter table at_partitioned alter column b type numeric using b::numeric;
+\d at_part_1
+\d at_part_2
+
-- disallow recursive containment of row types
create temp table recur1 (f1 int);
alter table recur1 add column f2 recur1; -- fails
diff --git a/src/test/regress/sql/indexing.sql b/src/test/regress/sql/indexing.sql
new file mode 100644
index 0000000000..41d50ffaa4
--- /dev/null
+++ b/src/test/regress/sql/indexing.sql
@@ -0,0 +1,199 @@
+-- Creating an index on a partitioned table makes the partitions
+-- automatically get the index
+create table idxpart (a int, b int, c text) partition by range (a);
+create table idxpart1 partition of idxpart for values from (0) to (10);
+create table idxpart2 partition of idxpart for values from (10) to (100)
+ partition by range (b);
+create table idxpart21 partition of idxpart2 for values from (0) to (100);
+create index on idxpart (a);
+select relname, relkind, indparentidx::regclass
+ from pg_class left join pg_index on (indexrelid = oid)
+ where relname like 'idxpart%' order by relname;
+drop table idxpart;
+
+-- Some unsupported features
+create table idxpart (a int, b int, c text) partition by range (a);
+create table idxpart1 partition of idxpart for values from (0) to (10);
+create unique index on idxpart (a);
+create index concurrently on idxpart (a);
+drop table idxpart;
+
+-- If a table without index is attached as partition to a table with
+-- an index, the index is automatically created
+create table idxpart (a int, b int, c text) partition by range (a);
+create index idxparti on idxpart (a);
+create index idxparti2 on idxpart (b, c);
+create table idxpart1 (like idxpart);
+\d idxpart1
+alter table idxpart attach partition idxpart1 for values from (0) to (10);
+\d idxpart1
+drop table idxpart;
+
+-- If a partition already has an index, don't create a duplicative one
+create table idxpart (a int, b int) partition by range (a, b);
+create table idxpart1 partition of idxpart for values from (0, 0) to (10, 10);
+create index on idxpart1 (a, b);
+create index on idxpart (a, b);
+\d idxpart1
+select indexrelid::regclass, indrelid::regclass, indparentidx::regclass
+ from pg_index where indexrelid::regclass::text like 'idxpart%'
+ order by indexrelid::regclass;
+drop table idxpart;
+
+-- DROP behavior for partitioned indexes
+create table idxpart (a int) partition by range (a);
+create index on idxpart (a);
+create table idxpart1 partition of idxpart for values from (0) to (10);
+drop index idxpart1_a_idx; -- no way
+drop index idxpart_a_idx; -- both indexes go away
+\di idxpart*
+\d idxpart*
+create index on idxpart (a);
+drop table idxpart1; -- the index on partition goes away too
+\di idxpart*
+\d idxpart*
+drop table idxpart;
+
+-- ALTER INDEX .. ATTACH, error cases
+create table idxpart (a int, b int) partition by range (a, b);
+create table idxpart1 partition of idxpart for values from (0, 0) to (10, 10);
+create index idxpart_a_b_idx on only idxpart (a, b);
+create index idxpart1_a_b_idx on idxpart1 (a, b);
+create index idxpart1_tst1 on idxpart1 (b, a);
+create index idxpart1_tst2 on idxpart1 using hash (a);
+create index idxpart1_tst3 on idxpart1 (a, b) where a > 10;
+
+alter index idxpart attach partition idxpart1;
+alter index idxpart_a_b_idx attach partition idxpart1;
+alter index idxpart_a_b_idx attach partition idxpart_a_b_idx;
+alter index idxpart_a_b_idx attach partition idxpart1_b_idx;
+alter index idxpart_a_b_idx attach partition idxpart1_tst1;
+alter index idxpart_a_b_idx attach partition idxpart1_tst2;
+alter index idxpart_a_b_idx attach partition idxpart1_tst3;
+-- OK
+alter index idxpart_a_b_idx attach partition idxpart1_a_b_idx;
+alter index idxpart_a_b_idx attach partition idxpart1_a_b_idx; -- quiet
+
+-- reject dupe
+create index idxpart1_2_a_b on idxpart1 (a, b);
+alter index idxpart_a_b_idx attach partition idxpart1_2_a_b;
+drop table idxpart;
+-- make sure everything's gone
+select indexrelid::regclass, indrelid::regclass, indparentidx::regclass
+ from pg_index where indexrelid::regclass::text like 'idxpart%';
+
+-- Don't auto-attach incompatible indexes
+create table idxpart (a int, b int) partition by range (a);
+create table idxpart1 (a int, b int);
+create index on idxpart1 using hash (a);
+create index on idxpart1 (a) where b > 1;
+create index on idxpart1 ((a + 0));
+create index on idxpart1 (a, a);
+create index on idxpart (a);
+alter table idxpart attach partition idxpart1 for values from (0) to (1000);
+\d idxpart1
+drop table idxpart;
+
+-- If CREATE INDEX ONLY, don't create indexes on partitions; and existing
+-- indexes on partitions don't change parent. ALTER INDEX ATTACH can change
+-- the parent after the fact.
+create table idxpart (a int) partition by range (a);
+create table idxpart1 partition of idxpart for values from (0) to (100);
+create table idxpart2 partition of idxpart for values from (100) to (1000)
+ partition by range (a);
+create table idxpart21 partition of idxpart2 for values from (100) to (200);
+create table idxpart22 partition of idxpart2 for values from (200) to (300);
+create index on idxpart22 (a);
+create index on only idxpart2 (a);
+create index on idxpart (a);
+-- Here we expect that idxpart1 and idxpart2 have a new index, but idxpart21
+-- does not; also, idxpart22 is not attached.
+\d idxpart1
+\d idxpart2
+\d idxpart21
+select indexrelid::regclass, indrelid::regclass, indparentidx::regclass
+ from pg_index where indexrelid::regclass::text like 'idxpart%'
+ order by indrelid::regclass;
+alter index idxpart2_a_idx attach partition idxpart22_a_idx;
+select indexrelid::regclass, indrelid::regclass, indparentidx::regclass
+ from pg_index where indexrelid::regclass::text like 'idxpart%'
+ order by indrelid::regclass;
+-- attaching idxpart22 is not enough to set idxpart22_a_idx valid ...
+alter index idxpart2_a_idx attach partition idxpart22_a_idx;
+\d idxpart2
+-- ... but this one is.
+create index on idxpart21 (a);
+alter index idxpart2_a_idx attach partition idxpart21_a_idx;
+\d idxpart2
+drop table idxpart;
+
+-- When a table is attached a partition and it already has an index, a
+-- duplicate index should not get created, but rather the index becomes
+-- attached to the parent's index.
+create table idxpart (a int, b int, c text) partition by range (a);
+create index idxparti on idxpart (a);
+create index idxparti2 on idxpart (b, c);
+create table idxpart1 (like idxpart including indexes);
+\d idxpart1
+select relname, relkind, indparentidx::regclass
+ from pg_class left join pg_index on (indexrelid = oid)
+ where relname like 'idxpart%' order by relname;
+alter table idxpart attach partition idxpart1 for values from (0) to (10);
+\d idxpart1
+select relname, relkind, indparentidx::regclass
+ from pg_class left join pg_index on (indexrelid = oid)
+ where relname like 'idxpart%' order by relname;
+drop table idxpart;
+
+-- Make sure the partition columns are mapped correctly
+create table idxpart (a int, b int, c text) partition by range (a);
+create index idxparti on idxpart (a);
+create index idxparti2 on idxpart (c, b);
+create table idxpart1 (c text, a int, b int);
+alter table idxpart attach partition idxpart1 for values from (0) to (10);
+\d idxpart1
+drop table idxpart;
+
+-- Column number mapping: dropped columns in the partition
+create table idxpart1 (drop_1 int, drop_2 int, col_keep int, drop_3 int);
+alter table idxpart1 drop column drop_1;
+alter table idxpart1 drop column drop_2;
+alter table idxpart1 drop column drop_3;
+create index on idxpart1 (col_keep);
+create table idxpart (col_keep int) partition by range (col_keep);
+create index on idxpart (col_keep);
+alter table idxpart attach partition idxpart1 for values from (0) to (1000);
+\d idxpart
+\d idxpart1
+select attrelid::regclass, attname, attnum from pg_attribute
+ where attrelid::regclass::text like 'idxpart%' and attnum > 0
+ order by attrelid::regclass, attnum;
+drop table idxpart;
+
+-- Column number mapping: dropped columns in the parent table
+create table idxpart(drop_1 int, drop_2 int, col_keep int, drop_3 int) partition by range (col_keep);
+alter table idxpart drop column drop_1;
+alter table idxpart drop column drop_2;
+alter table idxpart drop column drop_3;
+create table idxpart1 (col_keep int);
+create index on idxpart1 (col_keep);
+create index on idxpart (col_keep);
+alter table idxpart attach partition idxpart1 for values from (0) to (1000);
+\d idxpart
+\d idxpart1
+select attrelid::regclass, attname, attnum from pg_attribute
+ where attrelid::regclass::text like 'idxpart%' and attnum > 0
+ order by attrelid::regclass, attnum;
+drop table idxpart;
+
+-- intentionally leave some objects around
+create table idxpart (a int) partition by range (a);
+create table idxpart1 partition of idxpart for values from (0) to (100);
+create table idxpart2 partition of idxpart for values from (100) to (1000)
+ partition by range (a);
+create table idxpart21 partition of idxpart2 for values from (100) to (200);
+create table idxpart22 partition of idxpart2 for values from (200) to (300);
+create index on idxpart22 (a);
+create index on only idxpart2 (a);
+alter index idxpart2_a_idx attach partition idxpart22_a_idx;
+create index on idxpart (a);
--
2.11.0
v1-0002-allow-indexes-on-partitioned-tables-to-be-unique.patchtext/plain; charset=us-asciiDownload
From ed6d02a4f6196569acf73de5da22b619b694b879 Mon Sep 17 00:00:00 2001
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
Date: Mon, 6 Nov 2017 17:04:55 +0100
Subject: [PATCH v1 2/4] allow indexes on partitioned tables to be unique
---
doc/src/sgml/ref/alter_table.sgml | 9 +-
doc/src/sgml/ref/create_table.sgml | 16 +++-
src/backend/bootstrap/bootparse.y | 2 +
src/backend/catalog/index.c | 28 +++++-
src/backend/catalog/toasting.c | 4 +-
src/backend/commands/indexcmds.c | 80 ++++++++++++++--
src/backend/commands/tablecmds.c | 12 ++-
src/backend/parser/parse_utilcmd.c | 31 +------
src/backend/tcop/utility.c | 1 +
src/include/catalog/index.h | 5 +-
src/include/commands/defrem.h | 1 +
src/include/parser/parse_utilcmd.h | 3 +-
src/test/regress/expected/alter_table.out | 8 --
src/test/regress/expected/create_table.out | 12 ---
src/test/regress/expected/indexing.out | 142 ++++++++++++++++++++++++++++-
src/test/regress/sql/alter_table.sql | 2 -
src/test/regress/sql/create_table.sql | 8 --
src/test/regress/sql/indexing.sql | 73 ++++++++++++++-
18 files changed, 355 insertions(+), 82 deletions(-)
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index 0a2f3e3646..ee6a45c9ad 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -782,8 +782,9 @@ ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
This form attaches an existing table (which might itself be partitioned)
as a partition of the target table. The table can be attached
as a partition for specific values using <literal>FOR VALUES
- </literal> or as a default partition by using <literal>DEFAULT
- </literal>. For each index in the target table, a corresponding
+ </literal> or as a default partition by using
+ <literal>DEFAULT</literal>.
+ For each index in the target table, a corresponding
one will be created in the attached table; or, if an equivalent
index already exists, will be attached to the target table's index,
as if <command>ALTER INDEX ATTACH</command> had been executed.
@@ -798,8 +799,10 @@ ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
as the target table and no more; moreover, the column types must also
match. Also, it must have all the <literal>NOT NULL</literal> and
<literal>CHECK</literal> constraints of the target table. Currently
- <literal>UNIQUE</literal>, <literal>PRIMARY KEY</literal>, and
<literal>FOREIGN KEY</literal> constraints are not considered.
+ <literal>UNIQUE</literal> and <literal>PRIMARY KEY</literal> constraints
+ from the parent table will be created in the partition, if they don't
+ already exist.
If any of the <literal>CHECK</literal> constraints of the table being
attached is marked <literal>NO INHERIT</literal>, the command will fail;
such a constraint must be recreated without the <literal>NO INHERIT</literal>
diff --git a/doc/src/sgml/ref/create_table.sgml b/doc/src/sgml/ref/create_table.sgml
index a0c9a6d257..98ab39473b 100644
--- a/doc/src/sgml/ref/create_table.sgml
+++ b/doc/src/sgml/ref/create_table.sgml
@@ -546,8 +546,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</para>
<para>
- Partitioned tables do not support <literal>UNIQUE</literal>,
- <literal>PRIMARY KEY</literal>, <literal>EXCLUDE</literal>, or
+ Partitioned tables do not support
+ <literal>EXCLUDE</literal>, or
<literal>FOREIGN KEY</literal> constraints; however, you can define
these constraints on individual partitions.
</para>
@@ -786,6 +786,11 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
primary key constraint defined for the table. (Otherwise it
would just be the same constraint listed twice.)
</para>
+
+ <para>
+ When used on partitioned tables, <literal>UNIQUE<literal> constraints
+ must include all the columns of the partition key.
+ <para>
</listitem>
</varlistentry>
@@ -814,6 +819,13 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
about the design of the schema, since a primary key implies that other
tables can rely on this set of columns as a unique identifier for rows.
</para>
+
+ <para>
+ <literal>PRIMARY KEY</literal> constraints share the restrictions that
+ <literal>UNIQUE</literal> constraints have when placed on partitioned
+ tables.
+ <para>
+
</listitem>
</varlistentry>
diff --git a/src/backend/bootstrap/bootparse.y b/src/backend/bootstrap/bootparse.y
index 95835ac1e7..fb14f558ec 100644
--- a/src/backend/bootstrap/bootparse.y
+++ b/src/backend/bootstrap/bootparse.y
@@ -322,6 +322,7 @@ Boot_DeclareIndexStmt:
stmt,
$4,
InvalidOid,
+ InvalidOid,
false,
false,
false,
@@ -367,6 +368,7 @@ Boot_DeclareUniqueIndexStmt:
stmt,
$5,
InvalidOid,
+ InvalidOid,
false,
false,
false,
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index a4787c3ad6..e1e5fd1f45 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -677,6 +677,8 @@ UpdateIndexRelation(Oid indexoid,
* nonzero to specify a preselected OID.
* parentIndexRelid: if creating an index partition, the OID of the
* parent index; otherwise InvalidOid.
+ * parentConstraintId: if creating a constraint on a partition, the OID
+ * of the constraint in the parent; otherwise InvalidOid.
* relFileNode: normally, pass InvalidOid to get new storage. May be
* nonzero to attach an existing valid build.
* indexInfo: same info executor uses to insert into the index
@@ -708,6 +710,7 @@ UpdateIndexRelation(Oid indexoid,
* (only if INDEX_CREATE_ADD_CONSTRAINT is set)
* allow_system_table_mods: allow table to be a system catalog
* is_internal: if true, post creation hook for new index
+ * constraintId: if not NULL, receives OID of created constraint
*
* Returns the OID of the created index.
*/
@@ -716,6 +719,7 @@ index_create(Relation heapRelation,
const char *indexRelationName,
Oid indexRelationId,
Oid parentIndexRelid,
+ Oid parentConstraintId,
Oid relFileNode,
IndexInfo *indexInfo,
List *indexColNames,
@@ -728,7 +732,8 @@ index_create(Relation heapRelation,
bits16 flags,
bits16 constr_flags,
bool allow_system_table_mods,
- bool is_internal)
+ bool is_internal,
+ Oid *constraintId)
{
Oid heapRelationId = RelationGetRelid(heapRelation);
Relation pg_class;
@@ -971,6 +976,7 @@ index_create(Relation heapRelation,
if ((flags & INDEX_CREATE_ADD_CONSTRAINT) != 0)
{
char constraintType;
+ ObjectAddress localaddr;
if (isprimary)
constraintType = CONSTRAINT_PRIMARY;
@@ -984,14 +990,17 @@ index_create(Relation heapRelation,
constraintType = 0; /* keep compiler quiet */
}
- index_constraint_create(heapRelation,
+ localaddr = index_constraint_create(heapRelation,
indexRelationId,
+ parentConstraintId,
indexInfo,
indexRelationName,
constraintType,
constr_flags,
allow_system_table_mods,
is_internal);
+ if (constraintId)
+ *constraintId = localaddr.objectId;
}
else
{
@@ -1163,6 +1172,8 @@ index_create(Relation heapRelation,
*
* heapRelation: table owning the index (must be suitably locked by caller)
* indexRelationId: OID of the index
+ * parentConstraintId: if constraint is on a partition, the OID of the
+ * constraint in the parent.
* indexInfo: same info executor uses to insert into the index
* constraintName: what it say (generally, should match name of index)
* constraintType: one of CONSTRAINT_PRIMARY, CONSTRAINT_UNIQUE, or
@@ -1180,6 +1191,7 @@ index_create(Relation heapRelation,
ObjectAddress
index_constraint_create(Relation heapRelation,
Oid indexRelationId,
+ Oid parentConstraintId,
IndexInfo *indexInfo,
const char *constraintName,
char constraintType,
@@ -1277,6 +1289,18 @@ index_constraint_create(Relation heapRelation,
recordDependencyOn(&myself, &referenced, DEPENDENCY_INTERNAL);
/*
+ * Also, if this is a constraint on a partition, mark it as depending
+ * on the constraint in the parent.
+ */
+ if (OidIsValid(parentConstraintId))
+ {
+ ObjectAddress third;
+
+ ObjectAddressSet(third, ConstraintRelationId, parentConstraintId);
+ recordDependencyOn(&referenced, &third, DEPENDENCY_INTERNAL);
+ }
+
+ /*
* If the constraint is deferrable, create the deferred uniqueness
* checking trigger. (The trigger will be given an internal dependency on
* the constraint by CreateTrigger.)
diff --git a/src/backend/catalog/toasting.c b/src/backend/catalog/toasting.c
index 09582a8d52..94e167d775 100644
--- a/src/backend/catalog/toasting.c
+++ b/src/backend/catalog/toasting.c
@@ -329,13 +329,13 @@ create_toast_table(Relation rel, Oid toastOid, Oid toastIndexOid,
coloptions[1] = 0;
index_create(toast_rel, toast_idxname, toastIndexOid, InvalidOid,
- InvalidOid,
+ InvalidOid, InvalidOid,
indexInfo,
list_make2("chunk_id", "chunk_seq"),
BTREE_AM_OID,
rel->rd_rel->reltablespace,
collationObjectId, classObjectId, coloptions, (Datum) 0,
- INDEX_CREATE_IS_PRIMARY, 0, true, true);
+ INDEX_CREATE_IS_PRIMARY, 0, true, true, NULL);
heap_close(toast_rel, NoLock);
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index e925351056..e5b2c1d1e4 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -299,6 +299,8 @@ CheckIndexCompatible(Oid oldId,
* nonzero to specify a preselected OID for the index.
* 'parentIndexId': the OID of the parent index; InvalidOid if not the child
* of a partitioned index.
+ * 'parentConstraintId': the OID of the parent constraint; InvalidOid if not
+ * the child of a constraint (only used when recursing)
* 'is_alter_table': this is due to an ALTER rather than a CREATE operation.
* 'check_rights': check for CREATE rights in namespace and tablespace. (This
* should be true except when ALTER is deleting/recreating an index.)
@@ -315,6 +317,7 @@ DefineIndex(Oid relationId,
IndexStmt *stmt,
Oid indexRelationId,
Oid parentIndexId,
+ Oid parentConstraintId,
bool is_alter_table,
bool check_rights,
bool check_not_in_use,
@@ -329,6 +332,7 @@ DefineIndex(Oid relationId,
Oid accessMethodId;
Oid namespaceId;
Oid tablespaceId;
+ Oid createdConstraintId;
List *indexColNames;
Relation rel;
Relation indexRelation;
@@ -426,20 +430,11 @@ DefineIndex(Oid relationId,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot create index on partitioned table \"%s\" concurrently",
RelationGetRelationName(rel))));
- if (stmt->unique)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot create unique index on partitioned table \"%s\"",
- RelationGetRelationName(rel))));
if (stmt->excludeOpNames)
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot create exclusion constraints on partitioned table \"%s\"",
RelationGetRelationName(rel))));
- if (stmt->primary || stmt->isconstraint)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot create constraints on partitioned tables")));
}
/*
@@ -637,6 +632,68 @@ DefineIndex(Oid relationId,
index_check_primary_key(rel, indexInfo, is_alter_table);
/*
+ * If this table is partitioned and we're creating a unique index or a
+ * primary key, make sure that the indexed columns are part of the
+ * partition key. Otherwise it would be possible to violate uniqueness by
+ * putting values that ought to be unique in different partitions.
+ *
+ * We could lift this limitation if we had global indexes, but those have
+ * their own problems, so this is a useful feature combination.
+ */
+ if (partitioned && (stmt->unique || stmt->primary))
+ {
+ PartitionKey key = rel->rd_partkey;
+ int i;
+
+ /*
+ * A partitioned table can have unique indexes, as long as all the
+ * columns in the partition key appear in the unique key. A
+ * partition-local index can enforce global uniqueness iff the PK
+ * value completely determines the partition that a row is in.
+ *
+ * Thus, verify that all the columns in the partition key appear
+ * in the unique key definition.
+ */
+ for (i = 0; i < key->partnatts; i++)
+ {
+ bool found = false;
+ int j;
+
+ /*
+ * It may be possible to support UNIQUE constraints when partition
+ * keys are expressions, but is it worth it? Give up for now.
+ */
+ if (key->partattrs[i] == 0)
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ /* XXX reformulate error message? */
+ errmsg("UNIQUE constraints are not supported on partitioned tables using expressions as partition keys")));
+
+ for (j = 0; j < indexInfo->ii_NumIndexAttrs; j++)
+ {
+ if (key->partattrs[i] == indexInfo->ii_KeyAttrNumbers[j])
+ {
+ found = true;
+ break;
+ }
+ }
+ if (!found)
+ {
+ Form_pg_attribute att;
+
+ att = TupleDescAttr(RelationGetDescr(rel), key->partattrs[i] - 1);
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("insufficient columns in UNIQUE constraint definition"),
+ errdetail("UNIQUE constraint on table \"%s\" does not include column \"%s\" which is part of the partition key.",
+ RelationGetRelationName(rel),
+ NameStr(att->attname))));
+ }
+ }
+ }
+
+
+ /*
* We disallow indexes on system columns other than OID. They would not
* necessarily get updated correctly, and they don't seem useful anyway.
*/
@@ -733,12 +790,14 @@ DefineIndex(Oid relationId,
indexRelationId =
index_create(rel, indexRelationName, indexRelationId, parentIndexId,
+ parentConstraintId,
stmt->oldNode, indexInfo, indexColNames,
accessMethodId, tablespaceId,
collationObjectId, classObjectId,
coloptions, reloptions,
flags, constr_flags,
- allowSystemTableMods, !check_rights);
+ allowSystemTableMods, !check_rights,
+ &createdConstraintId);
ObjectAddressSet(address, RelationRelationId, indexRelationId);
@@ -850,6 +909,7 @@ DefineIndex(Oid relationId,
DefineIndex(childRelid, childStmt,
InvalidOid, /* no predefined OID */
indexRelationId, /* this is our child */
+ createdConstraintId,
false, check_rights, check_not_in_use,
false, quiet);
}
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 26cec0b480..b0ddfb9110 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -938,17 +938,20 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
Relation idxRel = index_open(lfirst_oid(cell), AccessShareLock);
AttrNumber *attmap;
IndexStmt *idxstmt;
+ Oid constraintOid;
attmap = convert_tuples_by_name_map(RelationGetDescr(rel),
RelationGetDescr(parent),
gettext_noop("could not convert row type"));
idxstmt =
generateClonedIndexStmt(NULL, RelationGetRelid(rel), idxRel,
- attmap, RelationGetDescr(rel)->natts);
+ attmap, RelationGetDescr(rel)->natts,
+ &constraintOid);
DefineIndex(RelationGetRelid(rel),
idxstmt,
InvalidOid,
RelationGetRelid(idxRel),
+ constraintOid,
false, false, false, false, false);
index_close(idxRel, AccessShareLock);
@@ -6877,6 +6880,7 @@ ATExecAddIndex(AlteredTableInfo *tab, Relation rel,
stmt,
InvalidOid, /* no predefined OID */
InvalidOid, /* no parent index */
+ InvalidOid, /* no parent constraint */
true, /* is_alter_table */
check_rights,
false, /* check_not_in_use - we did it already */
@@ -6969,6 +6973,7 @@ ATExecAddIndexConstraint(AlteredTableInfo *tab, Relation rel,
address = index_constraint_create(rel,
index_oid,
+ InvalidOid,
indexInfo,
constraintName,
constraintType,
@@ -14170,12 +14175,15 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
if (!found)
{
IndexStmt *stmt;
+ Oid constraintOid;
stmt = generateClonedIndexStmt(NULL, RelationGetRelid(attachrel),
idxRel, attmap,
- RelationGetDescr(rel)->natts);
+ RelationGetDescr(rel)->natts,
+ &constraintOid);
DefineIndex(RelationGetRelid(attachrel), stmt, InvalidOid,
RelationGetRelid(idxRel),
+ constraintOid,
false, false, false, false, false);
}
diff --git a/src/backend/parser/parse_utilcmd.c b/src/backend/parser/parse_utilcmd.c
index 45f6ec2820..917b9bd636 100644
--- a/src/backend/parser/parse_utilcmd.c
+++ b/src/backend/parser/parse_utilcmd.c
@@ -704,12 +704,6 @@ transformColumnDefinition(CreateStmtContext *cxt, ColumnDef *column)
errmsg("primary key constraints are not supported on foreign tables"),
parser_errposition(cxt->pstate,
constraint->location)));
- if (cxt->ispartitioned)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("primary key constraints are not supported on partitioned tables"),
- parser_errposition(cxt->pstate,
- constraint->location)));
/* FALL THRU */
case CONSTR_UNIQUE:
@@ -719,12 +713,6 @@ transformColumnDefinition(CreateStmtContext *cxt, ColumnDef *column)
errmsg("unique constraints are not supported on foreign tables"),
parser_errposition(cxt->pstate,
constraint->location)));
- if (cxt->ispartitioned)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("unique constraints are not supported on partitioned tables"),
- parser_errposition(cxt->pstate,
- constraint->location)));
if (constraint->keys == NIL)
constraint->keys = list_make1(makeString(column->colname));
cxt->ixconstraints = lappend(cxt->ixconstraints, constraint);
@@ -821,12 +809,6 @@ transformTableConstraint(CreateStmtContext *cxt, Constraint *constraint)
errmsg("primary key constraints are not supported on foreign tables"),
parser_errposition(cxt->pstate,
constraint->location)));
- if (cxt->ispartitioned)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("primary key constraints are not supported on partitioned tables"),
- parser_errposition(cxt->pstate,
- constraint->location)));
cxt->ixconstraints = lappend(cxt->ixconstraints, constraint);
break;
@@ -837,12 +819,6 @@ transformTableConstraint(CreateStmtContext *cxt, Constraint *constraint)
errmsg("unique constraints are not supported on foreign tables"),
parser_errposition(cxt->pstate,
constraint->location)));
- if (cxt->ispartitioned)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("unique constraints are not supported on partitioned tables"),
- parser_errposition(cxt->pstate,
- constraint->location)));
cxt->ixconstraints = lappend(cxt->ixconstraints, constraint);
break;
@@ -1184,7 +1160,7 @@ transformTableLikeClause(CreateStmtContext *cxt, TableLikeClause *table_like_cla
/* Build CREATE INDEX statement to recreate the parent_index */
index_stmt = generateClonedIndexStmt(cxt->relation, InvalidOid,
parent_index,
- attmap, tupleDesc->natts);
+ attmap, tupleDesc->natts, NULL);
/* Copy comment on index, if requested */
if (table_like_clause->options & CREATE_TABLE_LIKE_COMMENTS)
@@ -1267,7 +1243,7 @@ transformOfType(CreateStmtContext *cxt, TypeName *ofTypename)
*/
IndexStmt *
generateClonedIndexStmt(RangeVar *heapRel, Oid heapRelid, Relation source_idx,
- const AttrNumber *attmap, int attmap_length)
+ const AttrNumber *attmap, int attmap_length, Oid *constraintOid)
{
Oid source_relid = RelationGetRelid(source_idx);
HeapTuple ht_idxrel;
@@ -1365,6 +1341,9 @@ generateClonedIndexStmt(RangeVar *heapRel, Oid heapRelid, Relation source_idx,
HeapTuple ht_constr;
Form_pg_constraint conrec;
+ if (constraintOid)
+ *constraintOid = constraintId;
+
ht_constr = SearchSysCache1(CONSTROID,
ObjectIdGetDatum(constraintId));
if (!HeapTupleIsValid(ht_constr))
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 8b5fd95a96..974b2f2c33 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1337,6 +1337,7 @@ ProcessUtilitySlow(ParseState *pstate,
stmt,
InvalidOid, /* no predefined OID */
InvalidOid, /* no parent index */
+ InvalidOid, /* no parent constraint */
false, /* is_alter_table */
true, /* check_rights */
true, /* check_not_in_use */
diff --git a/src/include/catalog/index.h b/src/include/catalog/index.h
index 112d69debc..9a35ad6c0d 100644
--- a/src/include/catalog/index.h
+++ b/src/include/catalog/index.h
@@ -54,6 +54,7 @@ extern Oid index_create(Relation heapRelation,
const char *indexRelationName,
Oid indexRelationId,
Oid parentIndexRelid,
+ Oid parentConstraintId,
Oid relFileNode,
IndexInfo *indexInfo,
List *indexColNames,
@@ -66,7 +67,8 @@ extern Oid index_create(Relation heapRelation,
bits16 flags,
bits16 constr_flags,
bool allow_system_table_mods,
- bool is_internal);
+ bool is_internal,
+ Oid *constraintId);
#define INDEX_CONSTR_CREATE_MARK_AS_PRIMARY (1 << 0)
#define INDEX_CONSTR_CREATE_DEFERRABLE (1 << 1)
@@ -76,6 +78,7 @@ extern Oid index_create(Relation heapRelation,
extern ObjectAddress index_constraint_create(Relation heapRelation,
Oid indexRelationId,
+ Oid parentConstraintId,
IndexInfo *indexInfo,
const char *constraintName,
char constraintType,
diff --git a/src/include/commands/defrem.h b/src/include/commands/defrem.h
index 35f50c1175..1435a31940 100644
--- a/src/include/commands/defrem.h
+++ b/src/include/commands/defrem.h
@@ -26,6 +26,7 @@ extern ObjectAddress DefineIndex(Oid relationId,
IndexStmt *stmt,
Oid indexRelationId,
Oid parentIndexId,
+ Oid parentConstraintId,
bool is_alter_table,
bool check_rights,
bool check_not_in_use,
diff --git a/src/include/parser/parse_utilcmd.h b/src/include/parser/parse_utilcmd.h
index 599f0e8e29..c130b3a93a 100644
--- a/src/include/parser/parse_utilcmd.h
+++ b/src/include/parser/parse_utilcmd.h
@@ -29,6 +29,7 @@ extern PartitionBoundSpec *transformPartitionBound(ParseState *pstate, Relation
PartitionBoundSpec *spec);
extern IndexStmt *generateClonedIndexStmt(RangeVar *heapRel, Oid heapOid,
Relation source_idx,
- const AttrNumber *attmap, int attmap_length);
+ const AttrNumber *attmap, int attmap_length,
+ Oid *constraintOid);
#endif /* PARSE_UTILCMD_H */
diff --git a/src/test/regress/expected/alter_table.out b/src/test/regress/expected/alter_table.out
index 517fb080bd..2caf930242 100644
--- a/src/test/regress/expected/alter_table.out
+++ b/src/test/regress/expected/alter_table.out
@@ -3290,14 +3290,6 @@ CREATE TABLE partitioned (
a int,
b int
) PARTITION BY RANGE (a, (a+b+1));
-ALTER TABLE partitioned ADD UNIQUE (a);
-ERROR: unique constraints are not supported on partitioned tables
-LINE 1: ALTER TABLE partitioned ADD UNIQUE (a);
- ^
-ALTER TABLE partitioned ADD PRIMARY KEY (a);
-ERROR: primary key constraints are not supported on partitioned tables
-LINE 1: ALTER TABLE partitioned ADD PRIMARY KEY (a);
- ^
ALTER TABLE partitioned ADD FOREIGN KEY (a) REFERENCES blah;
ERROR: foreign key constraints are not supported on partitioned tables
LINE 1: ALTER TABLE partitioned ADD FOREIGN KEY (a) REFERENCES blah;
diff --git a/src/test/regress/expected/create_table.out b/src/test/regress/expected/create_table.out
index 8e745402ae..866cc99b9f 100644
--- a/src/test/regress/expected/create_table.out
+++ b/src/test/regress/expected/create_table.out
@@ -276,12 +276,6 @@ CREATE TABLE partitioned (
) PARTITION BY LIST (a1, a2); -- fail
ERROR: cannot use "list" partition strategy with more than one column
-- unsupported constraint type for partitioned tables
-CREATE TABLE partitioned (
- a int PRIMARY KEY
-) PARTITION BY RANGE (a);
-ERROR: primary key constraints are not supported on partitioned tables
-LINE 2: a int PRIMARY KEY
- ^
CREATE TABLE pkrel (
a int PRIMARY KEY
);
@@ -293,12 +287,6 @@ LINE 2: a int REFERENCES pkrel(a)
^
DROP TABLE pkrel;
CREATE TABLE partitioned (
- a int UNIQUE
-) PARTITION BY RANGE (a);
-ERROR: unique constraints are not supported on partitioned tables
-LINE 2: a int UNIQUE
- ^
-CREATE TABLE partitioned (
a int,
EXCLUDE USING gist (a WITH &&)
) PARTITION BY RANGE (a);
diff --git a/src/test/regress/expected/indexing.out b/src/test/regress/expected/indexing.out
index 7d463cde15..c888bdb6d5 100644
--- a/src/test/regress/expected/indexing.out
+++ b/src/test/regress/expected/indexing.out
@@ -25,8 +25,6 @@ drop table idxpart;
-- Some unsupported features
create table idxpart (a int, b int, c text) partition by range (a);
create table idxpart1 partition of idxpart for values from (0) to (10);
-create unique index on idxpart (a);
-ERROR: cannot create unique index on partitioned table "idxpart"
create index concurrently on idxpart (a);
ERROR: cannot create index on partitioned table "idxpart" concurrently
drop table idxpart;
@@ -458,6 +456,146 @@ select attrelid::regclass, attname, attnum from pg_attribute
(7 rows)
drop table idxpart;
+--
+-- Constraint-related indexes
+--
+-- Verify that it works to add primary key / unique to partitioned tables
+create table idxpart (a int primary key, b int) partition by range (a);
+\d idxpart
+ Table "public.idxpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | not null |
+ b | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "idxpart_pkey" PRIMARY KEY, btree (a)
+Number of partitions: 0
+
+drop table idxpart;
+-- but not if you fail to use the full partition key
+create table idxpart (a int unique, b int) partition by range (a, b);
+ERROR: insufficient columns in UNIQUE constraint definition
+DETAIL: UNIQUE constraint on table "idxpart" does not include column "b" which is part of the partition key.
+create table idxpart (a int, b int unique) partition by range (a, b);
+ERROR: insufficient columns in UNIQUE constraint definition
+DETAIL: UNIQUE constraint on table "idxpart" does not include column "a" which is part of the partition key.
+create table idxpart (a int primary key, b int) partition by range (b, a);
+ERROR: insufficient columns in UNIQUE constraint definition
+DETAIL: UNIQUE constraint on table "idxpart" does not include column "b" which is part of the partition key.
+create table idxpart (a int, b int primary key) partition by range (b, a);
+ERROR: insufficient columns in UNIQUE constraint definition
+DETAIL: UNIQUE constraint on table "idxpart" does not include column "a" which is part of the partition key.
+-- OK if you use them in some other order
+create table idxpart (a int, b int, c text, primary key (a, b, c)) partition by range (b, c, a);
+drop table idxpart;
+create table idxpart (a int primary key, b int) partition by range ((b + a));
+ERROR: UNIQUE constraints are not supported on partitioned tables using expressions as partition keys
+-- not other types of index-based constraints
+create table idxpart (a int, exclude (a with = )) partition by range (a);
+ERROR: exclusion constraints are not supported on partitioned tables
+LINE 1: create table idxpart (a int, exclude (a with = )) partition ...
+ ^
+-- It works to add primary keys after the partitioned table is created
+create table idxpart (a int, b int, c text) partition by range (a, b);
+alter table idxpart add primary key (a); -- not an incomplete one tho
+ERROR: insufficient columns in UNIQUE constraint definition
+DETAIL: UNIQUE constraint on table "idxpart" does not include column "b" which is part of the partition key.
+alter table idxpart add primary key (a, b);
+\d idxpart
+ Table "public.idxpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | not null |
+ b | integer | | not null |
+ c | text | | |
+Partition key: RANGE (a, b)
+Indexes:
+ "idxpart_pkey" PRIMARY KEY, btree (a, b)
+Number of partitions: 0
+
+create table idxpart1 partition of idxpart for values from (0, 0) to (1000, 1000);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | not null |
+ b | integer | | not null |
+ c | text | | |
+Partition of: idxpart FOR VALUES FROM (0, 0) TO (1000, 1000)
+Indexes:
+ "idxpart1_pkey" PRIMARY KEY, btree (a, b)
+
+drop table idxpart;
+-- It works to add unique constraints after the partitioned table is created
+create table idxpart (a int, b int) partition by range (a, b);
+alter table idxpart add unique (a); -- ... nope
+ERROR: insufficient columns in UNIQUE constraint definition
+DETAIL: UNIQUE constraint on table "idxpart" does not include column "b" which is part of the partition key.
+alter table idxpart add unique (b, a);
+\d idxpart
+ Table "public.idxpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | integer | | |
+Partition key: RANGE (a, b)
+Indexes:
+ "idxpart_b_a_key" UNIQUE CONSTRAINT, btree (b, a)
+Number of partitions: 0
+
+drop table idxpart;
+-- Exclusion constraints cannot be added
+create table idxpart (a int, b int) partition by range (a);
+alter table idxpart add exclude (a with =);
+ERROR: exclusion constraints are not supported on partitioned tables
+LINE 1: alter table idxpart add exclude (a with =);
+ ^
+drop table idxpart;
+-- When (sub)partitions are created, they also contain the constraint
+create table idxpart (a int, b int, primary key (a, b)) partition by range (a, b);
+create table idxpart1 partition of idxpart for values from (1, 1) to (10, 10);
+create table idxpart2 partition of idxpart for values from (10, 10) to (20, 20)
+ partition by range (b);
+create table idxpart21 partition of idxpart2 for values from (10) to (15);
+create table idxpart22 partition of idxpart2 for values from (15) to (20);
+create table idxpart3 (b int not null, a int not null);
+alter table idxpart attach partition idxpart3 for values from (20, 20) to (30, 30);
+select conname, contype, conrelid::regclass, conindid::regclass, conkey
+ from pg_constraint where conrelid::regclass::text like 'idxpart%'
+ order by conname;
+ conname | contype | conrelid | conindid | conkey
+----------------+---------+-----------+----------------+--------
+ idxpart1_pkey | p | idxpart1 | idxpart1_pkey | {1,2}
+ idxpart21_pkey | p | idxpart21 | idxpart21_pkey | {1,2}
+ idxpart22_pkey | p | idxpart22 | idxpart22_pkey | {1,2}
+ idxpart2_pkey | p | idxpart2 | idxpart2_pkey | {1,2}
+ idxpart3_pkey | p | idxpart3 | idxpart3_pkey | {2,1}
+ idxpart_pkey | p | idxpart | idxpart_pkey | {1,2}
+(6 rows)
+
+drop table idxpart;
+-- multi-layer partitioning honors the prohibition. So this fails:
+create table idxpart (a int, b int, primary key (a)) partition by range (a);
+create table idxpart2 partition of idxpart for values from (0) to (1000) partition by range (b);
+ERROR: insufficient columns in UNIQUE constraint definition
+DETAIL: UNIQUE constraint on table "idxpart2" does not include column "b" which is part of the partition key.
+drop table idxpart;
+-- but this works:
+create table idxpart (a int, b int, primary key (a, b)) partition by range (a);
+create table idxpart2 partition of idxpart for values from (0) to (1000) partition by range (b);
+create table idxpart21 partition of idxpart2 for values from (0) to (1000);
+select conname, contype, conrelid::regclass, conindid::regclass, conkey
+ from pg_constraint where conrelid::regclass::text like 'idxpart%'
+ order by conname;
+ conname | contype | conrelid | conindid | conkey
+----------------+---------+-----------+----------------+--------
+ idxpart21_pkey | p | idxpart21 | idxpart21_pkey | {1,2}
+ idxpart2_pkey | p | idxpart2 | idxpart2_pkey | {1,2}
+ idxpart_pkey | p | idxpart | idxpart_pkey | {1,2}
+(3 rows)
+
+drop table idxpart;
-- intentionally leave some objects around
create table idxpart (a int) partition by range (a);
create table idxpart1 partition of idxpart for values from (0) to (100);
diff --git a/src/test/regress/sql/alter_table.sql b/src/test/regress/sql/alter_table.sql
index af25ee9e77..ed0bb7845b 100644
--- a/src/test/regress/sql/alter_table.sql
+++ b/src/test/regress/sql/alter_table.sql
@@ -2016,8 +2016,6 @@ CREATE TABLE partitioned (
a int,
b int
) PARTITION BY RANGE (a, (a+b+1));
-ALTER TABLE partitioned ADD UNIQUE (a);
-ALTER TABLE partitioned ADD PRIMARY KEY (a);
ALTER TABLE partitioned ADD FOREIGN KEY (a) REFERENCES blah;
ALTER TABLE partitioned ADD EXCLUDE USING gist (a WITH &&);
diff --git a/src/test/regress/sql/create_table.sql b/src/test/regress/sql/create_table.sql
index 8f9991ef18..fefccf21a2 100644
--- a/src/test/regress/sql/create_table.sql
+++ b/src/test/regress/sql/create_table.sql
@@ -294,10 +294,6 @@ CREATE TABLE partitioned (
) PARTITION BY LIST (a1, a2); -- fail
-- unsupported constraint type for partitioned tables
-CREATE TABLE partitioned (
- a int PRIMARY KEY
-) PARTITION BY RANGE (a);
-
CREATE TABLE pkrel (
a int PRIMARY KEY
);
@@ -307,10 +303,6 @@ CREATE TABLE partitioned (
DROP TABLE pkrel;
CREATE TABLE partitioned (
- a int UNIQUE
-) PARTITION BY RANGE (a);
-
-CREATE TABLE partitioned (
a int,
EXCLUDE USING gist (a WITH &&)
) PARTITION BY RANGE (a);
diff --git a/src/test/regress/sql/indexing.sql b/src/test/regress/sql/indexing.sql
index 41d50ffaa4..b1f82c51d8 100644
--- a/src/test/regress/sql/indexing.sql
+++ b/src/test/regress/sql/indexing.sql
@@ -14,7 +14,6 @@ drop table idxpart;
-- Some unsupported features
create table idxpart (a int, b int, c text) partition by range (a);
create table idxpart1 partition of idxpart for values from (0) to (10);
-create unique index on idxpart (a);
create index concurrently on idxpart (a);
drop table idxpart;
@@ -186,6 +185,78 @@ select attrelid::regclass, attname, attnum from pg_attribute
order by attrelid::regclass, attnum;
drop table idxpart;
+--
+-- Constraint-related indexes
+--
+
+-- Verify that it works to add primary key / unique to partitioned tables
+create table idxpart (a int primary key, b int) partition by range (a);
+\d idxpart
+drop table idxpart;
+
+-- but not if you fail to use the full partition key
+create table idxpart (a int unique, b int) partition by range (a, b);
+create table idxpart (a int, b int unique) partition by range (a, b);
+create table idxpart (a int primary key, b int) partition by range (b, a);
+create table idxpart (a int, b int primary key) partition by range (b, a);
+
+-- OK if you use them in some other order
+create table idxpart (a int, b int, c text, primary key (a, b, c)) partition by range (b, c, a);
+drop table idxpart;
+
+create table idxpart (a int primary key, b int) partition by range ((b + a));
+-- not other types of index-based constraints
+create table idxpart (a int, exclude (a with = )) partition by range (a);
+
+-- It works to add primary keys after the partitioned table is created
+create table idxpart (a int, b int, c text) partition by range (a, b);
+alter table idxpart add primary key (a); -- not an incomplete one tho
+alter table idxpart add primary key (a, b);
+\d idxpart
+create table idxpart1 partition of idxpart for values from (0, 0) to (1000, 1000);
+\d idxpart1
+drop table idxpart;
+
+-- It works to add unique constraints after the partitioned table is created
+create table idxpart (a int, b int) partition by range (a, b);
+alter table idxpart add unique (a); -- ... nope
+alter table idxpart add unique (b, a);
+\d idxpart
+drop table idxpart;
+
+-- Exclusion constraints cannot be added
+create table idxpart (a int, b int) partition by range (a);
+alter table idxpart add exclude (a with =);
+drop table idxpart;
+
+-- When (sub)partitions are created, they also contain the constraint
+create table idxpart (a int, b int, primary key (a, b)) partition by range (a, b);
+create table idxpart1 partition of idxpart for values from (1, 1) to (10, 10);
+create table idxpart2 partition of idxpart for values from (10, 10) to (20, 20)
+ partition by range (b);
+create table idxpart21 partition of idxpart2 for values from (10) to (15);
+create table idxpart22 partition of idxpart2 for values from (15) to (20);
+create table idxpart3 (b int not null, a int not null);
+alter table idxpart attach partition idxpart3 for values from (20, 20) to (30, 30);
+select conname, contype, conrelid::regclass, conindid::regclass, conkey
+ from pg_constraint where conrelid::regclass::text like 'idxpart%'
+ order by conname;
+drop table idxpart;
+
+-- multi-layer partitioning honors the prohibition. So this fails:
+create table idxpart (a int, b int, primary key (a)) partition by range (a);
+create table idxpart2 partition of idxpart for values from (0) to (1000) partition by range (b);
+drop table idxpart;
+
+-- but this works:
+create table idxpart (a int, b int, primary key (a, b)) partition by range (a);
+create table idxpart2 partition of idxpart for values from (0) to (1000) partition by range (b);
+create table idxpart21 partition of idxpart2 for values from (0) to (1000);
+select conname, contype, conrelid::regclass, conindid::regclass, conkey
+ from pg_constraint where conrelid::regclass::text like 'idxpart%'
+ order by conname;
+drop table idxpart;
+
-- intentionally leave some objects around
create table idxpart (a int) partition by range (a);
create table idxpart1 partition of idxpart for values from (0) to (100);
--
2.11.0
v1-0003-on-conflict-do-update-not-supported.patchtext/plain; charset=us-asciiDownload
From ea3b02295fd431095b95ab319ff3034e038c83fa Mon Sep 17 00:00:00 2001
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
Date: Fri, 29 Dec 2017 18:43:00 -0300
Subject: [PATCH v1 3/4] on conflict do update not supported
---
src/backend/parser/analyze.c | 7 +++++++
1 file changed, 7 insertions(+)
diff --git a/src/backend/parser/analyze.c b/src/backend/parser/analyze.c
index d680d2285c..d8091a1435 100644
--- a/src/backend/parser/analyze.c
+++ b/src/backend/parser/analyze.c
@@ -1017,6 +1017,13 @@ transformOnConflictClause(ParseState *pstate,
TargetEntry *te;
int attno;
+ if (targetrel->rd_partdesc)
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("%s cannot be applied to partitioned table \"%s\"",
+ "ON CONFLICT DO UPDATE",
+ RelationGetRelationName(targetrel))));
+
/*
* All INSERT expressions have been parsed, get ready for potentially
* existing SET statements that need to be processed like an UPDATE.
--
2.11.0
v1-0004-Teach-executor-to-handle-ON-CONFLICT-key-on-parti.patchtext/plain; charset=us-asciiDownload
From 77817d057ce5fc6930305aebe3c87806d04e6acf Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Mon, 25 Dec 2017 18:45:33 +0900
Subject: [PATCH v1 4/4] Teach executor to handle ON CONFLICT (key) on
partitioned tables
---
src/backend/executor/execIndexing.c | 14 +++++++++---
src/backend/executor/nodeModifyTable.c | 32 +++++++++++++++++++++++++++
src/backend/parser/analyze.c | 7 ------
src/test/regress/expected/insert_conflict.out | 27 +++++++++++++++++-----
src/test/regress/sql/insert_conflict.sql | 18 ++++++++++-----
5 files changed, 77 insertions(+), 21 deletions(-)
diff --git a/src/backend/executor/execIndexing.c b/src/backend/executor/execIndexing.c
index 89e189fa71..f76a2ede76 100644
--- a/src/backend/executor/execIndexing.c
+++ b/src/backend/executor/execIndexing.c
@@ -531,10 +531,18 @@ ExecCheckIndexConstraints(TupleTableSlot *slot,
if (!indexInfo->ii_ReadyForInserts)
continue;
- /* When specific arbiter indexes requested, only examine them */
+ /*
+ * When specific arbiter indexes requested, only examine them. If
+ * this is a partition (after a tuple is routed to it from the
+ * parent into which the original tuple has been inserted), we must
+ * check the parent index id, instead of our own id, because that's
+ * the one that appears in the arbiterIndexes list.
+ */
if (arbiterIndexes != NIL &&
- !list_member_oid(arbiterIndexes,
- indexRelation->rd_index->indexrelid))
+ !(list_member_oid(arbiterIndexes,
+ indexRelation->rd_index->indexrelid) ||
+ list_member_oid(arbiterIndexes,
+ indexRelation->rd_index->indparentidx)))
continue;
if (!indexRelation->rd_index->indimmediate)
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 82cd4462a3..deced3585a 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -2184,6 +2184,38 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
resultRelInfo->ri_onConflictSetWhere = qualexpr;
}
+
+ /* Build the above information for each leaf partition rel */
+ for (i = 0; i < mtstate->mt_num_partitions; i++)
+ {
+ Relation partrel;
+ List *leaf_oc_set;
+
+ resultRelInfo = mtstate->mt_partitions[i];
+ partrel = resultRelInfo->ri_RelationDesc;
+
+ /* varno = node->nominalRelation */
+ leaf_oc_set = map_partition_varattnos(node->onConflictSet,
+ node->nominalRelation,
+ partrel, rel, NULL);
+ resultRelInfo->ri_onConflictSetProj =
+ ExecBuildProjectionInfo(leaf_oc_set, econtext,
+ mtstate->mt_conflproj, &mtstate->ps,
+ resultRelInfo->ri_RelationDesc->rd_att);
+
+ if (node->onConflictWhere)
+ {
+ List *leaf_oc_where;
+
+ /* varno = node->nominalRelation */
+ leaf_oc_where =
+ map_partition_varattnos((List *) node->onConflictWhere,
+ node->nominalRelation,
+ partrel, rel, NULL);
+ resultRelInfo->ri_onConflictSetWhere =
+ ExecInitQual(leaf_oc_where, &mtstate->ps);
+ }
+ }
}
/*
diff --git a/src/backend/parser/analyze.c b/src/backend/parser/analyze.c
index d8091a1435..d680d2285c 100644
--- a/src/backend/parser/analyze.c
+++ b/src/backend/parser/analyze.c
@@ -1017,13 +1017,6 @@ transformOnConflictClause(ParseState *pstate,
TargetEntry *te;
int attno;
- if (targetrel->rd_partdesc)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("%s cannot be applied to partitioned table \"%s\"",
- "ON CONFLICT DO UPDATE",
- RelationGetRelationName(targetrel))));
-
/*
* All INSERT expressions have been parsed, get ready for potentially
* existing SET statements that need to be processed like an UPDATE.
diff --git a/src/test/regress/expected/insert_conflict.out b/src/test/regress/expected/insert_conflict.out
index 8fd2027d6a..dcb07fc09e 100644
--- a/src/test/regress/expected/insert_conflict.out
+++ b/src/test/regress/expected/insert_conflict.out
@@ -792,10 +792,25 @@ create table parted_conflict_test (a int, b char) partition by list (a);
create table parted_conflict_test_1 partition of parted_conflict_test (b unique) for values in (1);
insert into parted_conflict_test values (1, 'a') on conflict do nothing;
insert into parted_conflict_test values (1, 'a') on conflict do nothing;
--- however, on conflict do update is not supported yet
-insert into parted_conflict_test values (1) on conflict (b) do update set a = excluded.a;
-ERROR: there is no unique or exclusion constraint matching the ON CONFLICT specification
--- but it works OK if we target the partition directly
-insert into parted_conflict_test_1 values (1) on conflict (b) do
-update set a = excluded.a;
+-- create one more partition and a partitioned unique index
+create table parted_conflict_test_2 partition of parted_conflict_test for values in (2);
+create unique index on parted_conflict_test (a);
+insert into parted_conflict_test values (1, 'a') on conflict (a) do nothing;
+insert into parted_conflict_test values (1, 'b') on conflict (a) do update set b = excluded.b;
+insert into parted_conflict_test values (1, 'b') on conflict (a) do update set b = excluded.b where parted_conflict_test.b = 'a';
+select * from parted_conflict_test;
+ a | b
+---+---
+ 1 | b
+(1 row)
+
+-- also works OK if we target the partition directly
+insert into parted_conflict_test_1 values (1, 'c') on conflict (a) do
+update set b = excluded.b;
+select * from parted_conflict_test;
+ a | b
+---+---
+ 1 | c
+(1 row)
+
drop table parted_conflict_test;
diff --git a/src/test/regress/sql/insert_conflict.sql b/src/test/regress/sql/insert_conflict.sql
index 32c647e3f8..264f67ce89 100644
--- a/src/test/regress/sql/insert_conflict.sql
+++ b/src/test/regress/sql/insert_conflict.sql
@@ -478,9 +478,17 @@ create table parted_conflict_test (a int, b char) partition by list (a);
create table parted_conflict_test_1 partition of parted_conflict_test (b unique) for values in (1);
insert into parted_conflict_test values (1, 'a') on conflict do nothing;
insert into parted_conflict_test values (1, 'a') on conflict do nothing;
--- however, on conflict do update is not supported yet
-insert into parted_conflict_test values (1) on conflict (b) do update set a = excluded.a;
--- but it works OK if we target the partition directly
-insert into parted_conflict_test_1 values (1) on conflict (b) do
-update set a = excluded.a;
+
+-- create one more partition and a partitioned unique index
+create table parted_conflict_test_2 partition of parted_conflict_test for values in (2);
+create unique index on parted_conflict_test (a);
+insert into parted_conflict_test values (1, 'a') on conflict (a) do nothing;
+insert into parted_conflict_test values (1, 'b') on conflict (a) do update set b = excluded.b;
+insert into parted_conflict_test values (1, 'b') on conflict (a) do update set b = excluded.b where parted_conflict_test.b = 'a';
+select * from parted_conflict_test;
+
+-- also works OK if we target the partition directly
+insert into parted_conflict_test_1 values (1, 'c') on conflict (a) do
+update set b = excluded.b;
+select * from parted_conflict_test;
drop table parted_conflict_test;
--
2.11.0
Re: unique indexes on partitioned tables
This new version fixes markup mistakes in the docs, and nothing else.
I'm not posting the ON CONFLICT DO UPDATE patch from Amit, since I
haven't fixed it; the 0003 patch has been squashed on 0002 instead, with
regression tests adapted. I'll see about the ON CONFLICT stuff as I
have time.
The 0001 patch is the same I posted last time in the other thread.
--
�lvaro Herrera https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
Attachments:
v2-0001-Local-partitioned-indexes.patchtext/plain; charset=us-asciiDownload
From 6ef141ed05bb4a03d1100dd5d39c9b91b9442990 Mon Sep 17 00:00:00 2001
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
Date: Mon, 23 Oct 2017 10:18:38 +0200
Subject: [PATCH v2 1/2] Local partitioned indexes
When CREATE INDEX is run on a partitioned table, create catalog entries
for an index on the partitioned table (which is just a placeholder since
the table proper has no data of its own), and recurse to create actual
indexes on the existing partitions; create them in future partitions
also.
As a convenience gadget, if the new index definition matches some
existing index in partitions, these are picked up and used instead of
creating new ones. Whichever way these indexes come about, they become
attached to the index on the parent table and are dropped alongside it,
and cannot be dropped on isolation unless they are detached first.
To support pg_dump'ing these indexes, add commands
CREATE INDEX ON ONLY <table>
(which creates the index on the parent partitioned table, without
recursing) and
ALTER INDEX ATTACH PARTITION
(which is used after the indexes have been created individually on each
partition, to attach them to the parent index). These reconstruct prior
database state exactly.
Reviewed-by: Robert Haas, Amit Langote, Jesper Pedersen, Simon Riggs, David Rowley
Discussion: https://postgr.es/m/20171113170646.gzweigyrgg6pwsg4@alvherre.pgsql
---
doc/src/sgml/catalogs.sgml | 18 +
doc/src/sgml/ref/alter_index.sgml | 14 +
doc/src/sgml/ref/alter_table.sgml | 8 +-
doc/src/sgml/ref/create_index.sgml | 33 +-
doc/src/sgml/ref/reindex.sgml | 5 +
src/backend/access/common/reloptions.c | 1 +
src/backend/access/heap/heapam.c | 9 +-
src/backend/access/index/indexam.c | 3 +-
src/backend/bootstrap/bootparse.y | 2 +
src/backend/catalog/aclchk.c | 9 +-
src/backend/catalog/dependency.c | 9 +-
src/backend/catalog/heap.c | 1 +
src/backend/catalog/index.c | 133 +++++++-
src/backend/catalog/objectaddress.c | 5 +-
src/backend/catalog/pg_depend.c | 13 +-
src/backend/catalog/toasting.c | 2 +
src/backend/commands/indexcmds.c | 273 ++++++++++++++-
src/backend/commands/tablecmds.c | 486 ++++++++++++++++++++++++++-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/nodes/outfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 10 +-
src/backend/parser/gram.y | 32 +-
src/backend/parser/parse_utilcmd.c | 65 ++--
src/backend/tcop/utility.c | 9 +
src/backend/utils/adt/amutils.c | 3 +-
src/backend/utils/adt/ruleutils.c | 17 +-
src/backend/utils/cache/relcache.c | 30 +-
src/bin/pg_dump/common.c | 104 ++++++
src/bin/pg_dump/pg_dump.c | 88 ++++-
src/bin/pg_dump/pg_dump.h | 11 +
src/bin/pg_dump/pg_dump_sort.c | 56 +++-
src/bin/psql/describe.c | 20 +-
src/bin/psql/tab-complete.c | 34 +-
src/include/catalog/catversion.h | 2 +-
src/include/catalog/dependency.h | 8 +
src/include/catalog/index.h | 7 +
src/include/catalog/indexing.h | 2 +
src/include/catalog/pg_class.h | 1 +
src/include/catalog/pg_index.h | 38 ++-
src/include/commands/defrem.h | 3 +-
src/include/nodes/execnodes.h | 1 +
src/include/nodes/parsenodes.h | 7 +-
src/include/parser/parse_utilcmd.h | 3 +
src/test/regress/expected/alter_table.out | 65 +++-
src/test/regress/expected/indexing.out | 533 ++++++++++++++++++++++++++++++
src/test/regress/parallel_schedule | 2 +-
src/test/regress/serial_schedule | 1 +
src/test/regress/sql/alter_table.sql | 16 +
src/test/regress/sql/indexing.sql | 238 +++++++++++++
50 files changed, 2291 insertions(+), 142 deletions(-)
create mode 100644 src/test/regress/expected/indexing.out
create mode 100644 src/test/regress/sql/indexing.sql
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3f02202caf..5db63985b2 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -2996,6 +2996,24 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
</varlistentry>
<varlistentry>
+ <term><symbol>DEPENDENCY_INTERNAL_AUTO</symbol> (<literal>I</literal>)</term>
+ <listitem>
+ <para>
+ The dependent object was created as part of creation of the
+ referenced object, and is really just a part of its internal
+ implementation. A <command>DROP</command> of the dependent object
+ will be disallowed outright (we'll tell the user to issue a
+ <command>DROP</command> against the referenced object, instead).
+ As opposed to <literal>DEPENDENCY_INTERNAL</literal>, [... explain the
+ difference ].
+ Example: a index
+ on a partition is made internal-auto-dependent on both the partition
+ itself as well as on the index on the partitioned table.
+ </para>
+ </listitem>
+ </varlistentry>
+
+ <varlistentry>
<term><symbol>DEPENDENCY_EXTENSION</symbol> (<literal>e</literal>)</term>
<listitem>
<para>
diff --git a/doc/src/sgml/ref/alter_index.sgml b/doc/src/sgml/ref/alter_index.sgml
index e54237272c..3984686d67 100644
--- a/doc/src/sgml/ref/alter_index.sgml
+++ b/doc/src/sgml/ref/alter_index.sgml
@@ -23,6 +23,7 @@ PostgreSQL documentation
<synopsis>
ALTER INDEX [ IF EXISTS ] <replaceable class="parameter">name</replaceable> RENAME TO <replaceable class="parameter">new_name</replaceable>
ALTER INDEX [ IF EXISTS ] <replaceable class="parameter">name</replaceable> SET TABLESPACE <replaceable class="parameter">tablespace_name</replaceable>
+ALTER INDEX <replaceable class="parameter">name</replaceable> ATTACH PARTITION <replaceable class="parameter">index_name</replaceable>
ALTER INDEX <replaceable class="parameter">name</replaceable> DEPENDS ON EXTENSION <replaceable class="parameter">extension_name</replaceable>
ALTER INDEX [ IF EXISTS ] <replaceable class="parameter">name</replaceable> SET ( <replaceable class="parameter">storage_parameter</replaceable> = <replaceable class="parameter">value</replaceable> [, ... ] )
ALTER INDEX [ IF EXISTS ] <replaceable class="parameter">name</replaceable> RESET ( <replaceable class="parameter">storage_parameter</replaceable> [, ... ] )
@@ -76,6 +77,19 @@ ALTER INDEX ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
</varlistentry>
<varlistentry>
+ <term><literal>ATTACH</literal></term>
+ <listitem>
+ <para>
+ Causes the named index to become attached to the altered index.
+ The named index must be on a partition of the table containing the
+ index being altered, and have an equivalent definition. An attached
+ index cannot be dropped by itself, and will automatically be dropped
+ if its parent index is dropped.
+ </para>
+ </listitem>
+ </varlistentry>
+
+ <varlistentry>
<term><literal>DEPENDS ON EXTENSION</literal></term>
<listitem>
<para>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index 7bcf242846..0a2f3e3646 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -783,7 +783,10 @@ ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
as a partition of the target table. The table can be attached
as a partition for specific values using <literal>FOR VALUES
</literal> or as a default partition by using <literal>DEFAULT
- </literal>.
+ </literal>. For each index in the target table, a corresponding
+ one will be created in the attached table; or, if an equivalent
+ index already exists, will be attached to the target table's index,
+ as if <command>ALTER INDEX ATTACH</command> had been executed.
</para>
<para>
@@ -844,7 +847,8 @@ ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
<para>
This form detaches specified partition of the target table. The detached
partition continues to exist as a standalone table, but no longer has any
- ties to the table from which it was detached.
+ ties to the table from which it was detached. Any indexes that were
+ attached to the target table's indexes are detached.
</para>
</listitem>
</varlistentry>
diff --git a/doc/src/sgml/ref/create_index.sgml b/doc/src/sgml/ref/create_index.sgml
index 025537575b..5137fe6383 100644
--- a/doc/src/sgml/ref/create_index.sgml
+++ b/doc/src/sgml/ref/create_index.sgml
@@ -21,7 +21,7 @@ PostgreSQL documentation
<refsynopsisdiv>
<synopsis>
-CREATE [ UNIQUE ] INDEX [ CONCURRENTLY ] [ [ IF NOT EXISTS ] <replaceable class="parameter">name</replaceable> ] ON <replaceable class="parameter">table_name</replaceable> [ USING <replaceable class="parameter">method</replaceable> ]
+CREATE [ UNIQUE ] INDEX [ CONCURRENTLY ] [ [ IF NOT EXISTS ] <replaceable class="parameter">name</replaceable> ] ON [ ONLY ] <replaceable class="parameter">table_name</replaceable> [ USING <replaceable class="parameter">method</replaceable> ]
( { <replaceable class="parameter">column_name</replaceable> | ( <replaceable class="parameter">expression</replaceable> ) } [ COLLATE <replaceable class="parameter">collation</replaceable> ] [ <replaceable class="parameter">opclass</replaceable> ] [ ASC | DESC ] [ NULLS { FIRST | LAST } ] [, ...] )
[ WITH ( <replaceable class="parameter">storage_parameter</replaceable> = <replaceable class="parameter">value</replaceable> [, ... ] ) ]
[ TABLESPACE <replaceable class="parameter">tablespace_name</replaceable> ]
@@ -152,6 +152,16 @@ CREATE [ UNIQUE ] INDEX [ CONCURRENTLY ] [ [ IF NOT EXISTS ] <replaceable class=
</varlistentry>
<varlistentry>
+ <term><literal>ONLY</literal></term>
+ <listitem>
+ <para>
+ Indicates not to recurse creating indexes on partitions, if the
+ table is partitioned. The default is to recurse.
+ </para>
+ </listitem>
+ </varlistentry>
+
+ <varlistentry>
<term><replaceable class="parameter">table_name</replaceable></term>
<listitem>
<para>
@@ -546,6 +556,27 @@ Indexes:
</para>
<para>
+ When <literal>CREATE INDEX</literal> is invoked on a partitioned
+ table, the default behavior is to recurse to all partitions to ensure
+ they all have matching indexes.
+ Each partition is first checked to determine whether an equivalent
+ index already exists, and if so, that index will become attached as a
+ partition index to the index being created, which will become its
+ parent index.
+ If no matching index exists, a new index will be created and
+ automatically attached; the name of the new index in each partition
+ will be determined as if no index name had been specified in the
+ command.
+ If the <literal>ONLY</literal> option is specified, no recursion
+ is done, and the index is marked invalid
+ (<command>ALTER INDEX ... ATTACH PARTITION</command> turns the index
+ valid, once all partitions acquire the index.) Note, however, that
+ any partition that is created in the future using
+ <command>CREATE TABLE ... PARTITION OF</command> will automatically
+ contain the index regardless of whether this option was specified.
+ </para>
+
+ <para>
For index methods that support ordered scans (currently, only B-tree),
the optional clauses <literal>ASC</literal>, <literal>DESC</literal>, <literal>NULLS
FIRST</literal>, and/or <literal>NULLS LAST</literal> can be specified to modify
diff --git a/doc/src/sgml/ref/reindex.sgml b/doc/src/sgml/ref/reindex.sgml
index 79f6931c6a..1c21fafb80 100644
--- a/doc/src/sgml/ref/reindex.sgml
+++ b/doc/src/sgml/ref/reindex.sgml
@@ -231,6 +231,11 @@ REINDEX [ ( VERBOSE ) ] { INDEX | TABLE | SCHEMA | DATABASE | SYSTEM } <replacea
reindex anything.
</para>
+ <para>
+ Reindexing partitioned tables or partitioned indexes is not supported.
+ Each individual partition can be reindexed separately instead.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/access/common/reloptions.c b/src/backend/access/common/reloptions.c
index 425bc5d06e..274f7aa8e9 100644
--- a/src/backend/access/common/reloptions.c
+++ b/src/backend/access/common/reloptions.c
@@ -993,6 +993,7 @@ extractRelOptions(HeapTuple tuple, TupleDesc tupdesc,
options = view_reloptions(datum, false);
break;
case RELKIND_INDEX:
+ case RELKIND_PARTITIONED_INDEX:
options = index_reloptions(amoptions, datum, false);
break;
case RELKIND_FOREIGN_TABLE:
diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c
index dbc8f2d6c7..be263850cd 100644
--- a/src/backend/access/heap/heapam.c
+++ b/src/backend/access/heap/heapam.c
@@ -1293,7 +1293,8 @@ heap_open(Oid relationId, LOCKMODE lockmode)
r = relation_open(relationId, lockmode);
- if (r->rd_rel->relkind == RELKIND_INDEX)
+ if (r->rd_rel->relkind == RELKIND_INDEX ||
+ r->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is an index",
@@ -1321,7 +1322,8 @@ heap_openrv(const RangeVar *relation, LOCKMODE lockmode)
r = relation_openrv(relation, lockmode);
- if (r->rd_rel->relkind == RELKIND_INDEX)
+ if (r->rd_rel->relkind == RELKIND_INDEX ||
+ r->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is an index",
@@ -1353,7 +1355,8 @@ heap_openrv_extended(const RangeVar *relation, LOCKMODE lockmode,
if (r)
{
- if (r->rd_rel->relkind == RELKIND_INDEX)
+ if (r->rd_rel->relkind == RELKIND_INDEX ||
+ r->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is an index",
diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c
index 1b61cd9515..91247f0fa5 100644
--- a/src/backend/access/index/indexam.c
+++ b/src/backend/access/index/indexam.c
@@ -154,7 +154,8 @@ index_open(Oid relationId, LOCKMODE lockmode)
r = relation_open(relationId, lockmode);
- if (r->rd_rel->relkind != RELKIND_INDEX)
+ if (r->rd_rel->relkind != RELKIND_INDEX &&
+ r->rd_rel->relkind != RELKIND_PARTITIONED_INDEX)
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is not an index",
diff --git a/src/backend/bootstrap/bootparse.y b/src/backend/bootstrap/bootparse.y
index 8c52846a92..dfd53fa054 100644
--- a/src/backend/bootstrap/bootparse.y
+++ b/src/backend/bootstrap/bootparse.y
@@ -321,6 +321,7 @@ Boot_DeclareIndexStmt:
DefineIndex(relationId,
stmt,
$4,
+ InvalidOid,
false,
false,
false,
@@ -365,6 +366,7 @@ Boot_DeclareUniqueIndexStmt:
DefineIndex(relationId,
stmt,
$5,
+ InvalidOid,
false,
false,
false,
diff --git a/src/backend/catalog/aclchk.c b/src/backend/catalog/aclchk.c
index fac80612b8..50a2e2681b 100644
--- a/src/backend/catalog/aclchk.c
+++ b/src/backend/catalog/aclchk.c
@@ -1824,7 +1824,8 @@ ExecGrant_Relation(InternalGrant *istmt)
pg_class_tuple = (Form_pg_class) GETSTRUCT(tuple);
/* Not sensible to grant on an index */
- if (pg_class_tuple->relkind == RELKIND_INDEX)
+ if (pg_class_tuple->relkind == RELKIND_INDEX ||
+ pg_class_tuple->relkind == RELKIND_PARTITIONED_INDEX)
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is an index",
@@ -5405,7 +5406,8 @@ recordExtObjInitPriv(Oid objoid, Oid classoid)
pg_class_tuple = (Form_pg_class) GETSTRUCT(tuple);
/* Indexes don't have permissions */
- if (pg_class_tuple->relkind == RELKIND_INDEX)
+ if (pg_class_tuple->relkind == RELKIND_INDEX ||
+ pg_class_tuple->relkind == RELKIND_PARTITIONED_INDEX)
return;
/* Composite types don't have permissions either */
@@ -5690,7 +5692,8 @@ removeExtObjInitPriv(Oid objoid, Oid classoid)
pg_class_tuple = (Form_pg_class) GETSTRUCT(tuple);
/* Indexes don't have permissions */
- if (pg_class_tuple->relkind == RELKIND_INDEX)
+ if (pg_class_tuple->relkind == RELKIND_INDEX ||
+ pg_class_tuple->relkind == RELKIND_PARTITIONED_INDEX)
return;
/* Composite types don't have permissions either */
diff --git a/src/backend/catalog/dependency.c b/src/backend/catalog/dependency.c
index 269111b4c1..62aa335abb 100644
--- a/src/backend/catalog/dependency.c
+++ b/src/backend/catalog/dependency.c
@@ -582,6 +582,7 @@ findDependentObjects(const ObjectAddress *object,
/* FALL THRU */
case DEPENDENCY_INTERNAL:
+ case DEPENDENCY_INTERNAL_AUTO:
/*
* This object is part of the internal implementation of
@@ -628,6 +629,9 @@ findDependentObjects(const ObjectAddress *object,
if (stack_address_present_add_flags(&otherObject, 0, stack))
break;
+ if (foundDep->deptype == DEPENDENCY_INTERNAL_AUTO)
+ break;
+
/*
* 3. Not all the owning objects have been visited, so
* transform this deletion request into a delete of this
@@ -675,6 +679,7 @@ findDependentObjects(const ObjectAddress *object,
/* And we're done here. */
systable_endscan(scan);
return;
+
case DEPENDENCY_PIN:
/*
@@ -762,6 +767,7 @@ findDependentObjects(const ObjectAddress *object,
case DEPENDENCY_AUTO_EXTENSION:
subflags = DEPFLAG_AUTO;
break;
+ case DEPENDENCY_INTERNAL_AUTO:
case DEPENDENCY_INTERNAL:
subflags = DEPFLAG_INTERNAL;
break;
@@ -1109,7 +1115,8 @@ doDeletion(const ObjectAddress *object, int flags)
{
char relKind = get_rel_relkind(object->objectId);
- if (relKind == RELKIND_INDEX)
+ if (relKind == RELKIND_INDEX ||
+ relKind == RELKIND_PARTITIONED_INDEX)
{
bool concurrent = ((flags & PERFORM_DELETION_CONCURRENTLY) != 0);
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 089b7965f2..99f4d59863 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -294,6 +294,7 @@ heap_create(const char *relname,
case RELKIND_COMPOSITE_TYPE:
case RELKIND_FOREIGN_TABLE:
case RELKIND_PARTITIONED_TABLE:
+ case RELKIND_PARTITIONED_INDEX:
create_storage = false;
/*
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 330488b96f..f34127c81b 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -41,6 +41,7 @@
#include "catalog/pg_collation.h"
#include "catalog/pg_constraint.h"
#include "catalog/pg_constraint_fn.h"
+#include "catalog/pg_depend.h"
#include "catalog/pg_operator.h"
#include "catalog/pg_opclass.h"
#include "catalog/pg_tablespace.h"
@@ -98,6 +99,7 @@ static void InitializeAttributeOids(Relation indexRelation,
int numatts, Oid indexoid);
static void AppendAttributeTuples(Relation indexRelation, int numatts);
static void UpdateIndexRelation(Oid indexoid, Oid heapoid,
+ Oid parentIndexId,
IndexInfo *indexInfo,
Oid *collationOids,
Oid *classOids,
@@ -105,7 +107,8 @@ static void UpdateIndexRelation(Oid indexoid, Oid heapoid,
bool primary,
bool isexclusion,
bool immediate,
- bool isvalid);
+ bool isvalid,
+ bool isready);
static void index_update_stats(Relation rel,
bool hasindex, bool isprimary,
double reltuples);
@@ -551,6 +554,7 @@ AppendAttributeTuples(Relation indexRelation, int numatts)
static void
UpdateIndexRelation(Oid indexoid,
Oid heapoid,
+ Oid parentIndexOid,
IndexInfo *indexInfo,
Oid *collationOids,
Oid *classOids,
@@ -558,7 +562,8 @@ UpdateIndexRelation(Oid indexoid,
bool primary,
bool isexclusion,
bool immediate,
- bool isvalid)
+ bool isvalid,
+ bool isready)
{
int2vector *indkey;
oidvector *indcollation;
@@ -624,6 +629,7 @@ UpdateIndexRelation(Oid indexoid,
values[Anum_pg_index_indexrelid - 1] = ObjectIdGetDatum(indexoid);
values[Anum_pg_index_indrelid - 1] = ObjectIdGetDatum(heapoid);
+ values[Anum_pg_index_indparentidx - 1] = ObjectIdGetDatum(parentIndexOid);
values[Anum_pg_index_indnatts - 1] = Int16GetDatum(indexInfo->ii_NumIndexAttrs);
values[Anum_pg_index_indisunique - 1] = BoolGetDatum(indexInfo->ii_Unique);
values[Anum_pg_index_indisprimary - 1] = BoolGetDatum(primary);
@@ -632,8 +638,7 @@ UpdateIndexRelation(Oid indexoid,
values[Anum_pg_index_indisclustered - 1] = BoolGetDatum(false);
values[Anum_pg_index_indisvalid - 1] = BoolGetDatum(isvalid);
values[Anum_pg_index_indcheckxmin - 1] = BoolGetDatum(false);
- /* we set isvalid and isready the same way */
- values[Anum_pg_index_indisready - 1] = BoolGetDatum(isvalid);
+ values[Anum_pg_index_indisready - 1] = BoolGetDatum(isready);
values[Anum_pg_index_indislive - 1] = BoolGetDatum(true);
values[Anum_pg_index_indisreplident - 1] = BoolGetDatum(false);
values[Anum_pg_index_indkey - 1] = PointerGetDatum(indkey);
@@ -670,6 +675,8 @@ UpdateIndexRelation(Oid indexoid,
* indexRelationId: normally, pass InvalidOid to let this routine
* generate an OID for the index. During bootstrap this may be
* nonzero to specify a preselected OID.
+ * parentIndexRelid: if creating an index partition, the OID of the
+ * parent index; otherwise InvalidOid.
* relFileNode: normally, pass InvalidOid to get new storage. May be
* nonzero to attach an existing valid build.
* indexInfo: same info executor uses to insert into the index
@@ -695,6 +702,8 @@ UpdateIndexRelation(Oid indexoid,
* INDEX_CREATE_IF_NOT_EXISTS:
* do not throw an error if a relation with the same name
* already exists.
+ * INDEX_CREATE_PARTITIONED:
+ * create a partitioned index (table must be partitioned)
* constr_flags: flags passed to index_constraint_create
* (only if INDEX_CREATE_ADD_CONSTRAINT is set)
* allow_system_table_mods: allow table to be a system catalog
@@ -706,6 +715,7 @@ Oid
index_create(Relation heapRelation,
const char *indexRelationName,
Oid indexRelationId,
+ Oid parentIndexRelid,
Oid relFileNode,
IndexInfo *indexInfo,
List *indexColNames,
@@ -731,12 +741,18 @@ index_create(Relation heapRelation,
int i;
char relpersistence;
bool isprimary = (flags & INDEX_CREATE_IS_PRIMARY) != 0;
+ bool invalid = (flags & INDEX_CREATE_INVALID) != 0;
bool concurrent = (flags & INDEX_CREATE_CONCURRENT) != 0;
+ bool partitioned = (flags & INDEX_CREATE_PARTITIONED) != 0;
+ char relkind;
/* constraint flags can only be set when a constraint is requested */
Assert((constr_flags == 0) ||
((flags & INDEX_CREATE_ADD_CONSTRAINT) != 0));
+ /* partitioned indexes must never be "built" by themselves */
+ Assert(!partitioned || (flags & INDEX_CREATE_SKIP_BUILD));
+ relkind = partitioned ? RELKIND_PARTITIONED_INDEX : RELKIND_INDEX;
is_exclusion = (indexInfo->ii_ExclusionOps != NULL);
pg_class = heap_open(RelationRelationId, RowExclusiveLock);
@@ -864,7 +880,7 @@ index_create(Relation heapRelation,
indexRelationId,
relFileNode,
indexTupDesc,
- RELKIND_INDEX,
+ relkind,
relpersistence,
shared_relation,
mapped_relation,
@@ -921,10 +937,12 @@ index_create(Relation heapRelation,
* (Or, could define a rule to maintain the predicate) --Nels, Feb '92
* ----------------
*/
- UpdateIndexRelation(indexRelationId, heapRelationId, indexInfo,
+ UpdateIndexRelation(indexRelationId, heapRelationId, parentIndexRelid,
+ indexInfo,
collationObjectId, classObjectId, coloptions,
isprimary, is_exclusion,
(constr_flags & INDEX_CONSTR_CREATE_DEFERRABLE) == 0,
+ !concurrent && !invalid,
!concurrent);
/*
@@ -978,6 +996,9 @@ index_create(Relation heapRelation,
else
{
bool have_simple_col = false;
+ DependencyType deptype;
+
+ deptype = OidIsValid(parentIndexRelid) ? DEPENDENCY_INTERNAL_AUTO : DEPENDENCY_AUTO;
/* Create auto dependencies on simply-referenced columns */
for (i = 0; i < indexInfo->ii_NumIndexAttrs; i++)
@@ -988,7 +1009,7 @@ index_create(Relation heapRelation,
referenced.objectId = heapRelationId;
referenced.objectSubId = indexInfo->ii_KeyAttrNumbers[i];
- recordDependencyOn(&myself, &referenced, DEPENDENCY_AUTO);
+ recordDependencyOn(&myself, &referenced, deptype);
have_simple_col = true;
}
@@ -1006,10 +1027,21 @@ index_create(Relation heapRelation,
referenced.objectId = heapRelationId;
referenced.objectSubId = 0;
- recordDependencyOn(&myself, &referenced, DEPENDENCY_AUTO);
+ recordDependencyOn(&myself, &referenced, deptype);
}
}
+ /* Store dependency on parent index, if any */
+ if (OidIsValid(parentIndexRelid))
+ {
+ referenced.classId = RelationRelationId;
+ referenced.objectId = parentIndexRelid;
+ referenced.objectSubId = 0;
+
+ recordDependencyOn(&myself, &referenced, DEPENDENCY_INTERNAL_AUTO);
+ }
+
+
/* Store dependency on collations */
/* The default collation is pinned, so don't bother recording it */
for (i = 0; i < indexInfo->ii_NumIndexAttrs; i++)
@@ -1555,9 +1587,10 @@ index_drop(Oid indexId, bool concurrent)
}
/*
- * Schedule physical removal of the files
+ * Schedule physical removal of the files (if any)
*/
- RelationDropStorage(userIndexRelation);
+ if (userIndexRelation->rd_rel->relkind != RELKIND_PARTITIONED_INDEX)
+ RelationDropStorage(userIndexRelation);
/*
* Close and flush the index's relcache entry, to ensure relcache doesn't
@@ -1694,12 +1727,65 @@ BuildIndexInfo(Relation index)
ii->ii_BrokenHotChain = false;
/* set up for possible use by index AM */
+ ii->ii_Am = index->rd_rel->relam;
ii->ii_AmCache = NULL;
ii->ii_Context = CurrentMemoryContext;
return ii;
}
+/*
+ * CompareIndexInfo
+ * Compare two IndexInfos, and return true if they are similar enough that
+ * an index built with one can pass as an index built with the other.
+ *
+ * attmap is an attribute map where info2 is input and info1 is output.
+ */
+bool
+CompareIndexInfo(IndexInfo *info1, IndexInfo *info2, AttrNumber *attmap)
+{
+ int i;
+
+ /* indexes are only equivalent if they have the same access method */
+ if (info1->ii_Am != info2->ii_Am)
+ return false;
+
+ /* and same number of attributes */
+ if (info1->ii_NumIndexAttrs != info2->ii_NumIndexAttrs)
+ return false;
+
+ /*
+ * and columns match through the attribute map (actual attribute numbers
+ * might differ!)
+ */
+ for (i = 0; i < info1->ii_NumIndexAttrs; i++)
+ {
+ if (attmap[info1->ii_KeyAttrNumbers[i] - 1] !=
+ info2->ii_KeyAttrNumbers[i])
+ return false;
+ }
+
+ /*
+ * Expression indexes are currently not considered equal. Not needed for
+ * current callers.
+ */
+ if (info1->ii_Expressions != NIL || info2->ii_Expressions != NIL)
+ return false;
+
+ /* Index predicates must be identical */
+ if (!equal(info1->ii_Predicate, info2->ii_Predicate))
+ return false;
+
+ /* No support currently for comparing exclusion indexes. */
+ if (info1->ii_ExclusionOps != NULL || info2->ii_ExclusionOps != NULL)
+ return false;
+
+ if (info1->ii_Unique != info2->ii_Unique)
+ return false;
+
+ return true;
+}
+
/* ----------------
* BuildSpeculativeIndexInfo
* Add extra state to IndexInfo record
@@ -1922,6 +2008,9 @@ index_update_stats(Relation rel,
elog(ERROR, "could not find tuple for relation %u", relid);
rd_rel = (Form_pg_class) GETSTRUCT(tuple);
+ /* Should this be a more comprehensive test? */
+ Assert(rd_rel->relkind != RELKIND_PARTITIONED_INDEX);
+
/* Apply required updates, if any, to copied tuple */
dirty = false;
@@ -3332,6 +3421,14 @@ reindex_index(Oid indexId, bool skip_constraint_checks, char persistence,
iRel = index_open(indexId, AccessExclusiveLock);
/*
+ * The case of reindexing partitioned tables and indexes is handled
+ * differently by upper layers, so this case shouldn't arise.
+ */
+ if (iRel->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
+ elog(ERROR, "unsupported relation kind for index \"%s\"",
+ RelationGetRelationName(iRel));
+
+ /*
* Don't allow reindex on temp tables of other backends ... their local
* buffer manager is not going to cope.
*/
@@ -3530,6 +3627,22 @@ reindex_relation(Oid relid, int flags, int options)
*/
rel = heap_open(relid, ShareLock);
+ /*
+ * This may be useful when implemented someday; but that day is not today.
+ * For now, avoid erroring out when called in a multi-table context
+ * (REINDEX SCHEMA) and happen to come across a partitioned table. The
+ * partitions may be reindexed on their own anyway.
+ */
+ if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
+ {
+ ereport(WARNING,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("REINDEX of partitioned tables is not yet implemented, skipping \"%s\"",
+ RelationGetRelationName(rel))));
+ heap_close(rel, ShareLock);
+ return false;
+ }
+
toast_relid = rel->rd_rel->reltoastrelid;
/*
diff --git a/src/backend/catalog/objectaddress.c b/src/backend/catalog/objectaddress.c
index bc999ca3c4..7576606c1b 100644
--- a/src/backend/catalog/objectaddress.c
+++ b/src/backend/catalog/objectaddress.c
@@ -1217,7 +1217,8 @@ get_relation_by_qualified_name(ObjectType objtype, List *object,
switch (objtype)
{
case OBJECT_INDEX:
- if (relation->rd_rel->relkind != RELKIND_INDEX)
+ if (relation->rd_rel->relkind != RELKIND_INDEX &&
+ relation->rd_rel->relkind != RELKIND_PARTITIONED_INDEX)
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is not an index",
@@ -3483,6 +3484,7 @@ getRelationDescription(StringInfo buffer, Oid relid)
relname);
break;
case RELKIND_INDEX:
+ case RELKIND_PARTITIONED_INDEX:
appendStringInfo(buffer, _("index %s"),
relname);
break;
@@ -3957,6 +3959,7 @@ getRelationTypeDescription(StringInfo buffer, Oid relid, int32 objectSubId)
appendStringInfoString(buffer, "table");
break;
case RELKIND_INDEX:
+ case RELKIND_PARTITIONED_INDEX:
appendStringInfoString(buffer, "index");
break;
case RELKIND_SEQUENCE:
diff --git a/src/backend/catalog/pg_depend.c b/src/backend/catalog/pg_depend.c
index 9dfbe123b5..2ea05f350b 100644
--- a/src/backend/catalog/pg_depend.c
+++ b/src/backend/catalog/pg_depend.c
@@ -656,14 +656,19 @@ get_constraint_index(Oid constraintId)
/*
* We assume any internal dependency of an index on the constraint
- * must be what we are looking for. (The relkind test is just
- * paranoia; there shouldn't be any such dependencies otherwise.)
+ * must be what we are looking for.
*/
if (deprec->classid == RelationRelationId &&
deprec->objsubid == 0 &&
- deprec->deptype == DEPENDENCY_INTERNAL &&
- get_rel_relkind(deprec->objid) == RELKIND_INDEX)
+ deprec->deptype == DEPENDENCY_INTERNAL)
{
+ char relkind = get_rel_relkind(deprec->objid);
+
+ /* This is pure paranoia; there shouldn't be any such */
+ if (relkind != RELKIND_INDEX &&
+ relkind != RELKIND_PARTITIONED_INDEX)
+ break;
+
indexId = deprec->objid;
break;
}
diff --git a/src/backend/catalog/toasting.c b/src/backend/catalog/toasting.c
index 0b4b5631a1..cf37011b73 100644
--- a/src/backend/catalog/toasting.c
+++ b/src/backend/catalog/toasting.c
@@ -315,6 +315,7 @@ create_toast_table(Relation rel, Oid toastOid, Oid toastIndexOid,
indexInfo->ii_ReadyForInserts = true;
indexInfo->ii_Concurrent = false;
indexInfo->ii_BrokenHotChain = false;
+ indexInfo->ii_Am = BTREE_AM_OID;
indexInfo->ii_AmCache = NULL;
indexInfo->ii_Context = CurrentMemoryContext;
@@ -328,6 +329,7 @@ create_toast_table(Relation rel, Oid toastOid, Oid toastIndexOid,
coloptions[1] = 0;
index_create(toast_rel, toast_idxname, toastIndexOid, InvalidOid,
+ InvalidOid,
indexInfo,
list_make2("chunk_id", "chunk_seq"),
BTREE_AM_OID,
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 9e6ba92008..c1cf83b045 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -23,7 +23,9 @@
#include "catalog/catalog.h"
#include "catalog/index.h"
#include "catalog/indexing.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits_fn.h"
#include "catalog/pg_opclass.h"
#include "catalog/pg_opfamily.h"
#include "catalog/pg_tablespace.h"
@@ -35,6 +37,7 @@
#include "commands/tablespace.h"
#include "mb/pg_wchar.h"
#include "miscadmin.h"
+#include "nodes/makefuncs.h"
#include "nodes/nodeFuncs.h"
#include "optimizer/clauses.h"
#include "optimizer/planner.h"
@@ -77,6 +80,7 @@ static char *ChooseIndexNameAddition(List *colnames);
static List *ChooseIndexColumnNames(List *indexElems);
static void RangeVarCallbackForReindexIndex(const RangeVar *relation,
Oid relId, Oid oldRelId, void *arg);
+static void ReindexPartitionedIndex(Relation parentIdx);
/*
* CheckIndexCompatible
@@ -183,6 +187,7 @@ CheckIndexCompatible(Oid oldId,
indexInfo->ii_ExclusionOps = NULL;
indexInfo->ii_ExclusionProcs = NULL;
indexInfo->ii_ExclusionStrats = NULL;
+ indexInfo->ii_Am = accessMethodId;
indexInfo->ii_AmCache = NULL;
indexInfo->ii_Context = CurrentMemoryContext;
typeObjectId = (Oid *) palloc(numberOfAttributes * sizeof(Oid));
@@ -292,14 +297,15 @@ CheckIndexCompatible(Oid oldId,
* 'stmt': IndexStmt describing the properties of the new index.
* 'indexRelationId': normally InvalidOid, but during bootstrap can be
* nonzero to specify a preselected OID for the index.
+ * 'parentIndexId': the OID of the parent index; InvalidOid if not the child
+ * of a partitioned index.
* 'is_alter_table': this is due to an ALTER rather than a CREATE operation.
* 'check_rights': check for CREATE rights in namespace and tablespace. (This
* should be true except when ALTER is deleting/recreating an index.)
* 'check_not_in_use': check for table not already in use in current session.
* This should be true unless caller is holding the table open, in which
* case the caller had better have checked it earlier.
- * 'skip_build': make the catalog entries but leave the index file empty;
- * it will be filled later.
+ * 'skip_build': make the catalog entries but don't create the index files
* 'quiet': suppress the NOTICE chatter ordinarily provided for constraints.
*
* Returns the object address of the created index.
@@ -308,6 +314,7 @@ ObjectAddress
DefineIndex(Oid relationId,
IndexStmt *stmt,
Oid indexRelationId,
+ Oid parentIndexId,
bool is_alter_table,
bool check_rights,
bool check_not_in_use,
@@ -330,6 +337,7 @@ DefineIndex(Oid relationId,
IndexAmRoutine *amRoutine;
bool amcanorder;
amoptions_function amoptions;
+ bool partitioned;
Datum reloptions;
int16 *coloptions;
IndexInfo *indexInfo;
@@ -382,23 +390,56 @@ DefineIndex(Oid relationId,
{
case RELKIND_RELATION:
case RELKIND_MATVIEW:
+ case RELKIND_PARTITIONED_TABLE:
/* OK */
break;
case RELKIND_FOREIGN_TABLE:
+ /*
+ * Custom error message for FOREIGN TABLE since the term is close
+ * to a regular table and can confuse the user.
+ */
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("cannot create index on foreign table \"%s\"",
RelationGetRelationName(rel))));
- case RELKIND_PARTITIONED_TABLE:
- ereport(ERROR,
- (errcode(ERRCODE_WRONG_OBJECT_TYPE),
- errmsg("cannot create index on partitioned table \"%s\"",
- RelationGetRelationName(rel))));
default:
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is not a table or materialized view",
RelationGetRelationName(rel))));
+ break;
+ }
+
+ /*
+ * Establish behavior for partitioned tables, and verify sanity of
+ * parameters.
+ *
+ * We do not build an actual index in this case; we only create a few
+ * catalog entries. The actual indexes are built by recursing for each
+ * partition.
+ */
+ partitioned = rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE;
+ if (partitioned)
+ {
+ if (stmt->concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("cannot create index on partitioned table \"%s\" concurrently",
+ RelationGetRelationName(rel))));
+ if (stmt->unique)
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("cannot create unique index on partitioned table \"%s\"",
+ RelationGetRelationName(rel))));
+ if (stmt->excludeOpNames)
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("cannot create exclusion constraints on partitioned table \"%s\"",
+ RelationGetRelationName(rel))));
+ if (stmt->primary || stmt->isconstraint)
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("cannot create constraints on partitioned tables")));
}
/*
@@ -574,6 +615,7 @@ DefineIndex(Oid relationId,
indexInfo->ii_ReadyForInserts = !stmt->concurrent;
indexInfo->ii_Concurrent = stmt->concurrent;
indexInfo->ii_BrokenHotChain = false;
+ indexInfo->ii_Am = accessMethodId;
indexInfo->ii_AmCache = NULL;
indexInfo->ii_Context = CurrentMemoryContext;
@@ -665,19 +707,24 @@ DefineIndex(Oid relationId,
/*
* Make the catalog entries for the index, including constraints. This
* step also actually builds the index, except if caller requested not to
- * or in concurrent mode, in which case it'll be done later.
+ * or in concurrent mode, in which case it'll be done later, or
+ * doing a partitioned index (because those don't have storage).
*/
flags = constr_flags = 0;
if (stmt->isconstraint)
flags |= INDEX_CREATE_ADD_CONSTRAINT;
- if (skip_build || stmt->concurrent)
+ if (skip_build || stmt->concurrent || partitioned)
flags |= INDEX_CREATE_SKIP_BUILD;
if (stmt->if_not_exists)
flags |= INDEX_CREATE_IF_NOT_EXISTS;
if (stmt->concurrent)
flags |= INDEX_CREATE_CONCURRENT;
+ if (partitioned)
+ flags |= INDEX_CREATE_PARTITIONED;
if (stmt->primary)
flags |= INDEX_CREATE_IS_PRIMARY;
+ if (partitioned && stmt->relation && !stmt->relation->inh)
+ flags |= INDEX_CREATE_INVALID;
if (stmt->deferrable)
constr_flags |= INDEX_CONSTR_CREATE_DEFERRABLE;
@@ -685,8 +732,8 @@ DefineIndex(Oid relationId,
constr_flags |= INDEX_CONSTR_CREATE_INIT_DEFERRED;
indexRelationId =
- index_create(rel, indexRelationName, indexRelationId, stmt->oldNode,
- indexInfo, indexColNames,
+ index_create(rel, indexRelationName, indexRelationId, parentIndexId,
+ stmt->oldNode, indexInfo, indexColNames,
accessMethodId, tablespaceId,
collationObjectId, classObjectId,
coloptions, reloptions,
@@ -706,6 +753,118 @@ DefineIndex(Oid relationId,
CreateComments(indexRelationId, RelationRelationId, 0,
stmt->idxcomment);
+ if (partitioned)
+ {
+ /*
+ * Unless caller specified to skip this step (via ONLY), process
+ * each partition to make sure they all contain a corresponding index.
+ *
+ * If we're called internally (no stmt->relation), recurse always.
+ */
+ if (!stmt->relation || stmt->relation->inh)
+ {
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ int nparts = partdesc->nparts;
+ Oid *part_oids;
+ TupleDesc parentDesc;
+
+ nparts = partdesc->nparts;
+ part_oids = palloc(sizeof(Oid) * nparts);
+ memcpy(part_oids, partdesc->oids, sizeof(Oid) * nparts);
+
+ parentDesc = CreateTupleDescCopy(RelationGetDescr(rel));
+
+ heap_close(rel, NoLock);
+
+ /*
+ * For each partition, scan all existing indexes; if one matches
+ * our index definition and is not already attached to some other
+ * parent index, attach it to the one we just created.
+ *
+ * If none matches, build a new index by calling ourselves
+ * recursively with the same options (except for the index name).
+ */
+ for (i = 0; i < nparts; i++)
+ {
+ Oid childRelid = part_oids[i];
+ Relation childrel;
+ List *childidxs;
+ ListCell *cell;
+ AttrNumber *attmap = NULL;
+ bool found = false;
+
+ childrel = heap_open(childRelid, lockmode);
+ childidxs = RelationGetIndexList(childrel);
+
+ foreach(cell, childidxs)
+ {
+ Oid cldidxid = lfirst_oid(cell);
+ Relation cldidx;
+ IndexInfo *cldIdxInfo;
+
+ cldidx = index_open(cldidxid, lockmode);
+
+ /* this index is already partition of another one */
+ if (cldidx->rd_index->indparentidx != 0)
+ {
+ index_close(cldidx, lockmode);
+ continue;
+ }
+
+ cldIdxInfo = BuildIndexInfo(cldidx);
+ if (attmap == NULL)
+ attmap =
+ convert_tuples_by_name_map(RelationGetDescr(childrel),
+ parentDesc,
+ gettext_noop("could not convert row type"));
+
+ if (CompareIndexInfo(cldIdxInfo, indexInfo, attmap))
+ {
+ /*
+ * Found a match. Attach index to parent and we're
+ * done, but keep lock till commit.
+ */
+ IndexSetParentIndex(cldidx, indexRelationId);
+ found = true;
+ index_close(cldidx, NoLock);
+ break;
+ }
+
+ index_close(cldidx, lockmode);
+ }
+
+ list_free(childidxs);
+ heap_close(childrel, NoLock);
+ if (attmap)
+ pfree(attmap);
+
+ /*
+ * If no matching index was found, create our own.
+ */
+ if (!found)
+ {
+ IndexStmt *childStmt = copyObject(stmt);
+
+ childStmt->idxname = NULL;
+ childStmt->relationId = childRelid;
+ DefineIndex(childRelid, childStmt,
+ InvalidOid, /* no predefined OID */
+ indexRelationId, /* this is our child */
+ false, check_rights, check_not_in_use,
+ false, quiet);
+ }
+ }
+ }
+ else
+ heap_close(rel, NoLock);
+
+ /*
+ * Indexes on partitioned tables are not themselves built, so we're
+ * done here.
+ */
+ return address;
+ }
+
if (!stmt->concurrent)
{
/* Close the heap and we're done, in the non-concurrent case */
@@ -1765,7 +1924,7 @@ ChooseIndexColumnNames(List *indexElems)
* ReindexIndex
* Recreate a specific index.
*/
-Oid
+void
ReindexIndex(RangeVar *indexRelation, int options)
{
Oid indOid;
@@ -1788,12 +1947,17 @@ ReindexIndex(RangeVar *indexRelation, int options)
* lock on the index.
*/
irel = index_open(indOid, NoLock);
+
+ if (irel->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
+ {
+ ReindexPartitionedIndex(irel);
+ return;
+ }
+
persistence = irel->rd_rel->relpersistence;
index_close(irel, NoLock);
reindex_index(indOid, false, persistence, options);
-
- return indOid;
}
/*
@@ -1832,7 +1996,8 @@ RangeVarCallbackForReindexIndex(const RangeVar *relation,
relkind = get_rel_relkind(relId);
if (!relkind)
return;
- if (relkind != RELKIND_INDEX)
+ if (relkind != RELKIND_INDEX &&
+ relkind != RELKIND_PARTITIONED_INDEX)
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is not an index", relation->relname)));
@@ -1976,6 +2141,12 @@ ReindexMultipleTables(const char *objectName, ReindexObjectType objectKind,
/*
* Only regular tables and matviews can have indexes, so ignore any
* other kind of relation.
+ *
+ * It is tempting to also consider partitioned tables here, but that
+ * has the problem that if the children are in the same schema, they
+ * would be processed twice. Maybe we could have a separate list of
+ * partitioned tables, and expand that afterwards into relids,
+ * ignoring any duplicates.
*/
if (classtuple->relkind != RELKIND_RELATION &&
classtuple->relkind != RELKIND_MATVIEW)
@@ -2038,3 +2209,75 @@ ReindexMultipleTables(const char *objectName, ReindexObjectType objectKind,
MemoryContextDelete(private_context);
}
+
+/*
+ * Reindex each child of a partitioned index.
+ *
+ * The parent index is given, locked in AccessExclusive mode; this routine
+ * obtains the list of children and releases the lock on parent before
+ * applying reindex on each child.
+ */
+static void
+ReindexPartitionedIndex(Relation parentIdx)
+{
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("REINDEX is not yet implemented for partitioned indexes")));
+}
+
+/*
+ * Update the pg_index tuple corresponding to the given index on a partition
+ * to indicate that the given index OID is now its parent partitioned index.
+ *
+ * This also corrects the pg_depend information for the affected index.
+ */
+void
+IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
+{
+ Relation pgindex;
+ HeapTuple indTup;
+ Form_pg_index indForm;
+ ObjectAddress partIdx;
+
+ /* Make sure this is an index */
+ Assert(partitionIdx->rd_rel->relkind == RELKIND_INDEX ||
+ partitionIdx->rd_rel->relkind == RELKIND_PARTITIONED_INDEX);
+
+ pgindex = heap_open(IndexRelationId, RowExclusiveLock);
+ indTup = partitionIdx->rd_indextuple;
+ indForm = (Form_pg_index) GETSTRUCT(indTup);
+ indForm->indparentidx = parentOid;
+
+ CatalogTupleUpdate(pgindex, &(indTup->t_self), indTup);
+
+ heap_close(pgindex, RowExclusiveLock);
+
+ ObjectAddressSet(partIdx,
+ RelationRelationId, RelationGetRelid(partitionIdx));
+
+ /*
+ * If setting a parent, add a pg_depend row; if making standalone, remove
+ * all existing rows.
+ */
+ if (OidIsValid(parentOid))
+ {
+ ObjectAddress parent;
+
+ ObjectAddressSet(parent, RelationRelationId, parentOid);
+ recordDependencyOn(&partIdx, &parent, DEPENDENCY_INTERNAL_AUTO);
+ }
+ else
+ {
+ ObjectAddress partitionTbl;
+
+ ObjectAddressSet(partitionTbl,
+ RelationRelationId, partitionIdx->rd_index->indrelid);
+
+ deleteDependencyRecordsForClass(RelationRelationId,
+ RelationGetRelid(partitionIdx),
+ RelationRelationId,
+ DEPENDENCY_INTERNAL_AUTO);
+
+ recordDependencyOn(&partIdx, &partitionTbl, DEPENDENCY_AUTO);
+ }
+}
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 62cf81e95a..4588d07fda 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -266,6 +266,12 @@ static const struct dropmsgstrings dropmsgstringarray[] = {
gettext_noop("table \"%s\" does not exist, skipping"),
gettext_noop("\"%s\" is not a table"),
gettext_noop("Use DROP TABLE to remove a table.")},
+ {RELKIND_PARTITIONED_INDEX,
+ ERRCODE_UNDEFINED_OBJECT,
+ gettext_noop("index \"%s\" does not exist"),
+ gettext_noop("index \"%s\" does not exist, skipping"),
+ gettext_noop("\"%s\" is not an index"),
+ gettext_noop("Use DROP INDEX to remove an index.")},
{'\0', 0, NULL, NULL, NULL, NULL}
};
@@ -481,6 +487,11 @@ static void ValidatePartitionConstraints(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
+ RangeVar *name);
+static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
+static void refuseDupeIndexAttach(Relation parentIdx, Relation partIdx,
+ Relation partitionTbl);
/* ----------------------------------------------------------------
@@ -898,6 +909,53 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
StorePartitionKey(rel, strategy, partnatts, partattrs, partexprs,
partopclass, partcollation);
+
+ /* make it all visible */
+ CommandCounterIncrement();
+ }
+
+ /*
+ * If we're creating a partition, create now all the indexes defined in
+ * the parent. We can't do it earlier, because DefineIndex wants to know
+ * the partition key which we just stored.
+ */
+ if (stmt->partbound)
+ {
+ Oid parentId = linitial_oid(inheritOids);
+ Relation parent;
+ List *idxlist;
+ ListCell *cell;
+
+ /* Already have strong enough lock on the parent */
+ parent = heap_open(parentId, NoLock);
+ idxlist = RelationGetIndexList(parent);
+
+ /*
+ * For each index in the parent table, create one in the partition
+ */
+ foreach(cell, idxlist)
+ {
+ Relation idxRel = index_open(lfirst_oid(cell), AccessShareLock);
+ AttrNumber *attmap;
+ IndexStmt *idxstmt;
+
+ attmap = convert_tuples_by_name_map(RelationGetDescr(rel),
+ RelationGetDescr(parent),
+ gettext_noop("could not convert row type"));
+ idxstmt =
+ generateClonedIndexStmt(NULL, RelationGetRelid(rel), idxRel,
+ attmap, RelationGetDescr(rel)->natts);
+ DefineIndex(RelationGetRelid(rel),
+ idxstmt,
+ InvalidOid,
+ RelationGetRelid(idxRel),
+ false, false, false, false, false);
+
+ index_close(idxRel, AccessShareLock);
+ }
+
+ list_free(idxlist);
+ heap_close(parent, NoLock);
}
/*
@@ -1180,10 +1238,13 @@ RangeVarCallbackForDropRelation(const RangeVar *rel, Oid relOid, Oid oldRelOid,
* but RemoveRelations() can only pass one relkind for a given relation.
* It chooses RELKIND_RELATION for both regular and partitioned tables.
* That means we must be careful before giving the wrong type error when
- * the relation is RELKIND_PARTITIONED_TABLE.
+ * the relation is RELKIND_PARTITIONED_TABLE. An equivalent problem
+ * exists with indexes.
*/
if (classform->relkind == RELKIND_PARTITIONED_TABLE)
expected_relkind = RELKIND_RELATION;
+ else if (classform->relkind == RELKIND_PARTITIONED_INDEX)
+ expected_relkind = RELKIND_INDEX;
else
expected_relkind = classform->relkind;
@@ -1211,7 +1272,8 @@ RangeVarCallbackForDropRelation(const RangeVar *rel, Oid relOid, Oid oldRelOid,
* we do it the other way around. No error if we don't find a pg_index
* entry, though --- the relation may have been dropped.
*/
- if (relkind == RELKIND_INDEX && relOid != oldRelOid)
+ if ((relkind == RELKIND_INDEX || relkind == RELKIND_PARTITIONED_INDEX) &&
+ relOid != oldRelOid)
{
state->heapOid = IndexGetRelation(relOid, true);
if (OidIsValid(state->heapOid))
@@ -2541,6 +2603,7 @@ renameatt_check(Oid myrelid, Form_pg_class classform, bool recursing)
relkind != RELKIND_MATVIEW &&
relkind != RELKIND_COMPOSITE_TYPE &&
relkind != RELKIND_INDEX &&
+ relkind != RELKIND_PARTITIONED_INDEX &&
relkind != RELKIND_FOREIGN_TABLE &&
relkind != RELKIND_PARTITIONED_TABLE)
ereport(ERROR,
@@ -3020,7 +3083,8 @@ RenameRelationInternal(Oid myrelid, const char *newrelname, bool is_internal)
/*
* Also rename the associated constraint, if any.
*/
- if (targetrelation->rd_rel->relkind == RELKIND_INDEX)
+ if (targetrelation->rd_rel->relkind == RELKIND_INDEX ||
+ targetrelation->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
{
Oid constraintId = get_index_constraint(myrelid);
@@ -3074,6 +3138,7 @@ CheckTableNotInUse(Relation rel, const char *stmt)
stmt, RelationGetRelationName(rel))));
if (rel->rd_rel->relkind != RELKIND_INDEX &&
+ rel->rd_rel->relkind != RELKIND_PARTITIONED_INDEX &&
AfterTriggerPendingOnRel(RelationGetRelid(rel)))
ereport(ERROR,
(errcode(ERRCODE_OBJECT_IN_USE),
@@ -3765,6 +3830,10 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
pass = AT_PASS_MISC;
break;
case AT_AttachPartition:
+ ATSimplePermissions(rel, ATT_TABLE | ATT_INDEX);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
case AT_DetachPartition:
ATSimplePermissions(rel, ATT_TABLE);
/* No command-specific prep needed */
@@ -4113,9 +4182,14 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab, Relation rel,
ATExecGenericOptions(rel, (List *) cmd->def);
break;
case AT_AttachPartition:
- ATExecAttachPartition(wqueue, rel, (PartitionCmd *) cmd->def);
+ if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
+ ATExecAttachPartition(wqueue, rel, (PartitionCmd *) cmd->def);
+ else
+ ATExecAttachPartitionIdx(wqueue, rel,
+ ((PartitionCmd *) cmd->def)->name);
break;
case AT_DetachPartition:
+ Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
@@ -4751,6 +4825,7 @@ ATSimplePermissions(Relation rel, int allowed_targets)
actual_target = ATT_MATVIEW;
break;
case RELKIND_INDEX:
+ case RELKIND_PARTITIONED_INDEX:
actual_target = ATT_INDEX;
break;
case RELKIND_COMPOSITE_TYPE:
@@ -6195,6 +6270,7 @@ ATPrepSetStatistics(Relation rel, const char *colName, int16 colNum, Node *newVa
if (rel->rd_rel->relkind != RELKIND_RELATION &&
rel->rd_rel->relkind != RELKIND_MATVIEW &&
rel->rd_rel->relkind != RELKIND_INDEX &&
+ rel->rd_rel->relkind != RELKIND_PARTITIONED_INDEX &&
rel->rd_rel->relkind != RELKIND_FOREIGN_TABLE &&
rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
ereport(ERROR,
@@ -6206,7 +6282,9 @@ ATPrepSetStatistics(Relation rel, const char *colName, int16 colNum, Node *newVa
* We allow referencing columns by numbers only for indexes, since table
* column numbers could contain gaps if columns are later dropped.
*/
- if (rel->rd_rel->relkind != RELKIND_INDEX && !colName)
+ if (rel->rd_rel->relkind != RELKIND_INDEX &&
+ rel->rd_rel->relkind != RELKIND_PARTITIONED_INDEX &&
+ !colName)
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot refer to non-index column by number")));
@@ -6284,7 +6362,8 @@ ATExecSetStatistics(Relation rel, const char *colName, int16 colNum, Node *newVa
errmsg("cannot alter system column \"%s\"",
colName)));
- if (rel->rd_rel->relkind == RELKIND_INDEX &&
+ if ((rel->rd_rel->relkind == RELKIND_INDEX ||
+ rel->rd_rel->relkind == RELKIND_PARTITIONED_INDEX) &&
rel->rd_index->indkey.values[attnum - 1] != 0)
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
@@ -6797,6 +6876,7 @@ ATExecAddIndex(AlteredTableInfo *tab, Relation rel,
address = DefineIndex(RelationGetRelid(rel),
stmt,
InvalidOid, /* no predefined OID */
+ InvalidOid, /* no parent index */
true, /* is_alter_table */
check_rights,
false, /* check_not_in_use - we did it already */
@@ -9198,7 +9278,8 @@ ATExecAlterColumnType(AlteredTableInfo *tab, Relation rel,
{
char relKind = get_rel_relkind(foundObject.objectId);
- if (relKind == RELKIND_INDEX)
+ if (relKind == RELKIND_INDEX ||
+ relKind == RELKIND_PARTITIONED_INDEX)
{
Assert(foundObject.objectSubId == 0);
if (!list_member_oid(tab->changedIndexOids, foundObject.objectId))
@@ -10041,6 +10122,15 @@ ATExecChangeOwner(Oid relationOid, Oid newOwnerId, bool recursing, LOCKMODE lock
newOwnerId = tuple_class->relowner;
}
break;
+ case RELKIND_PARTITIONED_INDEX:
+ if (recursing)
+ break;
+ ereport(ERROR,
+ (errcode(ERRCODE_WRONG_OBJECT_TYPE),
+ errmsg("cannot change owner of index \"%s\"",
+ NameStr(tuple_class->relname)),
+ errhint("Change the ownership of the index's table, instead.")));
+ break;
case RELKIND_SEQUENCE:
if (!recursing &&
tuple_class->relowner != newOwnerId)
@@ -10162,6 +10252,7 @@ ATExecChangeOwner(Oid relationOid, Oid newOwnerId, bool recursing, LOCKMODE lock
*/
if (tuple_class->relkind != RELKIND_COMPOSITE_TYPE &&
tuple_class->relkind != RELKIND_INDEX &&
+ tuple_class->relkind != RELKIND_PARTITIONED_INDEX &&
tuple_class->relkind != RELKIND_TOASTVALUE)
changeDependencyOnOwner(RelationRelationId, relationOid,
newOwnerId);
@@ -10169,7 +10260,8 @@ ATExecChangeOwner(Oid relationOid, Oid newOwnerId, bool recursing, LOCKMODE lock
/*
* Also change the ownership of the table's row type, if it has one
*/
- if (tuple_class->relkind != RELKIND_INDEX)
+ if (tuple_class->relkind != RELKIND_INDEX &&
+ tuple_class->relkind != RELKIND_PARTITIONED_INDEX)
AlterTypeOwnerInternal(tuple_class->reltype, newOwnerId);
/*
@@ -10178,6 +10270,7 @@ ATExecChangeOwner(Oid relationOid, Oid newOwnerId, bool recursing, LOCKMODE lock
* relation, as well as its toast table (if it has one).
*/
if (tuple_class->relkind == RELKIND_RELATION ||
+ tuple_class->relkind == RELKIND_PARTITIONED_TABLE ||
tuple_class->relkind == RELKIND_MATVIEW ||
tuple_class->relkind == RELKIND_TOASTVALUE)
{
@@ -10486,6 +10579,7 @@ ATExecSetRelOptions(Relation rel, List *defList, AlterTableType operation,
(void) view_reloptions(newOptions, true);
break;
case RELKIND_INDEX:
+ case RELKIND_PARTITIONED_INDEX:
(void) index_reloptions(rel->rd_amroutine->amoptions, newOptions, true);
break;
default:
@@ -10898,7 +10992,8 @@ AlterTableMoveAll(AlterTableMoveAllStmt *stmt)
relForm->relkind != RELKIND_RELATION &&
relForm->relkind != RELKIND_PARTITIONED_TABLE) ||
(stmt->objtype == OBJECT_INDEX &&
- relForm->relkind != RELKIND_INDEX) ||
+ relForm->relkind != RELKIND_INDEX &&
+ relForm->relkind != RELKIND_PARTITIONED_INDEX) ||
(stmt->objtype == OBJECT_MATVIEW &&
relForm->relkind != RELKIND_MATVIEW))
continue;
@@ -13285,7 +13380,8 @@ RangeVarCallbackForAlterRelation(const RangeVar *rv, Oid relid, Oid oldrelid,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is not a composite type", rv->relname)));
- if (reltype == OBJECT_INDEX && relkind != RELKIND_INDEX
+ if (reltype == OBJECT_INDEX && relkind != RELKIND_INDEX &&
+ relkind != RELKIND_PARTITIONED_INDEX
&& !IsA(stmt, RenameStmt))
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
@@ -14006,6 +14102,103 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
StorePartitionBound(attachrel, rel, cmd->bound);
/*
+ * Ensure a correct set of indexes in the partition. This either creates
+ * a new index in the table being attached, or re-parents an existing one.
+ */
+ {
+ AttrNumber *attmap = NULL;
+ List *idxes;
+ List *attachRelIdxs;
+ Relation *attachrelIdxRels;
+ IndexInfo **attachInfos;
+ int i;
+ ListCell *cell;
+
+ idxes = RelationGetIndexList(rel);
+ attachRelIdxs = RelationGetIndexList(attachrel);
+ attachrelIdxRels = palloc(sizeof(Relation) * list_length(attachRelIdxs));
+ attachInfos = palloc(sizeof(IndexInfo *) * list_length(attachRelIdxs));
+
+ /* Build arrays of all existing indexes and their IndexInfos */
+ i = 0;
+ foreach(cell, attachRelIdxs)
+ {
+ Oid cldIdxId = lfirst_oid(cell);
+
+ attachrelIdxRels[i] = index_open(cldIdxId, AccessShareLock);
+ attachInfos[i] = BuildIndexInfo(attachrelIdxRels[i]);
+ i++;
+ }
+
+ /*
+ * For each index on the partitioned table, find a matching one in the
+ * partition-to-be; if one is not found, create one.
+ */
+ foreach(cell, idxes)
+ {
+ Oid idx = lfirst_oid(cell);
+ Relation idxRel = index_open(idx, AccessShareLock);
+ IndexInfo *info;
+ bool found = false;
+
+ if (idxRel->rd_rel->relkind != RELKIND_PARTITIONED_INDEX)
+ {
+ index_close(idxRel, AccessShareLock);
+ continue;
+ }
+ info = BuildIndexInfo(idxRel);
+ if (attmap == NULL)
+ attmap =
+ convert_tuples_by_name_map(RelationGetDescr(attachrel),
+ RelationGetDescr(rel),
+ gettext_noop("could not convert row type"));
+
+ for (i = 0; i < list_length(attachRelIdxs); i++)
+ {
+ /* already used it */
+ if (attachrelIdxRels[i]->rd_index->indparentidx != 0)
+ continue;
+
+ if (CompareIndexInfo(info, attachInfos[i], attmap))
+ {
+ /* bingo. */
+ IndexSetParentIndex(attachrelIdxRels[i], idx);
+ found = true;
+ break;
+ }
+ }
+ if (!found)
+ {
+ IndexStmt *stmt;
+
+ stmt = generateClonedIndexStmt(NULL, RelationGetRelid(attachrel),
+ idxRel, attmap,
+ RelationGetDescr(rel)->natts);
+ DefineIndex(RelationGetRelid(attachrel), stmt, InvalidOid,
+ RelationGetRelid(idxRel),
+ false, false, false, false, false);
+ }
+
+ index_close(idxRel, AccessShareLock);
+ }
+
+ /* Clean up. */
+ if (attmap)
+ pfree(attmap);
+
+ for (i = 0; i < list_length(attachRelIdxs); i++)
+ {
+ pfree(attachInfos[i]);
+ index_close(attachrelIdxRels[i], AccessShareLock);
+ }
+
+ if (idxes)
+ pfree(idxes);
+ if (attachRelIdxs)
+ pfree(attachRelIdxs);
+ }
+
+ /*
* Generate partition constraint from the partition bound specification.
* If the parent itself is a partition, make sure to include its
* constraint as well.
@@ -14092,6 +14285,8 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
new_repl[Natts_pg_class];
ObjectAddress address;
Oid defaultPartOid;
+ List *indexes;
+ ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
@@ -14153,6 +14348,24 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
}
}
+ /* detach indexes too */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx = index_open(idxid, AccessExclusiveLock);
+
+ if (idx->rd_index->indparentidx != InvalidOid)
+ {
+ Assert(IndexGetRelation(idx->rd_index->indparentidx, false) ==
+ RelationGetRelid(rel));
+
+ IndexSetParentIndex(idx, InvalidOid);
+ }
+
+ relation_close(idx, AccessExclusiveLock);
+ }
+
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
@@ -14166,3 +14379,256 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
return address;
}
+
+/*
+ * Before acquiring lock on an index, acquire the same lock on the owning
+ * table.
+ */
+struct AttachIndexCallbackState
+{
+ Oid partitionOid;
+ Oid parentTblOid;
+ bool lockedParentTbl;
+};
+
+static void
+RangeVarCallbackForAttachIndex(const RangeVar *rv, Oid relOid, Oid oldRelOid,
+ void *arg)
+{
+ struct AttachIndexCallbackState *state;
+ Form_pg_class classform;
+ HeapTuple tuple;
+
+ state = (struct AttachIndexCallbackState *) arg;
+
+ if (!state->lockedParentTbl)
+ {
+ LockRelationOid(state->parentTblOid, AccessShareLock);
+ state->lockedParentTbl = true;
+ }
+
+ /*
+ * If we previously locked some other heap, and the name we're looking up
+ * no longer refers to an index on that relation, release the now-useless
+ * lock. XXX maybe we should do *after* we verify whether the index does
+ * not actually belong to the same relation ...
+ */
+ if (relOid != oldRelOid && OidIsValid(state->partitionOid))
+ {
+ UnlockRelationOid(state->partitionOid, AccessShareLock);
+ state->partitionOid = InvalidOid;
+ }
+
+ /* Didn't find a relation, so no need for locking or permission checks. */
+ if (!OidIsValid(relOid))
+ return;
+
+ tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(relOid));
+ if (!HeapTupleIsValid(tuple))
+ return; /* concurrently dropped, so nothing to do */
+ classform = (Form_pg_class) GETSTRUCT(tuple);
+ if (classform->relkind != RELKIND_PARTITIONED_INDEX &&
+ classform->relkind != RELKIND_INDEX)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_OBJECT_DEFINITION),
+ errmsg("\"%s\" is not an index", rv->relname)));
+ ReleaseSysCache(tuple);
+
+ /*
+ * Since we need only examine the heap's tupledesc, an access share lock
+ * on it (preventing any DDL) is sufficient.
+ */
+ state->partitionOid = IndexGetRelation(relOid, false);
+ LockRelationOid(state->partitionOid, AccessShareLock);
+}
+
+/*
+ * ALTER INDEX i1 ATTACH PARTITION i2
+ */
+static ObjectAddress
+ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
+{
+ Relation partIdx;
+ Relation partTbl;
+ Relation parentTbl;
+ ObjectAddress address;
+ Oid partIdxId;
+ struct AttachIndexCallbackState state;
+
+ /*
+ * We need to obtain lock on the index 'name' to modify it, but we also
+ * need to read its owning table's tuple descriptor -- so we need to lock
+ * both. To avoid deadlocks, obtain lock on the table before doing so on
+ * the index. Furthermore, we need to examine the parent table of the
+ * partition, so lock that one too.
+ */
+ state.partitionOid = InvalidOid;
+ state.parentTblOid = parentIdx->rd_index->indrelid;
+ state.lockedParentTbl = false;
+ partIdxId =
+ RangeVarGetRelidExtended(name, AccessExclusiveLock, false, false,
+ RangeVarCallbackForAttachIndex,
+ (void *) &state);
+ /* Not there? */
+ if (!OidIsValid(partIdxId))
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_OBJECT),
+ errmsg("index \"%s\" does not exist", name->relname)));
+
+ /* no deadlock risk: our callback above already acquired the lock */
+ partIdx = relation_open(partIdxId, AccessExclusiveLock);
+
+ /* we already hold lock on both tables, so this is safe: */
+ parentTbl = relation_open(parentIdx->rd_index->indrelid, AccessShareLock);
+ partTbl = relation_open(partIdx->rd_index->indrelid, NoLock);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partIdx));
+
+ /* Silently do nothing if already the right state */
+ if (partIdx->rd_index->indparentidx != RelationGetRelid(parentIdx))
+ {
+ IndexInfo *childInfo;
+ IndexInfo *parentInfo;
+ AttrNumber *attmap;
+ bool found;
+ int i;
+ PartitionDesc partDesc;
+
+ /*
+ * If this partition already has an index attached, refuse the operation.
+ */
+ refuseDupeIndexAttach(parentIdx, partIdx, partTbl);
+
+ if (OidIsValid(partIdx->rd_index->indparentidx))
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot attach index \"%s\" as a partition of index \"%s\"",
+ RelationGetRelationName(partIdx),
+ RelationGetRelationName(parentIdx)),
+ errdetail("Index \"%s\" is already attached to another index.",
+ RelationGetRelationName(partIdx))));
+
+ /* Make sure it indexes a partition of the other index's table */
+ partDesc = RelationGetPartitionDesc(parentTbl);
+ found = false;
+ for (i = 0; i < partDesc->nparts; i++)
+ {
+ if (partDesc->oids[i] == state.partitionOid)
+ {
+ found = true;
+ break;
+ }
+ }
+ if (!found)
+ ereport(ERROR,
+ (errmsg("cannot attach index \"%s\" as a partition of index \"%s\"",
+ RelationGetRelationName(partIdx),
+ RelationGetRelationName(parentIdx)),
+ errdetail("Index \"%s\" is not on a partition of table \"%s\".",
+ RelationGetRelationName(partIdx),
+ RelationGetRelationName(parentTbl))));
+
+ /* Ensure the indexes are compatible */
+ childInfo = BuildIndexInfo(partIdx);
+ parentInfo = BuildIndexInfo(parentIdx);
+ attmap = convert_tuples_by_name_map(RelationGetDescr(parentTbl),
+ RelationGetDescr(partTbl),
+ gettext_noop("could not convert row type"));
+ if (!CompareIndexInfo(parentInfo, childInfo, attmap))
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_OBJECT_DEFINITION),
+ errmsg("cannot attach index \"%s\" as a partition of index \"%s\"",
+ RelationGetRelationName(partIdx),
+ RelationGetRelationName(parentIdx)),
+ errdetail("The index definitions do not match.")));
+
+ /* All good -- do it */
+ IndexSetParentIndex(partIdx, RelationGetRelid(parentIdx));
+ pfree(attmap);
+
+ CommandCounterIncrement();
+
+ validatePartitionedIndex(parentIdx, parentTbl);
+ }
+
+ relation_close(parentTbl, AccessShareLock);
+ /* keep these locks till commit */
+ relation_close(partTbl, NoLock);
+ relation_close(partIdx, NoLock);
+
+ return address;
+}
+
+/*
+ * Verify whether the given partition already contains an index attached
+ * to the given partitioned index. If so, raise an error.
+ */
+static void
+refuseDupeIndexAttach(Relation parentIdx, Relation partIdx, Relation partitionTbl)
+{
+ Relation idxRel;
+ SysScanDesc scan;
+ ScanKeyData key[2];
+
+ idxRel = heap_open(IndexRelationId, AccessShareLock);
+ ScanKeyInit(&key[0], Anum_pg_index_indparentidx,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(parentIdx)));
+ ScanKeyInit(&key[1], Anum_pg_index_indrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(partitionTbl)));
+ scan = systable_beginscan(idxRel, IndexParentidxRelidIndexId, true,
+ NULL, 2, key);
+ if (systable_getnext(scan))
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot attach index \"%s\" as a partition of index \"%s\"",
+ RelationGetRelationName(partIdx),
+ RelationGetRelationName(parentIdx)),
+ errdetail("Another index is already attached for partition \"%s\".",
+ RelationGetRelationName(partitionTbl))));
+
+ systable_endscan(scan);
+ heap_close(idxRel, AccessShareLock);
+}
+
+/*
+ * Every time a partitioned index is attached a partition, verify whether the
+ * set is complete. If it is, mark it valid.
+ */
+static void
+validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
+{
+ Relation idxRel;
+ SysScanDesc scan;
+ ScanKeyData key;
+ int tuples = 0;
+ HeapTuple tuple;
+ PartitionDesc partDesc;
+
+ Assert(partedIdx->rd_rel->relkind == RELKIND_PARTITIONED_INDEX);
+
+ idxRel = heap_open(IndexRelationId, RowExclusiveLock);
+ ScanKeyInit(&key, Anum_pg_index_indparentidx,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(partedIdx)));
+ scan = systable_beginscan(idxRel, IndexParentidxRelidIndexId, true,
+ NULL, 1, &key);
+ while ((tuple = systable_getnext(scan)) != NULL)
+ tuples += 1;
+
+ partDesc = RelationGetPartitionDesc(partedTbl);
+ if (tuples == partDesc->nparts)
+ {
+ HeapTuple newtup;
+
+ newtup = heap_copytuple(partedIdx->rd_indextuple);
+ ((Form_pg_index) GETSTRUCT(newtup))->indisvalid = true;
+
+ CatalogTupleUpdate(idxRel, &partedIdx->rd_indextuple->t_self, newtup);
+ }
+
+ systable_endscan(scan);
+
+ heap_close(idxRel, RowExclusiveLock);
+}
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index ddbbc79823..65d8c77d7a 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -3379,6 +3379,7 @@ _copyIndexStmt(const IndexStmt *from)
COPY_STRING_FIELD(idxname);
COPY_NODE_FIELD(relation);
+ COPY_SCALAR_FIELD(relationId);
COPY_STRING_FIELD(accessMethod);
COPY_STRING_FIELD(tableSpace);
COPY_NODE_FIELD(indexParams);
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index 30ccc9c5ae..0bd12e862e 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -1332,6 +1332,7 @@ _equalIndexStmt(const IndexStmt *a, const IndexStmt *b)
{
COMPARE_STRING_FIELD(idxname);
COMPARE_NODE_FIELD(relation);
+ COMPARE_SCALAR_FIELD(relationId);
COMPARE_STRING_FIELD(accessMethod);
COMPARE_STRING_FIELD(tableSpace);
COMPARE_NODE_FIELD(indexParams);
diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index 5e72df137e..b1cdfc36a6 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -2650,6 +2650,7 @@ _outIndexStmt(StringInfo str, const IndexStmt *node)
WRITE_STRING_FIELD(idxname);
WRITE_NODE_FIELD(relation);
+ WRITE_OID_FIELD(relationId);
WRITE_STRING_FIELD(accessMethod);
WRITE_STRING_FIELD(tableSpace);
WRITE_NODE_FIELD(indexParams);
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 8c60b35068..5f263395c5 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -372,7 +372,12 @@ get_relation_info(PlannerInfo *root, Oid relationObjectId, bool inhparent,
* a table, except we can be sure that the index is not larger
* than the table.
*/
- if (info->indpred == NIL)
+ if (indexRelation->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
+ {
+ info->pages = 0;
+ info->tuples = 0;
+ }
+ else if (info->indpred == NIL)
{
info->pages = RelationGetNumberOfBlocks(indexRelation);
info->tuples = rel->tuples;
@@ -387,7 +392,8 @@ get_relation_info(PlannerInfo *root, Oid relationObjectId, bool inhparent,
info->tuples = rel->tuples;
}
- if (info->relam == BTREE_AM_OID)
+ if (indexRelation->rd_rel->relkind == RELKIND_INDEX &&
+ info->relam == BTREE_AM_OID)
{
/* For btrees, get tree height while we have the index open */
info->tree_height = _bt_getrootheight(indexRelation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 16923e853a..04ef741a4c 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -290,7 +290,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
%type <ival> add_drop opt_asc_desc opt_nulls_order
%type <node> alter_table_cmd alter_type_cmd opt_collate_clause
- replica_identity partition_cmd
+ replica_identity partition_cmd index_partition_cmd
%type <list> alter_table_cmds alter_type_cmds
%type <list> alter_identity_column_option_list
%type <defelt> alter_identity_column_option
@@ -1891,6 +1891,15 @@ AlterTableStmt:
n->missing_ok = true;
$$ = (Node *)n;
}
+ | ALTER INDEX qualified_name index_partition_cmd
+ {
+ AlterTableStmt *n = makeNode(AlterTableStmt);
+ n->relation = $3;
+ n->cmds = list_make1($4);
+ n->relkind = OBJECT_INDEX;
+ n->missing_ok = false;
+ $$ = (Node *)n;
+ }
| ALTER INDEX ALL IN_P TABLESPACE name SET TABLESPACE name opt_nowait
{
AlterTableMoveAllStmt *n =
@@ -2025,6 +2034,22 @@ partition_cmd:
}
;
+index_partition_cmd:
+ /* ALTER INDEX <name> ATTACH PARTITION <index_name> */
+ ATTACH PARTITION qualified_name
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_AttachPartition;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ n->def = (Node *) cmd;
+
+ $$ = (Node *) n;
+ }
+ ;
+
alter_table_cmd:
/* ALTER TABLE <name> ADD <coldef> */
ADD_P columnDef
@@ -7330,7 +7355,7 @@ defacl_privilege_target:
*****************************************************************************/
IndexStmt: CREATE opt_unique INDEX opt_concurrently opt_index_name
- ON qualified_name access_method_clause '(' index_params ')'
+ ON relation_expr access_method_clause '(' index_params ')'
opt_reloptions OptTableSpace where_clause
{
IndexStmt *n = makeNode(IndexStmt);
@@ -7338,6 +7363,7 @@ IndexStmt: CREATE opt_unique INDEX opt_concurrently opt_index_name
n->concurrent = $4;
n->idxname = $5;
n->relation = $7;
+ n->relationId = InvalidOid;
n->accessMethod = $8;
n->indexParams = $10;
n->options = $12;
@@ -7356,7 +7382,7 @@ IndexStmt: CREATE opt_unique INDEX opt_concurrently opt_index_name
$$ = (Node *)n;
}
| CREATE opt_unique INDEX opt_concurrently IF_P NOT EXISTS index_name
- ON qualified_name access_method_clause '(' index_params ')'
+ ON relation_expr access_method_clause '(' index_params ')'
opt_reloptions OptTableSpace where_clause
{
IndexStmt *n = makeNode(IndexStmt);
diff --git a/src/backend/parser/parse_utilcmd.c b/src/backend/parser/parse_utilcmd.c
index 128f1679c6..90bb356df8 100644
--- a/src/backend/parser/parse_utilcmd.c
+++ b/src/backend/parser/parse_utilcmd.c
@@ -118,9 +118,6 @@ static void transformTableLikeClause(CreateStmtContext *cxt,
TableLikeClause *table_like_clause);
static void transformOfType(CreateStmtContext *cxt,
TypeName *ofTypename);
-static IndexStmt *generateClonedIndexStmt(CreateStmtContext *cxt,
- Relation source_idx,
- const AttrNumber *attmap, int attmap_length);
static List *get_collation(Oid collation, Oid actual_datatype);
static List *get_opclass(Oid opclass, Oid actual_datatype);
static void transformIndexConstraints(CreateStmtContext *cxt);
@@ -1185,7 +1182,8 @@ transformTableLikeClause(CreateStmtContext *cxt, TableLikeClause *table_like_cla
parent_index = index_open(parent_index_oid, AccessShareLock);
/* Build CREATE INDEX statement to recreate the parent_index */
- index_stmt = generateClonedIndexStmt(cxt, parent_index,
+ index_stmt = generateClonedIndexStmt(cxt->relation, InvalidOid,
+ parent_index,
attmap, tupleDesc->natts);
/* Copy comment on index, if requested */
@@ -1263,10 +1261,12 @@ transformOfType(CreateStmtContext *cxt, TypeName *ofTypename)
/*
* Generate an IndexStmt node using information from an already existing index
- * "source_idx". Attribute numbers should be adjusted according to attmap.
+ * "source_idx", for the rel identified either by heapRel or heapRelid.
+ *
+ * Attribute numbers should be adjusted according to attmap.
*/
-static IndexStmt *
-generateClonedIndexStmt(CreateStmtContext *cxt, Relation source_idx,
+IndexStmt *
+generateClonedIndexStmt(RangeVar *heapRel, Oid heapRelid, Relation source_idx,
const AttrNumber *attmap, int attmap_length)
{
Oid source_relid = RelationGetRelid(source_idx);
@@ -1287,6 +1287,9 @@ generateClonedIndexStmt(CreateStmtContext *cxt, Relation source_idx,
Datum datum;
bool isnull;
+ Assert((heapRel == NULL && OidIsValid(heapRelid)) ||
+ (heapRel != NULL && !OidIsValid(heapRelid)));
+
/*
* Fetch pg_class tuple of source index. We can't use the copy in the
* relcache entry because it doesn't include optional fields.
@@ -1322,7 +1325,8 @@ generateClonedIndexStmt(CreateStmtContext *cxt, Relation source_idx,
/* Begin building the IndexStmt */
index = makeNode(IndexStmt);
- index->relation = cxt->relation;
+ index->relation = heapRel;
+ index->relationId = heapRelid;
index->accessMethod = pstrdup(NameStr(amrec->amname));
if (OidIsValid(idxrelrec->reltablespace))
index->tableSpace = get_tablespace_name(idxrelrec->reltablespace);
@@ -3289,18 +3293,39 @@ transformPartitionCmd(CreateStmtContext *cxt, PartitionCmd *cmd)
{
Relation parentRel = cxt->rel;
- /* the table must be partitioned */
- if (parentRel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_OBJECT_DEFINITION),
- errmsg("\"%s\" is not partitioned",
- RelationGetRelationName(parentRel))));
-
- /* transform the partition bound, if any */
- Assert(RelationGetPartitionKey(parentRel) != NULL);
- if (cmd->bound != NULL)
- cxt->partbound = transformPartitionBound(cxt->pstate, parentRel,
- cmd->bound);
+ switch (parentRel->rd_rel->relkind)
+ {
+ case RELKIND_PARTITIONED_TABLE:
+ /* transform the partition bound, if any */
+ Assert(RelationGetPartitionKey(parentRel) != NULL);
+ if (cmd->bound != NULL)
+ cxt->partbound = transformPartitionBound(cxt->pstate, parentRel,
+ cmd->bound);
+ break;
+ case RELKIND_PARTITIONED_INDEX:
+ /* nothing to check */
+ Assert(cmd->bound == NULL);
+ break;
+ case RELKIND_RELATION:
+ /* the table must be partitioned */
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_OBJECT_DEFINITION),
+ errmsg("table \"%s\" is not partitioned",
+ RelationGetRelationName(parentRel))));
+ break;
+ case RELKIND_INDEX:
+ /* the index must be partitioned */
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_OBJECT_DEFINITION),
+ errmsg("index \"%s\" is not partitioned",
+ RelationGetRelationName(parentRel))));
+ break;
+ default:
+ /* parser shouldn't let this case through */
+ elog(ERROR, "\"%s\" is not a partitioned table or index",
+ RelationGetRelationName(parentRel));
+ break;
+ }
}
/*
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index ec98a612ec..16c4f8fad4 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -23,6 +23,7 @@
#include "access/xlog.h"
#include "catalog/catalog.h"
#include "catalog/namespace.h"
+#include "catalog/pg_inherits_fn.h"
#include "catalog/toasting.h"
#include "commands/alter.h"
#include "commands/async.h"
@@ -1300,6 +1301,7 @@ ProcessUtilitySlow(ParseState *pstate,
IndexStmt *stmt = (IndexStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ List *inheritors = NIL;
if (stmt->concurrent)
PreventTransactionChain(isTopLevel,
@@ -1321,6 +1323,9 @@ ProcessUtilitySlow(ParseState *pstate,
false, false,
RangeVarCallbackOwnsRelation,
NULL);
+ /* Also, lock any descendant tables if recursive */
+ if (stmt->relation->inh)
+ inheritors = find_all_inheritors(relid, lockmode, NULL);
/* Run parse analysis ... */
stmt = transformIndexStmt(relid, stmt, queryString);
@@ -1331,6 +1336,7 @@ ProcessUtilitySlow(ParseState *pstate,
DefineIndex(relid, /* OID of heap relation */
stmt,
InvalidOid, /* no predefined OID */
+ InvalidOid, /* no parent index */
false, /* is_alter_table */
true, /* check_rights */
true, /* check_not_in_use */
@@ -1346,6 +1352,9 @@ ProcessUtilitySlow(ParseState *pstate,
parsetree);
commandCollected = true;
EventTriggerAlterTableEnd();
+
+ if (inheritors)
+ list_free(inheritors);
}
break;
diff --git a/src/backend/utils/adt/amutils.c b/src/backend/utils/adt/amutils.c
index a6d8feea5b..0f7ceb62eb 100644
--- a/src/backend/utils/adt/amutils.c
+++ b/src/backend/utils/adt/amutils.c
@@ -183,7 +183,8 @@ indexam_property(FunctionCallInfo fcinfo,
if (!HeapTupleIsValid(tuple))
PG_RETURN_NULL();
rd_rel = (Form_pg_class) GETSTRUCT(tuple);
- if (rd_rel->relkind != RELKIND_INDEX)
+ if (rd_rel->relkind != RELKIND_INDEX &&
+ rd_rel->relkind != RELKIND_PARTITIONED_INDEX)
{
ReleaseSysCache(tuple);
PG_RETURN_NULL();
diff --git a/src/backend/utils/adt/ruleutils.c b/src/backend/utils/adt/ruleutils.c
index 9cdbb06add..c5f5a1ca3f 100644
--- a/src/backend/utils/adt/ruleutils.c
+++ b/src/backend/utils/adt/ruleutils.c
@@ -317,7 +317,7 @@ static void decompile_column_index_array(Datum column_index_array, Oid relId,
static char *pg_get_ruledef_worker(Oid ruleoid, int prettyFlags);
static char *pg_get_indexdef_worker(Oid indexrelid, int colno,
const Oid *excludeOps,
- bool attrsOnly, bool showTblSpc,
+ bool attrsOnly, bool showTblSpc, bool inherits,
int prettyFlags, bool missing_ok);
static char *pg_get_statisticsobj_worker(Oid statextid, bool missing_ok);
static char *pg_get_partkeydef_worker(Oid relid, int prettyFlags,
@@ -1086,7 +1086,7 @@ pg_get_indexdef(PG_FUNCTION_ARGS)
prettyFlags = PRETTYFLAG_INDENT;
- res = pg_get_indexdef_worker(indexrelid, 0, NULL, false, false,
+ res = pg_get_indexdef_worker(indexrelid, 0, NULL, false, false, false,
prettyFlags, true);
if (res == NULL)
@@ -1107,7 +1107,7 @@ pg_get_indexdef_ext(PG_FUNCTION_ARGS)
prettyFlags = pretty ? PRETTYFLAG_PAREN | PRETTYFLAG_INDENT : PRETTYFLAG_INDENT;
res = pg_get_indexdef_worker(indexrelid, colno, NULL, colno != 0, false,
- prettyFlags, true);
+ false, prettyFlags, true);
if (res == NULL)
PG_RETURN_NULL();
@@ -1123,7 +1123,7 @@ pg_get_indexdef_ext(PG_FUNCTION_ARGS)
char *
pg_get_indexdef_string(Oid indexrelid)
{
- return pg_get_indexdef_worker(indexrelid, 0, NULL, false, true, 0, false);
+ return pg_get_indexdef_worker(indexrelid, 0, NULL, false, true, true, 0, false);
}
/* Internal version that just reports the column definitions */
@@ -1133,7 +1133,7 @@ pg_get_indexdef_columns(Oid indexrelid, bool pretty)
int prettyFlags;
prettyFlags = pretty ? PRETTYFLAG_PAREN | PRETTYFLAG_INDENT : PRETTYFLAG_INDENT;
- return pg_get_indexdef_worker(indexrelid, 0, NULL, true, false,
+ return pg_get_indexdef_worker(indexrelid, 0, NULL, true, false, false,
prettyFlags, false);
}
@@ -1146,7 +1146,7 @@ pg_get_indexdef_columns(Oid indexrelid, bool pretty)
static char *
pg_get_indexdef_worker(Oid indexrelid, int colno,
const Oid *excludeOps,
- bool attrsOnly, bool showTblSpc,
+ bool attrsOnly, bool showTblSpc, bool inherits,
int prettyFlags, bool missing_ok)
{
/* might want a separate isConstraint parameter later */
@@ -1259,9 +1259,11 @@ pg_get_indexdef_worker(Oid indexrelid, int colno,
if (!attrsOnly)
{
if (!isConstraint)
- appendStringInfo(&buf, "CREATE %sINDEX %s ON %s USING %s (",
+ appendStringInfo(&buf, "CREATE %sINDEX %s ON %s%s USING %s (",
idxrec->indisunique ? "UNIQUE " : "",
quote_identifier(NameStr(idxrelrec->relname)),
+ idxrelrec->relkind == RELKIND_PARTITIONED_INDEX
+ && !inherits ? "ONLY " : "",
generate_relation_name(indrelid, NIL),
quote_identifier(NameStr(amrec->amname)));
else /* currently, must be EXCLUDE constraint */
@@ -2148,6 +2150,7 @@ pg_get_constraintdef_worker(Oid constraintId, bool fullCommand,
operators,
false,
false,
+ false,
prettyFlags,
false));
break;
diff --git a/src/backend/utils/cache/relcache.c b/src/backend/utils/cache/relcache.c
index 28a4483434..a28046f857 100644
--- a/src/backend/utils/cache/relcache.c
+++ b/src/backend/utils/cache/relcache.c
@@ -430,6 +430,7 @@ static void
RelationParseRelOptions(Relation relation, HeapTuple tuple)
{
bytea *options;
+ bool isindex;
relation->rd_options = NULL;
@@ -439,6 +440,7 @@ RelationParseRelOptions(Relation relation, HeapTuple tuple)
case RELKIND_RELATION:
case RELKIND_TOASTVALUE:
case RELKIND_INDEX:
+ case RELKIND_PARTITIONED_INDEX:
case RELKIND_VIEW:
case RELKIND_MATVIEW:
case RELKIND_PARTITIONED_TABLE:
@@ -452,10 +454,12 @@ RelationParseRelOptions(Relation relation, HeapTuple tuple)
* we might not have any other for pg_class yet (consider executing this
* code for pg_class itself)
*/
+ isindex = relation->rd_rel->relkind == RELKIND_INDEX ||
+ relation->rd_rel->relkind == RELKIND_PARTITIONED_INDEX;
options = extractRelOptions(tuple,
GetPgClassDescriptor(),
- relation->rd_rel->relkind == RELKIND_INDEX ?
- relation->rd_amroutine->amoptions : NULL);
+ isindex ? relation->rd_amroutine->amoptions :
+ NULL);
/*
* Copy parsed data into CacheMemoryContext. To guard against the
@@ -2053,7 +2057,8 @@ RelationIdGetRelation(Oid relationId)
* and we don't want to use the full-blown procedure because it's
* a headache for indexes that reload itself depends on.
*/
- if (rd->rd_rel->relkind == RELKIND_INDEX)
+ if (rd->rd_rel->relkind == RELKIND_INDEX ||
+ rd->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
RelationReloadIndexInfo(rd);
else
RelationClearRelation(rd, true);
@@ -2167,7 +2172,8 @@ RelationReloadIndexInfo(Relation relation)
Form_pg_class relp;
/* Should be called only for invalidated indexes */
- Assert(relation->rd_rel->relkind == RELKIND_INDEX &&
+ Assert((relation->rd_rel->relkind == RELKIND_INDEX ||
+ relation->rd_rel->relkind == RELKIND_PARTITIONED_INDEX) &&
!relation->rd_isvalid);
/* Ensure it's closed at smgr level */
@@ -2387,7 +2393,8 @@ RelationClearRelation(Relation relation, bool rebuild)
{
RelationInitPhysicalAddr(relation);
- if (relation->rd_rel->relkind == RELKIND_INDEX)
+ if (relation->rd_rel->relkind == RELKIND_INDEX ||
+ relation->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
{
relation->rd_isvalid = false; /* needs to be revalidated */
if (relation->rd_refcnt > 1 && IsTransactionState())
@@ -2403,7 +2410,8 @@ RelationClearRelation(Relation relation, bool rebuild)
* re-read the pg_class row to handle possible physical relocation of the
* index, and we check for pg_index updates too.
*/
- if (relation->rd_rel->relkind == RELKIND_INDEX &&
+ if ((relation->rd_rel->relkind == RELKIND_INDEX ||
+ relation->rd_rel->relkind == RELKIND_PARTITIONED_INDEX) &&
relation->rd_refcnt > 0 &&
relation->rd_indexcxt != NULL)
{
@@ -5461,7 +5469,10 @@ load_relcache_init_file(bool shared)
rel->rd_att->constr = constr;
}
- /* If it's an index, there's more to do */
+ /*
+ * If it's an index, there's more to do. Note we explicitly ignore
+ * partitioned indexes here.
+ */
if (rel->rd_rel->relkind == RELKIND_INDEX)
{
MemoryContext indexcxt;
@@ -5825,7 +5836,10 @@ write_relcache_init_file(bool shared)
(rel->rd_options ? VARSIZE(rel->rd_options) : 0),
fp);
- /* If it's an index, there's more to do */
+ /*
+ * If it's an index, there's more to do. Note we explicitly ignore
+ * partitioned indexes here.
+ */
if (rel->rd_rel->relkind == RELKIND_INDEX)
{
/* write the pg_index tuple */
diff --git a/src/bin/pg_dump/common.c b/src/bin/pg_dump/common.c
index 7f5f351486..778c127c8e 100644
--- a/src/bin/pg_dump/common.c
+++ b/src/bin/pg_dump/common.c
@@ -68,6 +68,7 @@ static int numextmembers;
static void flagInhTables(Archive *fout, TableInfo *tbinfo, int numTables,
InhInfo *inhinfo, int numInherits);
+static void flagInhIndexes(Archive *fout, TableInfo *tblinfo, int numTables);
static void flagInhAttrs(DumpOptions *dopt, TableInfo *tblinfo, int numTables);
static DumpableObject **buildIndexArray(void *objArray, int numObjs,
Size objSize);
@@ -76,6 +77,8 @@ static int ExtensionMemberIdCompare(const void *p1, const void *p2);
static void findParentsByOid(TableInfo *self,
InhInfo *inhinfo, int numInherits);
static int strInArray(const char *pattern, char **arr, int arr_size);
+static IndxInfo *findIndexByOid(Oid oid, DumpableObject **idxinfoindex,
+ int numIndexes);
/*
@@ -258,6 +261,10 @@ getSchemaData(Archive *fout, int *numTablesPtr)
getIndexes(fout, tblinfo, numTables);
if (g_verbose)
+ write_msg(NULL, "flagging indexes in partitioned tables\n");
+ flagInhIndexes(fout, tblinfo, numTables);
+
+ if (g_verbose)
write_msg(NULL, "reading extended statistics\n");
getExtendedStatistics(fout, tblinfo, numTables);
@@ -354,6 +361,91 @@ flagInhTables(Archive *fout, TableInfo *tblinfo, int numTables,
}
}
+/*
+ * flagInhIndexes -
+ * Create AttachIndexInfo objects for partitioned indexes, and add
+ * appropriate dependency links.
+ */
+static void
+flagInhIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
+{
+ int i,
+ j,
+ k;
+ DumpableObject ***parentIndexArray;
+
+ parentIndexArray = (DumpableObject ***)
+ pg_malloc0(getMaxDumpId() * sizeof(DumpableObject **));
+
+ for (i = 0; i < numTables; i++)
+ {
+ TableInfo *parenttbl;
+ IndexAttachInfo *attachinfo;
+
+ if (!tblinfo[i].ispartition || tblinfo[i].numParents == 0)
+ continue;
+
+ Assert(tblinfo[i].numParents == 1);
+ parenttbl = tblinfo[i].parents[0];
+
+ /*
+ * We need access to each parent table's index list, but there is no
+ * index to cover them outside of this function. To avoid having to
+ * sort every parent table's indexes each time we come across each of
+ * its partitions, create an indexed array for each parent the first
+ * time it is required.
+ */
+ if (parentIndexArray[parenttbl->dobj.dumpId] == NULL)
+ parentIndexArray[parenttbl->dobj.dumpId] =
+ buildIndexArray(parenttbl->indexes,
+ parenttbl->numIndexes,
+ sizeof(IndxInfo));
+
+ attachinfo = (IndexAttachInfo *)
+ pg_malloc0(tblinfo[i].numIndexes * sizeof(IndexAttachInfo));
+ for (j = 0, k = 0; j < tblinfo[i].numIndexes; j++)
+ {
+ IndxInfo *index = &(tblinfo[i].indexes[j]);
+ IndxInfo *parentidx;
+
+ if (index->indparentidx == 0)
+ continue;
+
+ parentidx = findIndexByOid(index->indparentidx,
+ parentIndexArray[parenttbl->dobj.dumpId],
+ parenttbl->numIndexes);
+ if (parentidx == NULL)
+ continue;
+
+ attachinfo[k].dobj.objType = DO_INDEX_ATTACH;
+ attachinfo[k].dobj.catId.tableoid = 0;
+ attachinfo[k].dobj.catId.oid = 0;
+ AssignDumpId(&attachinfo[k].dobj);
+ attachinfo[k].dobj.name = pg_strdup(index->dobj.name);
+ attachinfo[k].parentIdx = parentidx;
+ attachinfo[k].partitionIdx = index;
+
+ /*
+ * Normal dependency goes from child idx to parent idx; remove it
+ * to avoid creating a cycle only to resolve it later. We want
+ * one from parent to partition (so that the partition index is
+ * created first), and another one from attach object to parent
+ * (so that the partition index is attached once the parent index
+ * has been created).
+ */
+ addObjectDependency(&parentidx->dobj, index->dobj.dumpId);
+ addObjectDependency(&attachinfo[k].dobj, parentidx->dobj.dumpId);
+
+ k++;
+ }
+ }
+
+ for (i = 0; i < numTables; i++)
+ if (parentIndexArray[i])
+ pg_free(parentIndexArray[i]);
+ pg_free(parentIndexArray);
+}
+
/* flagInhAttrs -
* for each dumpable table in tblinfo, flag its inherited attributes
*
@@ -827,6 +919,18 @@ findExtensionByOid(Oid oid)
return (ExtensionInfo *) findObjectByOid(oid, extinfoindex, numExtensions);
}
+/*
+ * findIndexByOid
+ * find the entry of the index with the given oid
+ *
+ * This one's signature is different from the previous ones because we lack a
+ * global array of all indexes, so caller must pass their array as argument.
+ */
+static IndxInfo *
+findIndexByOid(Oid oid, DumpableObject **idxinfoindex, int numIndexes)
+{
+ return (IndxInfo *) findObjectByOid(oid, idxinfoindex, numIndexes);
+}
/*
* setExtensionMembership
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index 27628a397c..28245a741e 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -193,6 +193,7 @@ static void dumpAttrDef(Archive *fout, AttrDefInfo *adinfo);
static void dumpSequence(Archive *fout, TableInfo *tbinfo);
static void dumpSequenceData(Archive *fout, TableDataInfo *tdinfo);
static void dumpIndex(Archive *fout, IndxInfo *indxinfo);
+static void dumpIndexAttach(Archive *fout, IndexAttachInfo *attachinfo);
static void dumpStatisticsExt(Archive *fout, StatsExtInfo *statsextinfo);
static void dumpConstraint(Archive *fout, ConstraintInfo *coninfo);
static void dumpTableConstraintComment(Archive *fout, ConstraintInfo *coninfo);
@@ -6509,6 +6510,7 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
int i_tableoid,
i_oid,
i_indexname,
+ i_parentidx,
i_indexdef,
i_indnkeys,
i_indkey,
@@ -6530,10 +6532,6 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
{
TableInfo *tbinfo = &tblinfo[i];
- /* Only plain tables and materialized views have indexes. */
- if (tbinfo->relkind != RELKIND_RELATION &&
- tbinfo->relkind != RELKIND_MATVIEW)
- continue;
if (!tbinfo->hasindex)
continue;
@@ -6561,7 +6559,37 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
* is not.
*/
resetPQExpBuffer(query);
- if (fout->remoteVersion >= 90400)
+ if (fout->remoteVersion >= 11000)
+ {
+ appendPQExpBuffer(query,
+ "SELECT t.tableoid, t.oid, "
+ "t.relname AS indexname, "
+ "i.indparentidx, "
+ "pg_catalog.pg_get_indexdef(i.indexrelid) AS indexdef, "
+ "t.relnatts AS indnkeys, "
+ "i.indkey, i.indisclustered, "
+ "i.indisreplident, t.relpages, "
+ "c.contype, c.conname, "
+ "c.condeferrable, c.condeferred, "
+ "c.tableoid AS contableoid, "
+ "c.oid AS conoid, "
+ "pg_catalog.pg_get_constraintdef(c.oid, false) AS condef, "
+ "(SELECT spcname FROM pg_catalog.pg_tablespace s WHERE s.oid = t.reltablespace) AS tablespace, "
+ "t.reloptions AS indreloptions "
+ "FROM pg_catalog.pg_index i "
+ "JOIN pg_catalog.pg_class t ON (t.oid = i.indexrelid) "
+ "JOIN pg_catalog.pg_class t2 ON (t2.oid = i.indrelid) "
+ "LEFT JOIN pg_catalog.pg_constraint c "
+ "ON (i.indrelid = c.conrelid AND "
+ "i.indexrelid = c.conindid AND "
+ "c.contype IN ('p','u','x')) "
+ "WHERE i.indrelid = '%u'::pg_catalog.oid "
+ "AND (i.indisvalid OR t2.relkind = 'p') "
+ "AND i.indisready "
+ "ORDER BY indexname",
+ tbinfo->dobj.catId.oid);
+ }
+ else if (fout->remoteVersion >= 90400)
{
/*
* the test on indisready is necessary in 9.2, and harmless in
@@ -6570,6 +6598,7 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
appendPQExpBuffer(query,
"SELECT t.tableoid, t.oid, "
"t.relname AS indexname, "
+ "0 AS indparentidx, "
"pg_catalog.pg_get_indexdef(i.indexrelid) AS indexdef, "
"t.relnatts AS indnkeys, "
"i.indkey, i.indisclustered, "
@@ -6601,6 +6630,7 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
appendPQExpBuffer(query,
"SELECT t.tableoid, t.oid, "
"t.relname AS indexname, "
+ "0 AS indparentidx, "
"pg_catalog.pg_get_indexdef(i.indexrelid) AS indexdef, "
"t.relnatts AS indnkeys, "
"i.indkey, i.indisclustered, "
@@ -6628,6 +6658,7 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
appendPQExpBuffer(query,
"SELECT t.tableoid, t.oid, "
"t.relname AS indexname, "
+ "0 AS indparentidx, "
"pg_catalog.pg_get_indexdef(i.indexrelid) AS indexdef, "
"t.relnatts AS indnkeys, "
"i.indkey, i.indisclustered, "
@@ -6658,6 +6689,7 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
appendPQExpBuffer(query,
"SELECT t.tableoid, t.oid, "
"t.relname AS indexname, "
+ "0 AS indparentidx, "
"pg_catalog.pg_get_indexdef(i.indexrelid) AS indexdef, "
"t.relnatts AS indnkeys, "
"i.indkey, i.indisclustered, "
@@ -6690,6 +6722,7 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
i_tableoid = PQfnumber(res, "tableoid");
i_oid = PQfnumber(res, "oid");
i_indexname = PQfnumber(res, "indexname");
+ i_parentidx = PQfnumber(res, "indparentidx");
i_indexdef = PQfnumber(res, "indexdef");
i_indnkeys = PQfnumber(res, "indnkeys");
i_indkey = PQfnumber(res, "indkey");
@@ -6706,8 +6739,10 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
i_tablespace = PQfnumber(res, "tablespace");
i_indreloptions = PQfnumber(res, "indreloptions");
- indxinfo = (IndxInfo *) pg_malloc(ntups * sizeof(IndxInfo));
+ tbinfo->indexes = indxinfo =
+ (IndxInfo *) pg_malloc(ntups * sizeof(IndxInfo));
constrinfo = (ConstraintInfo *) pg_malloc(ntups * sizeof(ConstraintInfo));
+ tbinfo->numIndexes = ntups;
for (j = 0; j < ntups; j++)
{
@@ -6729,6 +6764,7 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
indxinfo[j].indkeys, indxinfo[j].indnkeys);
indxinfo[j].indisclustered = (PQgetvalue(res, j, i_indisclustered)[0] == 't');
indxinfo[j].indisreplident = (PQgetvalue(res, j, i_indisreplident)[0] == 't');
+ indxinfo[j].indparentidx = atooid(PQgetvalue(res, j, i_parentidx));
indxinfo[j].relpages = atoi(PQgetvalue(res, j, i_relpages));
contype = *(PQgetvalue(res, j, i_contype));
@@ -9512,6 +9548,9 @@ dumpDumpableObject(Archive *fout, DumpableObject *dobj)
case DO_INDEX:
dumpIndex(fout, (IndxInfo *) dobj);
break;
+ case DO_INDEX_ATTACH:
+ dumpIndexAttach(fout, (IndexAttachInfo *) dobj);
+ break;
case DO_STATSEXT:
dumpStatisticsExt(fout, (StatsExtInfo *) dobj);
break;
@@ -16173,6 +16212,42 @@ dumpIndex(Archive *fout, IndxInfo *indxinfo)
}
/*
+ * dumpIndexAttach
+ * write out to fout a partitioned-index attachment clause
+ */
+void
+dumpIndexAttach(Archive *fout, IndexAttachInfo *attachinfo)
+{
+ if (fout->dopt->dataOnly)
+ return;
+
+ if (attachinfo->partitionIdx->dobj.dump & DUMP_COMPONENT_DEFINITION)
+ {
+ PQExpBuffer q = createPQExpBuffer();
+
+ appendPQExpBuffer(q, "\nALTER INDEX %s ",
+ fmtQualifiedId(fout->remoteVersion,
+ attachinfo->parentIdx->dobj.namespace->dobj.name,
+ attachinfo->parentIdx->dobj.name));
+ appendPQExpBuffer(q, "ATTACH PARTITION %s;\n",
+ fmtQualifiedId(fout->remoteVersion,
+ attachinfo->partitionIdx->dobj.namespace->dobj.name,
+ attachinfo->partitionIdx->dobj.name));
+
+ ArchiveEntry(fout, attachinfo->dobj.catId, attachinfo->dobj.dumpId,
+ attachinfo->dobj.name,
+ NULL, NULL,
+ "",
+ false, "INDEX ATTACH", SECTION_POST_DATA,
+ q->data, "", NULL,
+ NULL, 0,
+ NULL, NULL);
+
+ destroyPQExpBuffer(q);
+ }
+}
+
+/*
* dumpStatisticsExt
* write out to fout an extended statistics object
*/
@@ -17803,6 +17878,7 @@ addBoundaryDependencies(DumpableObject **dobjs, int numObjs,
addObjectDependency(postDataBound, dobj->dumpId);
break;
case DO_INDEX:
+ case DO_INDEX_ATTACH:
case DO_STATSEXT:
case DO_REFRESH_MATVIEW:
case DO_TRIGGER:
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index 49a02b4fa8..c40cb68a7c 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -56,6 +56,7 @@ typedef enum
DO_TABLE,
DO_ATTRDEF,
DO_INDEX,
+ DO_INDEX_ATTACH,
DO_STATSEXT,
DO_RULE,
DO_TRIGGER,
@@ -328,6 +329,8 @@ typedef struct _tableInfo
*/
int numParents; /* number of (immediate) parent tables */
struct _tableInfo **parents; /* TableInfos of immediate parents */
+ int numIndexes; /* number of indexes */
+ struct _indxInfo *indexes; /* indexes */
struct _tableDataInfo *dataObj; /* TableDataInfo, if dumping its data */
int numTriggers; /* number of triggers for table */
struct _triggerInfo *triggers; /* array of TriggerInfo structs */
@@ -361,11 +364,19 @@ typedef struct _indxInfo
Oid *indkeys;
bool indisclustered;
bool indisreplident;
+ Oid indparentidx; /* if partitioned, parent index OID */
/* if there is an associated constraint object, its dumpId: */
DumpId indexconstraint;
int relpages; /* relpages of the underlying table */
} IndxInfo;
+typedef struct _indexAttachInfo
+{
+ DumpableObject dobj;
+ IndxInfo *parentIdx; /* link to index on partitioned table */
+ IndxInfo *partitionIdx; /* link to index on partition */
+} IndexAttachInfo;
+
typedef struct _statsExtInfo
{
DumpableObject dobj;
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index 6da1c35a42..171cc01e9b 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -35,6 +35,10 @@ static const char *modulename = gettext_noop("sorter");
* pg_dump.c; that is, PRE_DATA objects must sort before DO_PRE_DATA_BOUNDARY,
* POST_DATA objects must sort after DO_POST_DATA_BOUNDARY, and DATA objects
* must sort between them.
+ *
+ * Note: sortDataAndIndexObjectsBySize wants to have all DO_TABLE_DATA and
+ * DO_INDEX objects in contiguous chunks, so do not reuse the values for those
+ * for other object types.
*/
static const int dbObjectTypePriority[] =
{
@@ -53,11 +57,12 @@ static const int dbObjectTypePriority[] =
18, /* DO_TABLE */
20, /* DO_ATTRDEF */
28, /* DO_INDEX */
- 29, /* DO_STATSEXT */
- 30, /* DO_RULE */
- 31, /* DO_TRIGGER */
+ 29, /* DO_INDEX_ATTACH */
+ 30, /* DO_STATSEXT */
+ 31, /* DO_RULE */
+ 32, /* DO_TRIGGER */
27, /* DO_CONSTRAINT */
- 32, /* DO_FK_CONSTRAINT */
+ 33, /* DO_FK_CONSTRAINT */
2, /* DO_PROCLANG */
10, /* DO_CAST */
23, /* DO_TABLE_DATA */
@@ -69,18 +74,18 @@ static const int dbObjectTypePriority[] =
15, /* DO_TSCONFIG */
16, /* DO_FDW */
17, /* DO_FOREIGN_SERVER */
- 32, /* DO_DEFAULT_ACL */
+ 33, /* DO_DEFAULT_ACL */
3, /* DO_TRANSFORM */
21, /* DO_BLOB */
25, /* DO_BLOB_DATA */
22, /* DO_PRE_DATA_BOUNDARY */
26, /* DO_POST_DATA_BOUNDARY */
- 33, /* DO_EVENT_TRIGGER */
- 38, /* DO_REFRESH_MATVIEW */
- 34, /* DO_POLICY */
- 35, /* DO_PUBLICATION */
- 36, /* DO_PUBLICATION_REL */
- 37 /* DO_SUBSCRIPTION */
+ 34, /* DO_EVENT_TRIGGER */
+ 39, /* DO_REFRESH_MATVIEW */
+ 35, /* DO_POLICY */
+ 36, /* DO_PUBLICATION */
+ 37, /* DO_PUBLICATION_REL */
+ 38 /* DO_SUBSCRIPTION */
};
static DumpId preDataBoundId;
@@ -937,6 +942,13 @@ repairDomainConstraintMultiLoop(DumpableObject *domainobj,
addObjectDependency(constraintobj, postDataBoundId);
}
+static void
+repairIndexLoop(DumpableObject *partedindex,
+ DumpableObject *partindex)
+{
+ removeObjectDependency(partedindex, partindex->dumpId);
+}
+
/*
* Fix a dependency loop, or die trying ...
*
@@ -1099,6 +1111,23 @@ repairDependencyLoop(DumpableObject **loop,
return;
}
+ /* index on partitioned table and corresponding index on partition */
+ if (nLoop == 2 &&
+ loop[0]->objType == DO_INDEX &&
+ loop[1]->objType == DO_INDEX)
+ {
+ if (((IndxInfo *) loop[0])->indparentidx == loop[1]->catId.oid)
+ {
+ repairIndexLoop(loop[0], loop[1]);
+ return;
+ }
+ else if (((IndxInfo *) loop[1])->indparentidx == loop[0]->catId.oid)
+ {
+ repairIndexLoop(loop[1], loop[0]);
+ return;
+ }
+ }
+
/* Indirect loop involving table and attribute default */
if (nLoop > 2)
{
@@ -1292,6 +1321,11 @@ describeDumpableObject(DumpableObject *obj, char *buf, int bufsize)
"INDEX %s (ID %d OID %u)",
obj->name, obj->dumpId, obj->catId.oid);
return;
+ case DO_INDEX_ATTACH:
+ snprintf(buf, bufsize,
+ "INDEX ATTACH %s (ID %d)",
+ obj->name, obj->dumpId);
+ return;
case DO_STATSEXT:
snprintf(buf, bufsize,
"STATISTICS %s (ID %d OID %u)",
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index f2e62946d8..7a0c2423ac 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -1705,7 +1705,8 @@ describeOneTableDetails(const char *schemaname,
appendPQExpBufferStr(&buf, ",\n a.attidentity");
else
appendPQExpBufferStr(&buf, ",\n ''::pg_catalog.char AS attidentity");
- if (tableinfo.relkind == RELKIND_INDEX)
+ if (tableinfo.relkind == RELKIND_INDEX ||
+ tableinfo.relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ",\n pg_catalog.pg_get_indexdef(a.attrelid, a.attnum, TRUE) AS indexdef");
else
appendPQExpBufferStr(&buf, ",\n NULL AS indexdef");
@@ -1766,6 +1767,7 @@ describeOneTableDetails(const char *schemaname,
schemaname, relationname);
break;
case RELKIND_INDEX:
+ case RELKIND_PARTITIONED_INDEX:
if (tableinfo.relpersistence == 'u')
printfPQExpBuffer(&title, _("Unlogged index \"%s.%s\""),
schemaname, relationname);
@@ -1823,7 +1825,8 @@ describeOneTableDetails(const char *schemaname,
show_column_details = true;
}
- if (tableinfo.relkind == RELKIND_INDEX)
+ if (tableinfo.relkind == RELKIND_INDEX ||
+ tableinfo.relkind == RELKIND_PARTITIONED_INDEX)
headers[cols++] = gettext_noop("Definition");
if (tableinfo.relkind == RELKIND_FOREIGN_TABLE && pset.sversion >= 90200)
@@ -1834,6 +1837,7 @@ describeOneTableDetails(const char *schemaname,
headers[cols++] = gettext_noop("Storage");
if (tableinfo.relkind == RELKIND_RELATION ||
tableinfo.relkind == RELKIND_INDEX ||
+ tableinfo.relkind == RELKIND_PARTITIONED_INDEX ||
tableinfo.relkind == RELKIND_MATVIEW ||
tableinfo.relkind == RELKIND_FOREIGN_TABLE ||
tableinfo.relkind == RELKIND_PARTITIONED_TABLE)
@@ -1906,7 +1910,8 @@ describeOneTableDetails(const char *schemaname,
}
/* Expression for index column */
- if (tableinfo.relkind == RELKIND_INDEX)
+ if (tableinfo.relkind == RELKIND_INDEX ||
+ tableinfo.relkind == RELKIND_PARTITIONED_INDEX)
printTableAddCell(&cont, PQgetvalue(res, i, 7), false, false);
/* FDW options for foreign table column, only for 9.2 or later */
@@ -1930,6 +1935,7 @@ describeOneTableDetails(const char *schemaname,
/* Statistics target, if the relkind supports this feature */
if (tableinfo.relkind == RELKIND_RELATION ||
tableinfo.relkind == RELKIND_INDEX ||
+ tableinfo.relkind == RELKIND_PARTITIONED_INDEX ||
tableinfo.relkind == RELKIND_MATVIEW ||
tableinfo.relkind == RELKIND_FOREIGN_TABLE ||
tableinfo.relkind == RELKIND_PARTITIONED_TABLE)
@@ -2021,7 +2027,8 @@ describeOneTableDetails(const char *schemaname,
PQclear(result);
}
- if (tableinfo.relkind == RELKIND_INDEX)
+ if (tableinfo.relkind == RELKIND_INDEX ||
+ tableinfo.relkind == RELKIND_PARTITIONED_INDEX)
{
/* Footer information about an index */
PGresult *result;
@@ -3397,6 +3404,7 @@ listTables(const char *tabtypes, const char *pattern, bool verbose, bool showSys
" WHEN 's' THEN '%s'"
" WHEN " CppAsString2(RELKIND_FOREIGN_TABLE) " THEN '%s'"
" WHEN " CppAsString2(RELKIND_PARTITIONED_TABLE) " THEN '%s'"
+ " WHEN " CppAsString2(RELKIND_PARTITIONED_INDEX) " THEN '%s'"
" END as \"%s\",\n"
" pg_catalog.pg_get_userbyid(c.relowner) as \"%s\"",
gettext_noop("Schema"),
@@ -3409,6 +3417,7 @@ listTables(const char *tabtypes, const char *pattern, bool verbose, bool showSys
gettext_noop("special"),
gettext_noop("foreign table"),
gettext_noop("table"), /* partitioned table */
+ gettext_noop("index"), /* partitioned index */
gettext_noop("Type"),
gettext_noop("Owner"));
@@ -3454,7 +3463,8 @@ listTables(const char *tabtypes, const char *pattern, bool verbose, bool showSys
if (showMatViews)
appendPQExpBufferStr(&buf, CppAsString2(RELKIND_MATVIEW) ",");
if (showIndexes)
- appendPQExpBufferStr(&buf, CppAsString2(RELKIND_INDEX) ",");
+ appendPQExpBufferStr(&buf, CppAsString2(RELKIND_INDEX) ","
+ CppAsString2(RELKIND_PARTITIONED_INDEX) ",");
if (showSeq)
appendPQExpBufferStr(&buf, CppAsString2(RELKIND_SEQUENCE) ",");
if (showSystem || pattern)
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index b51098deca..8bc4a194a5 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -412,7 +412,8 @@ static const SchemaQuery Query_for_list_of_indexes = {
/* catname */
"pg_catalog.pg_class c",
/* selcondition */
- "c.relkind IN (" CppAsString2(RELKIND_INDEX) ")",
+ "c.relkind IN (" CppAsString2(RELKIND_INDEX) ", "
+ CppAsString2(RELKIND_PARTITIONED_INDEX) ")",
/* viscondition */
"pg_catalog.pg_table_is_visible(c.oid)",
/* namespace */
@@ -600,6 +601,23 @@ static const SchemaQuery Query_for_list_of_tmf = {
NULL
};
+static const SchemaQuery Query_for_list_of_tpm = {
+ /* catname */
+ "pg_catalog.pg_class c",
+ /* selcondition */
+ "c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
+ CppAsString2(RELKIND_MATVIEW) ")",
+ /* viscondition */
+ "pg_catalog.pg_table_is_visible(c.oid)",
+ /* namespace */
+ "c.relnamespace",
+ /* result */
+ "pg_catalog.quote_ident(c.relname)",
+ /* qualresult */
+ NULL
+};
+
static const SchemaQuery Query_for_list_of_tm = {
/* catname */
"pg_catalog.pg_class c",
@@ -1676,7 +1694,12 @@ psql_completion(const char *text, int start, int end)
"UNION SELECT 'ALL IN TABLESPACE'");
/* ALTER INDEX <name> */
else if (Matches3("ALTER", "INDEX", MatchAny))
- COMPLETE_WITH_LIST5("ALTER COLUMN", "OWNER TO", "RENAME TO", "SET", "RESET");
+ COMPLETE_WITH_LIST6("ALTER COLUMN", "OWNER TO", "RENAME TO", "SET",
+ "RESET", "ATTACH PARTITION");
+ else if (Matches4("ALTER", "INDEX", MatchAny, "ATTACH"))
+ COMPLETE_WITH_CONST("PARTITION");
+ else if (Matches5("ALTER", "INDEX", MatchAny, "ATTACH", "PARTITION"))
+ COMPLETE_WITH_SCHEMA_QUERY(Query_for_list_of_indexes, NULL);
/* ALTER INDEX <name> ALTER COLUMN <colnum> */
else if (Matches6("ALTER", "INDEX", MatchAny, "ALTER", "COLUMN", MatchAny))
COMPLETE_WITH_CONST("SET STATISTICS");
@@ -2338,10 +2361,13 @@ psql_completion(const char *text, int start, int end)
COMPLETE_WITH_SCHEMA_QUERY(Query_for_list_of_indexes,
" UNION SELECT 'ON'"
" UNION SELECT 'CONCURRENTLY'");
- /* Complete ... INDEX|CONCURRENTLY [<name>] ON with a list of tables */
+ /*
+ * Complete ... INDEX|CONCURRENTLY [<name>] ON with a list of relations
+ * that can indexes can be created on
+ */
else if (TailMatches3("INDEX|CONCURRENTLY", MatchAny, "ON") ||
TailMatches2("INDEX|CONCURRENTLY", "ON"))
- COMPLETE_WITH_SCHEMA_QUERY(Query_for_list_of_tm, NULL);
+ COMPLETE_WITH_SCHEMA_QUERY(Query_for_list_of_tpm, NULL);
/*
* Complete CREATE|UNIQUE INDEX CONCURRENTLY with "ON" and existing
diff --git a/src/include/catalog/catversion.h b/src/include/catalog/catversion.h
index f1765af4ba..2b2c166266 100644
--- a/src/include/catalog/catversion.h
+++ b/src/include/catalog/catversion.h
@@ -53,6 +53,6 @@
*/
/* yyyymmddN */
-#define CATALOG_VERSION_NO 201712251
+#define CATALOG_VERSION_NO 201712291
#endif
diff --git a/src/include/catalog/dependency.h b/src/include/catalog/dependency.h
index 6f290d5c6f..5dfb79347b 100644
--- a/src/include/catalog/dependency.h
+++ b/src/include/catalog/dependency.h
@@ -49,6 +49,13 @@
* Example: a trigger that's created to enforce a foreign-key constraint
* is made internally dependent on the constraint's pg_constraint entry.
*
+ * DEPENDENCY_INTERNAL_AUTO ('I'): the dependent object was created as
+ * part of creation of the referenced object, and is really just a part
+ * of its internal implementation. A DROP of the dependent object will
+ * be disallowed outright (we'll tell the user to issue a DROP against the
+ * referenced object, instead). As opposed to DROP_INTERNAL [... explain
+ * the difference ... ]
+ *
* DEPENDENCY_EXTENSION ('e'): the dependent object is a member of the
* extension that is the referenced object. The dependent object can be
* dropped only via DROP EXTENSION on the referenced object. Functionally
@@ -75,6 +82,7 @@ typedef enum DependencyType
DEPENDENCY_NORMAL = 'n',
DEPENDENCY_AUTO = 'a',
DEPENDENCY_INTERNAL = 'i',
+ DEPENDENCY_INTERNAL_AUTO = 'I',
DEPENDENCY_EXTENSION = 'e',
DEPENDENCY_AUTO_EXTENSION = 'x',
DEPENDENCY_PIN = 'p'
diff --git a/src/include/catalog/index.h b/src/include/catalog/index.h
index 12bf35567a..139365c3b3 100644
--- a/src/include/catalog/index.h
+++ b/src/include/catalog/index.h
@@ -47,10 +47,13 @@ extern void index_check_primary_key(Relation heapRel,
#define INDEX_CREATE_SKIP_BUILD (1 << 2)
#define INDEX_CREATE_CONCURRENT (1 << 3)
#define INDEX_CREATE_IF_NOT_EXISTS (1 << 4)
+#define INDEX_CREATE_PARTITIONED (1 << 5)
+#define INDEX_CREATE_INVALID (1 << 6)
extern Oid index_create(Relation heapRelation,
const char *indexRelationName,
Oid indexRelationId,
+ Oid parentIndexRelid,
Oid relFileNode,
IndexInfo *indexInfo,
List *indexColNames,
@@ -84,6 +87,8 @@ extern void index_drop(Oid indexId, bool concurrent);
extern IndexInfo *BuildIndexInfo(Relation index);
+extern bool CompareIndexInfo(IndexInfo *info1, IndexInfo *info2, AttrNumber *attmap);
+
extern void BuildSpeculativeIndexInfo(Relation index, IndexInfo *ii);
extern void FormIndexDatum(IndexInfo *indexInfo,
@@ -134,4 +139,6 @@ extern bool ReindexIsProcessingHeap(Oid heapOid);
extern bool ReindexIsProcessingIndex(Oid indexOid);
extern Oid IndexGetRelation(Oid indexId, bool missing_ok);
+extern void IndexSetParentIndex(Relation idx, Oid parentOid);
+
#endif /* INDEX_H */
diff --git a/src/include/catalog/indexing.h b/src/include/catalog/indexing.h
index 0bb875441e..8dfd3c92a5 100644
--- a/src/include/catalog/indexing.h
+++ b/src/include/catalog/indexing.h
@@ -162,6 +162,8 @@ DECLARE_INDEX(pg_index_indrelid_index, 2678, on pg_index using btree(indrelid oi
#define IndexIndrelidIndexId 2678
DECLARE_UNIQUE_INDEX(pg_index_indexrelid_index, 2679, on pg_index using btree(indexrelid oid_ops));
#define IndexRelidIndexId 2679
+DECLARE_INDEX(pg_index_parentidx_relid_index, 2579, on pg_index using btree(indparentidx oid_ops, indrelid oid_ops));
+#define IndexParentidxRelidIndexId 2579
DECLARE_UNIQUE_INDEX(pg_inherits_relid_seqno_index, 2680, on pg_inherits using btree(inhrelid oid_ops, inhseqno int4_ops));
#define InheritsRelidSeqnoIndexId 2680
diff --git a/src/include/catalog/pg_class.h b/src/include/catalog/pg_class.h
index e7049438eb..26b1866c69 100644
--- a/src/include/catalog/pg_class.h
+++ b/src/include/catalog/pg_class.h
@@ -166,6 +166,7 @@ DESCR("");
#define RELKIND_COMPOSITE_TYPE 'c' /* composite type */
#define RELKIND_FOREIGN_TABLE 'f' /* foreign table */
#define RELKIND_PARTITIONED_TABLE 'p' /* partitioned table */
+#define RELKIND_PARTITIONED_INDEX 'I' /* partitioned index */
#define RELPERSISTENCE_PERMANENT 'p' /* regular table */
#define RELPERSISTENCE_UNLOGGED 'u' /* unlogged permanent table */
diff --git a/src/include/catalog/pg_index.h b/src/include/catalog/pg_index.h
index 057a9f7fe4..11defe1803 100644
--- a/src/include/catalog/pg_index.h
+++ b/src/include/catalog/pg_index.h
@@ -32,6 +32,7 @@ CATALOG(pg_index,2610) BKI_WITHOUT_OIDS BKI_SCHEMA_MACRO
{
Oid indexrelid; /* OID of the index */
Oid indrelid; /* OID of the relation it indexes */
+ Oid indparentidx; /* OID of parent index, 0 if not partitioned */
int16 indnatts; /* number of columns in index */
bool indisunique; /* is this a unique index? */
bool indisprimary; /* is this index for primary key? */
@@ -70,26 +71,27 @@ typedef FormData_pg_index *Form_pg_index;
* compiler constants for pg_index
* ----------------
*/
-#define Natts_pg_index 19
+#define Natts_pg_index 20
#define Anum_pg_index_indexrelid 1
#define Anum_pg_index_indrelid 2
-#define Anum_pg_index_indnatts 3
-#define Anum_pg_index_indisunique 4
-#define Anum_pg_index_indisprimary 5
-#define Anum_pg_index_indisexclusion 6
-#define Anum_pg_index_indimmediate 7
-#define Anum_pg_index_indisclustered 8
-#define Anum_pg_index_indisvalid 9
-#define Anum_pg_index_indcheckxmin 10
-#define Anum_pg_index_indisready 11
-#define Anum_pg_index_indislive 12
-#define Anum_pg_index_indisreplident 13
-#define Anum_pg_index_indkey 14
-#define Anum_pg_index_indcollation 15
-#define Anum_pg_index_indclass 16
-#define Anum_pg_index_indoption 17
-#define Anum_pg_index_indexprs 18
-#define Anum_pg_index_indpred 19
+#define Anum_pg_index_indparentidx 3
+#define Anum_pg_index_indnatts 4
+#define Anum_pg_index_indisunique 5
+#define Anum_pg_index_indisprimary 6
+#define Anum_pg_index_indisexclusion 7
+#define Anum_pg_index_indimmediate 8
+#define Anum_pg_index_indisclustered 9
+#define Anum_pg_index_indisvalid 10
+#define Anum_pg_index_indcheckxmin 11
+#define Anum_pg_index_indisready 12
+#define Anum_pg_index_indislive 13
+#define Anum_pg_index_indisreplident 14
+#define Anum_pg_index_indkey 15
+#define Anum_pg_index_indcollation 16
+#define Anum_pg_index_indclass 17
+#define Anum_pg_index_indoption 18
+#define Anum_pg_index_indexprs 19
+#define Anum_pg_index_indpred 20
/*
* Index AMs that support ordered scans must support these two indoption
diff --git a/src/include/commands/defrem.h b/src/include/commands/defrem.h
index 1f18cad963..41007162aa 100644
--- a/src/include/commands/defrem.h
+++ b/src/include/commands/defrem.h
@@ -25,12 +25,13 @@ extern void RemoveObjects(DropStmt *stmt);
extern ObjectAddress DefineIndex(Oid relationId,
IndexStmt *stmt,
Oid indexRelationId,
+ Oid parentIndexId,
bool is_alter_table,
bool check_rights,
bool check_not_in_use,
bool skip_build,
bool quiet);
-extern Oid ReindexIndex(RangeVar *indexRelation, int options);
+extern void ReindexIndex(RangeVar *indexRelation, int options);
extern Oid ReindexTable(RangeVar *relation, int options);
extern void ReindexMultipleTables(const char *objectName, ReindexObjectType objectKind,
int options);
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 3ad58cdfe7..cbc8d5fca8 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -158,6 +158,7 @@ typedef struct IndexInfo
bool ii_ReadyForInserts;
bool ii_Concurrent;
bool ii_BrokenHotChain;
+ Oid ii_Am;
void *ii_AmCache;
MemoryContext ii_Context;
} IndexInfo;
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index b72178efd1..0296784726 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -839,7 +839,7 @@ typedef struct PartitionRangeDatum
} PartitionRangeDatum;
/*
- * PartitionCmd - info for ALTER TABLE ATTACH/DETACH PARTITION commands
+ * PartitionCmd - info for ALTER TABLE/INDEX ATTACH/DETACH PARTITION commands
*/
typedef struct PartitionCmd
{
@@ -2702,6 +2702,10 @@ typedef struct FetchStmt
* index, just a UNIQUE/PKEY constraint using an existing index. isconstraint
* must always be true in this case, and the fields describing the index
* properties are empty.
+ *
+ * The relation to build the index on can be represented either by name
+ * (in which case the RangeVar indicates whether to recurse or not) or by OID
+ * (in which case the command is always recursive).
* ----------------------
*/
typedef struct IndexStmt
@@ -2709,6 +2713,7 @@ typedef struct IndexStmt
NodeTag type;
char *idxname; /* name of new index, or NULL for default */
RangeVar *relation; /* relation to build index on */
+ Oid relationId; /* OID of relation to build index on */
char *accessMethod; /* name of access method (eg. btree) */
char *tableSpace; /* tablespace, or NULL for default */
List *indexParams; /* columns to index: a list of IndexElem */
diff --git a/src/include/parser/parse_utilcmd.h b/src/include/parser/parse_utilcmd.h
index a7f5e0caea..64aa8234e5 100644
--- a/src/include/parser/parse_utilcmd.h
+++ b/src/include/parser/parse_utilcmd.h
@@ -27,5 +27,8 @@ extern void transformRuleStmt(RuleStmt *stmt, const char *queryString,
extern List *transformCreateSchemaStmt(CreateSchemaStmt *stmt);
extern PartitionBoundSpec *transformPartitionBound(ParseState *pstate, Relation parent,
PartitionBoundSpec *spec);
+extern IndexStmt *generateClonedIndexStmt(RangeVar *heapRel, Oid heapOid,
+ Relation source_idx,
+ const AttrNumber *attmap, int attmap_length);
#endif /* PARSE_UTILCMD_H */
diff --git a/src/test/regress/expected/alter_table.out b/src/test/regress/expected/alter_table.out
index 11f0baa11b..517fb080bd 100644
--- a/src/test/regress/expected/alter_table.out
+++ b/src/test/regress/expected/alter_table.out
@@ -1965,6 +1965,67 @@ create table tab1 (a int, b text);
create table tab2 (x int, y tab1);
alter table tab1 alter column b type varchar; -- fails
ERROR: cannot alter table "tab1" because column "tab2.y" uses its row type
+-- Alter column type that's part of a partitioned index
+create table at_partitioned (a int, b text) partition by range (a);
+create table at_part_1 partition of at_partitioned for values from (0) to (1000);
+insert into at_partitioned values (512, '0.123');
+create table at_part_2 (b text, a int);
+insert into at_part_2 values ('1.234', 1024);
+create index on at_partitioned (b);
+create index on at_partitioned (a);
+\d at_part_1
+ Table "public.at_part_1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | text | | |
+Partition of: at_partitioned FOR VALUES FROM (0) TO (1000)
+Indexes:
+ "at_part_1_a_idx" btree (a)
+ "at_part_1_b_idx" btree (b)
+
+\d at_part_2
+ Table "public.at_part_2"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ b | text | | |
+ a | integer | | |
+
+alter table at_partitioned attach partition at_part_2 for values from (1000) to (2000);
+\d at_part_2
+ Table "public.at_part_2"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ b | text | | |
+ a | integer | | |
+Partition of: at_partitioned FOR VALUES FROM (1000) TO (2000)
+Indexes:
+ "at_part_2_a_idx" btree (a)
+ "at_part_2_b_idx" btree (b)
+
+alter table at_partitioned alter column b type numeric using b::numeric;
+\d at_part_1
+ Table "public.at_part_1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | numeric | | |
+Partition of: at_partitioned FOR VALUES FROM (0) TO (1000)
+Indexes:
+ "at_part_1_a_idx" btree (a)
+ "at_part_1_b_idx" btree (b)
+
+\d at_part_2
+ Table "public.at_part_2"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ b | numeric | | |
+ a | integer | | |
+Partition of: at_partitioned FOR VALUES FROM (1000) TO (2000)
+Indexes:
+ "at_part_2_a_idx" btree (a)
+ "at_part_2_b_idx" btree (b)
+
-- disallow recursive containment of row types
create temp table recur1 (f1 int);
alter table recur1 add column f2 recur1; -- fails
@@ -3276,7 +3337,7 @@ CREATE TABLE unparted (
);
CREATE TABLE fail_part (like unparted);
ALTER TABLE unparted ATTACH PARTITION fail_part FOR VALUES IN ('a');
-ERROR: "unparted" is not partitioned
+ERROR: table "unparted" is not partitioned
DROP TABLE unparted, fail_part;
-- check that partition bound is compatible
CREATE TABLE list_parted (
@@ -3656,7 +3717,7 @@ DROP TABLE fail_part;
-- check that the table is partitioned at all
CREATE TABLE regular_table (a int);
ALTER TABLE regular_table DETACH PARTITION any_name;
-ERROR: "regular_table" is not partitioned
+ERROR: table "regular_table" is not partitioned
DROP TABLE regular_table;
-- check that the partition being detached exists at all
ALTER TABLE list_parted2 DETACH PARTITION part_4;
diff --git a/src/test/regress/expected/indexing.out b/src/test/regress/expected/indexing.out
new file mode 100644
index 0000000000..b64b466b11
--- /dev/null
+++ b/src/test/regress/expected/indexing.out
@@ -0,0 +1,533 @@
+-- Creating an index on a partitioned table makes the partitions
+-- automatically get the index
+create table idxpart (a int, b int, c text) partition by range (a);
+create table idxpart1 partition of idxpart for values from (0) to (10);
+create table idxpart2 partition of idxpart for values from (10) to (100)
+ partition by range (b);
+create table idxpart21 partition of idxpart2 for values from (0) to (100);
+create index on idxpart (a);
+select relname, relkind, indparentidx::regclass
+ from pg_class left join pg_index on (indexrelid = oid)
+ where relname like 'idxpart%' order by relname;
+ relname | relkind | indparentidx
+-----------------+---------+----------------
+ idxpart | p |
+ idxpart1 | r |
+ idxpart1_a_idx | i | idxpart_a_idx
+ idxpart2 | p |
+ idxpart21 | r |
+ idxpart21_a_idx | i | idxpart2_a_idx
+ idxpart2_a_idx | I | idxpart_a_idx
+ idxpart_a_idx | I | -
+(8 rows)
+
+drop table idxpart;
+-- Some unsupported features
+create table idxpart (a int, b int, c text) partition by range (a);
+create table idxpart1 partition of idxpart for values from (0) to (10);
+create unique index on idxpart (a);
+ERROR: cannot create unique index on partitioned table "idxpart"
+create index concurrently on idxpart (a);
+ERROR: cannot create index on partitioned table "idxpart" concurrently
+drop table idxpart;
+-- If a table without index is attached as partition to a table with
+-- an index, the index is automatically created
+create table idxpart (a int, b int, c text) partition by range (a);
+create index idxparti on idxpart (a);
+create index idxparti2 on idxpart (b, c);
+create table idxpart1 (like idxpart);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | integer | | |
+ c | text | | |
+
+alter table idxpart attach partition idxpart1 for values from (0) to (10);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | integer | | |
+ c | text | | |
+Partition of: idxpart FOR VALUES FROM (0) TO (10)
+Indexes:
+ "idxpart1_a_idx" btree (a)
+ "idxpart1_b_c_idx" btree (b, c)
+
+drop table idxpart;
+-- If a partition already has an index, don't create a duplicative one
+create table idxpart (a int, b int) partition by range (a, b);
+create table idxpart1 partition of idxpart for values from (0, 0) to (10, 10);
+create index on idxpart1 (a, b);
+create index on idxpart (a, b);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | integer | | |
+Partition of: idxpart FOR VALUES FROM (0, 0) TO (10, 10)
+Indexes:
+ "idxpart1_a_b_idx" btree (a, b)
+
+select indexrelid::regclass, indrelid::regclass, indparentidx::regclass
+ from pg_index where indexrelid::regclass::text like 'idxpart%'
+ order by indexrelid::regclass;
+ indexrelid | indrelid | indparentidx
+------------------+----------+-----------------
+ idxpart1_a_b_idx | idxpart1 | idxpart_a_b_idx
+ idxpart_a_b_idx | idxpart | -
+(2 rows)
+
+drop table idxpart;
+-- DROP behavior for partitioned indexes
+create table idxpart (a int) partition by range (a);
+create index on idxpart (a);
+create table idxpart1 partition of idxpart for values from (0) to (10);
+drop index idxpart1_a_idx; -- no way
+ERROR: cannot drop index idxpart1_a_idx because index idxpart_a_idx requires it
+HINT: You can drop index idxpart_a_idx instead.
+drop index idxpart_a_idx; -- both indexes go away
+select relname, relkind from pg_class
+ where relname like 'idxpart%' order by relname;
+ relname | relkind
+----------+---------
+ idxpart | p
+ idxpart1 | r
+(2 rows)
+
+create index on idxpart (a);
+drop table idxpart1; -- the index on partition goes away too
+select relname, relkind from pg_class
+ where relname like 'idxpart%' order by relname;
+ relname | relkind
+---------------+---------
+ idxpart | p
+ idxpart_a_idx | I
+(2 rows)
+
+drop table idxpart;
+-- ALTER INDEX .. ATTACH, error cases
+create table idxpart (a int, b int) partition by range (a, b);
+create table idxpart1 partition of idxpart for values from (0, 0) to (10, 10);
+create index idxpart_a_b_idx on only idxpart (a, b);
+create index idxpart1_a_b_idx on idxpart1 (a, b);
+create index idxpart1_tst1 on idxpart1 (b, a);
+create index idxpart1_tst2 on idxpart1 using hash (a);
+create index idxpart1_tst3 on idxpart1 (a, b) where a > 10;
+alter index idxpart attach partition idxpart1;
+ERROR: "idxpart" is not an index
+alter index idxpart_a_b_idx attach partition idxpart1;
+ERROR: "idxpart1" is not an index
+alter index idxpart_a_b_idx attach partition idxpart_a_b_idx;
+ERROR: cannot attach index "idxpart_a_b_idx" as a partition of index "idxpart_a_b_idx"
+DETAIL: Index "idxpart_a_b_idx" is not on a partition of table "idxpart".
+alter index idxpart_a_b_idx attach partition idxpart1_b_idx;
+ERROR: relation "idxpart1_b_idx" does not exist
+alter index idxpart_a_b_idx attach partition idxpart1_tst1;
+ERROR: cannot attach index "idxpart1_tst1" as a partition of index "idxpart_a_b_idx"
+DETAIL: The index definitions do not match.
+alter index idxpart_a_b_idx attach partition idxpart1_tst2;
+ERROR: cannot attach index "idxpart1_tst2" as a partition of index "idxpart_a_b_idx"
+DETAIL: The index definitions do not match.
+alter index idxpart_a_b_idx attach partition idxpart1_tst3;
+ERROR: cannot attach index "idxpart1_tst3" as a partition of index "idxpart_a_b_idx"
+DETAIL: The index definitions do not match.
+-- OK
+alter index idxpart_a_b_idx attach partition idxpart1_a_b_idx;
+alter index idxpart_a_b_idx attach partition idxpart1_a_b_idx; -- quiet
+-- reject dupe
+create index idxpart1_2_a_b on idxpart1 (a, b);
+alter index idxpart_a_b_idx attach partition idxpart1_2_a_b;
+ERROR: cannot attach index "idxpart1_2_a_b" as a partition of index "idxpart_a_b_idx"
+DETAIL: Another index is already attached for partition "idxpart1".
+drop table idxpart;
+-- make sure everything's gone
+select indexrelid::regclass, indrelid::regclass, indparentidx::regclass
+ from pg_index where indexrelid::regclass::text like 'idxpart%';
+ indexrelid | indrelid | indparentidx
+------------+----------+--------------
+(0 rows)
+
+-- Don't auto-attach incompatible indexes
+create table idxpart (a int, b int) partition by range (a);
+create table idxpart1 (a int, b int);
+create index on idxpart1 using hash (a);
+create index on idxpart1 (a) where b > 1;
+create index on idxpart1 ((a + 0));
+create index on idxpart1 (a, a);
+create index on idxpart (a);
+alter table idxpart attach partition idxpart1 for values from (0) to (1000);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | integer | | |
+Partition of: idxpart FOR VALUES FROM (0) TO (1000)
+Indexes:
+ "idxpart1_a_a1_idx" btree (a, a)
+ "idxpart1_a_idx" hash (a)
+ "idxpart1_a_idx1" btree (a) WHERE b > 1
+ "idxpart1_a_idx2" btree (a)
+ "idxpart1_expr_idx" btree ((a + 0))
+
+drop table idxpart;
+-- If CREATE INDEX ONLY, don't create indexes on partitions; and existing
+-- indexes on partitions don't change parent. ALTER INDEX ATTACH can change
+-- the parent after the fact.
+create table idxpart (a int) partition by range (a);
+create table idxpart1 partition of idxpart for values from (0) to (100);
+create table idxpart2 partition of idxpart for values from (100) to (1000)
+ partition by range (a);
+create table idxpart21 partition of idxpart2 for values from (100) to (200);
+create table idxpart22 partition of idxpart2 for values from (200) to (300);
+create index on idxpart22 (a);
+create index on only idxpart2 (a);
+create index on idxpart (a);
+-- Here we expect that idxpart1 and idxpart2 have a new index, but idxpart21
+-- does not; also, idxpart22 is not attached.
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition of: idxpart FOR VALUES FROM (0) TO (100)
+Indexes:
+ "idxpart1_a_idx" btree (a)
+
+\d idxpart2
+ Table "public.idxpart2"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition of: idxpart FOR VALUES FROM (100) TO (1000)
+Partition key: RANGE (a)
+Indexes:
+ "idxpart2_a_idx" btree (a) INVALID
+Number of partitions: 2 (Use \d+ to list them.)
+
+\d idxpart21
+ Table "public.idxpart21"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition of: idxpart2 FOR VALUES FROM (100) TO (200)
+
+select indexrelid::regclass, indrelid::regclass, indparentidx::regclass
+ from pg_index where indexrelid::regclass::text like 'idxpart%'
+ order by indrelid::regclass;
+ indexrelid | indrelid | indparentidx
+-----------------+-----------+---------------
+ idxpart_a_idx | idxpart | -
+ idxpart1_a_idx | idxpart1 | idxpart_a_idx
+ idxpart2_a_idx | idxpart2 | idxpart_a_idx
+ idxpart22_a_idx | idxpart22 | -
+(4 rows)
+
+alter index idxpart2_a_idx attach partition idxpart22_a_idx;
+select indexrelid::regclass, indrelid::regclass, indparentidx::regclass
+ from pg_index where indexrelid::regclass::text like 'idxpart%'
+ order by indrelid::regclass;
+ indexrelid | indrelid | indparentidx
+-----------------+-----------+----------------
+ idxpart_a_idx | idxpart | -
+ idxpart1_a_idx | idxpart1 | idxpart_a_idx
+ idxpart2_a_idx | idxpart2 | idxpart_a_idx
+ idxpart22_a_idx | idxpart22 | idxpart2_a_idx
+(4 rows)
+
+-- attaching idxpart22 is not enough to set idxpart22_a_idx valid ...
+alter index idxpart2_a_idx attach partition idxpart22_a_idx;
+\d idxpart2
+ Table "public.idxpart2"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition of: idxpart FOR VALUES FROM (100) TO (1000)
+Partition key: RANGE (a)
+Indexes:
+ "idxpart2_a_idx" btree (a) INVALID
+Number of partitions: 2 (Use \d+ to list them.)
+
+-- ... but this one is.
+create index on idxpart21 (a);
+alter index idxpart2_a_idx attach partition idxpart21_a_idx;
+\d idxpart2
+ Table "public.idxpart2"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition of: idxpart FOR VALUES FROM (100) TO (1000)
+Partition key: RANGE (a)
+Indexes:
+ "idxpart2_a_idx" btree (a)
+Number of partitions: 2 (Use \d+ to list them.)
+
+drop table idxpart;
+-- When a table is attached a partition and it already has an index, a
+-- duplicate index should not get created, but rather the index becomes
+-- attached to the parent's index.
+create table idxpart (a int, b int, c text) partition by range (a);
+create index idxparti on idxpart (a);
+create index idxparti2 on idxpart (b, c);
+create table idxpart1 (like idxpart including indexes);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | integer | | |
+ c | text | | |
+Indexes:
+ "idxpart1_a_idx" btree (a)
+ "idxpart1_b_c_idx" btree (b, c)
+
+select relname, relkind, indparentidx::regclass
+ from pg_class left join pg_index on (indexrelid = oid)
+ where relname like 'idxpart%' order by relname;
+ relname | relkind | indparentidx
+------------------+---------+--------------
+ idxpart | p |
+ idxpart1 | r |
+ idxpart1_a_idx | i | -
+ idxpart1_b_c_idx | i | -
+ idxparti | I | -
+ idxparti2 | I | -
+(6 rows)
+
+alter table idxpart attach partition idxpart1 for values from (0) to (10);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | integer | | |
+ c | text | | |
+Partition of: idxpart FOR VALUES FROM (0) TO (10)
+Indexes:
+ "idxpart1_a_idx" btree (a)
+ "idxpart1_b_c_idx" btree (b, c)
+
+select relname, relkind, indparentidx::regclass
+ from pg_class left join pg_index on (indexrelid = oid)
+ where relname like 'idxpart%' order by relname;
+ relname | relkind | indparentidx
+------------------+---------+--------------
+ idxpart | p |
+ idxpart1 | r |
+ idxpart1_a_idx | i | idxparti
+ idxpart1_b_c_idx | i | idxparti2
+ idxparti | I | -
+ idxparti2 | I | -
+(6 rows)
+
+drop table idxpart;
+-- verify dependency handling during ALTER TABLE DETACH PARTITION
+create table idxpart (a int) partition by range (a);
+create table idxpart1 (like idxpart);
+create index on idxpart1 (a);
+create index on idxpart (a);
+create table idxpart2 (like idxpart);
+alter table idxpart attach partition idxpart1 for values from (0000) to (1000);
+alter table idxpart attach partition idxpart2 for values from (1000) to (2000);
+create table idxpart3 partition of idxpart for values from (2000) to (3000);
+select relname, relkind from pg_class where relname like 'idxpart%';
+ relname | relkind
+----------------+---------
+ idxpart | p
+ idxpart1 | r
+ idxpart1_a_idx | i
+ idxpart2 | r
+ idxpart2_a_idx | i
+ idxpart3 | r
+ idxpart3_a_idx | i
+ idxpart_a_idx | I
+(8 rows)
+
+-- a) after detaching partitions, the indexes can be dropped independently
+alter table idxpart detach partition idxpart1;
+alter table idxpart detach partition idxpart2;
+alter table idxpart detach partition idxpart3;
+drop index idxpart1_a_idx;
+drop index idxpart2_a_idx;
+drop index idxpart3_a_idx;
+select relname, relkind from pg_class where relname like 'idxpart%';
+ relname | relkind
+---------------+---------
+ idxpart | p
+ idxpart1 | r
+ idxpart2 | r
+ idxpart3 | r
+ idxpart_a_idx | I
+(5 rows)
+
+drop table idxpart, idxpart1, idxpart2, idxpart3;
+select relname, relkind from pg_class where relname like 'idxpart%';
+ relname | relkind
+---------+---------
+(0 rows)
+
+create table idxpart (a int) partition by range (a);
+create table idxpart1 (like idxpart);
+create index on idxpart1 (a);
+create index on idxpart (a);
+create table idxpart2 (like idxpart);
+alter table idxpart attach partition idxpart1 for values from (0000) to (1000);
+alter table idxpart attach partition idxpart2 for values from (1000) to (2000);
+create table idxpart3 partition of idxpart for values from (2000) to (3000);
+-- b) after detaching, dropping the index on parent does not remove the others
+select relname, relkind from pg_class where relname like 'idxpart%';
+ relname | relkind
+----------------+---------
+ idxpart | p
+ idxpart1 | r
+ idxpart1_a_idx | i
+ idxpart2 | r
+ idxpart2_a_idx | i
+ idxpart3 | r
+ idxpart3_a_idx | i
+ idxpart_a_idx | I
+(8 rows)
+
+alter table idxpart detach partition idxpart1;
+alter table idxpart detach partition idxpart2;
+alter table idxpart detach partition idxpart3;
+drop index idxpart_a_idx;
+select relname, relkind from pg_class where relname like 'idxpart%';
+ relname | relkind
+----------------+---------
+ idxpart | p
+ idxpart1 | r
+ idxpart1_a_idx | i
+ idxpart2 | r
+ idxpart2_a_idx | i
+ idxpart3 | r
+ idxpart3_a_idx | i
+(7 rows)
+
+drop table idxpart, idxpart1, idxpart2, idxpart3;
+select relname, relkind from pg_class where relname like 'idxpart%';
+ relname | relkind
+---------+---------
+(0 rows)
+
+-- Make sure the partition columns are mapped correctly
+create table idxpart (a int, b int, c text) partition by range (a);
+create index idxparti on idxpart (a);
+create index idxparti2 on idxpart (c, b);
+create table idxpart1 (c text, a int, b int);
+alter table idxpart attach partition idxpart1 for values from (0) to (10);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ c | text | | |
+ a | integer | | |
+ b | integer | | |
+Partition of: idxpart FOR VALUES FROM (0) TO (10)
+Indexes:
+ "idxpart1_a_idx" btree (a)
+ "idxpart1_c_b_idx" btree (c, b)
+
+drop table idxpart;
+-- Column number mapping: dropped columns in the partition
+create table idxpart1 (drop_1 int, drop_2 int, col_keep int, drop_3 int);
+alter table idxpart1 drop column drop_1;
+alter table idxpart1 drop column drop_2;
+alter table idxpart1 drop column drop_3;
+create index on idxpart1 (col_keep);
+create table idxpart (col_keep int) partition by range (col_keep);
+create index on idxpart (col_keep);
+alter table idxpart attach partition idxpart1 for values from (0) to (1000);
+\d idxpart
+ Table "public.idxpart"
+ Column | Type | Collation | Nullable | Default
+----------+---------+-----------+----------+---------
+ col_keep | integer | | |
+Partition key: RANGE (col_keep)
+Indexes:
+ "idxpart_col_keep_idx" btree (col_keep)
+Number of partitions: 1 (Use \d+ to list them.)
+
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+----------+---------+-----------+----------+---------
+ col_keep | integer | | |
+Partition of: idxpart FOR VALUES FROM (0) TO (1000)
+Indexes:
+ "idxpart1_col_keep_idx" btree (col_keep)
+
+select attrelid::regclass, attname, attnum from pg_attribute
+ where attrelid::regclass::text like 'idxpart%' and attnum > 0
+ order by attrelid::regclass, attnum;
+ attrelid | attname | attnum
+-----------------------+------------------------------+--------
+ idxpart1 | ........pg.dropped.1........ | 1
+ idxpart1 | ........pg.dropped.2........ | 2
+ idxpart1 | col_keep | 3
+ idxpart1 | ........pg.dropped.4........ | 4
+ idxpart1_col_keep_idx | col_keep | 1
+ idxpart | col_keep | 1
+ idxpart_col_keep_idx | col_keep | 1
+(7 rows)
+
+drop table idxpart;
+-- Column number mapping: dropped columns in the parent table
+create table idxpart(drop_1 int, drop_2 int, col_keep int, drop_3 int) partition by range (col_keep);
+alter table idxpart drop column drop_1;
+alter table idxpart drop column drop_2;
+alter table idxpart drop column drop_3;
+create table idxpart1 (col_keep int);
+create index on idxpart1 (col_keep);
+create index on idxpart (col_keep);
+alter table idxpart attach partition idxpart1 for values from (0) to (1000);
+\d idxpart
+ Table "public.idxpart"
+ Column | Type | Collation | Nullable | Default
+----------+---------+-----------+----------+---------
+ col_keep | integer | | |
+Partition key: RANGE (col_keep)
+Indexes:
+ "idxpart_col_keep_idx" btree (col_keep)
+Number of partitions: 1 (Use \d+ to list them.)
+
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+----------+---------+-----------+----------+---------
+ col_keep | integer | | |
+Partition of: idxpart FOR VALUES FROM (0) TO (1000)
+Indexes:
+ "idxpart1_col_keep_idx" btree (col_keep)
+
+select attrelid::regclass, attname, attnum from pg_attribute
+ where attrelid::regclass::text like 'idxpart%' and attnum > 0
+ order by attrelid::regclass, attnum;
+ attrelid | attname | attnum
+-----------------------+------------------------------+--------
+ idxpart | ........pg.dropped.1........ | 1
+ idxpart | ........pg.dropped.2........ | 2
+ idxpart | col_keep | 3
+ idxpart | ........pg.dropped.4........ | 4
+ idxpart1 | col_keep | 1
+ idxpart1_col_keep_idx | col_keep | 1
+ idxpart_col_keep_idx | col_keep | 1
+(7 rows)
+
+drop table idxpart;
+-- intentionally leave some objects around
+create table idxpart (a int) partition by range (a);
+create table idxpart1 partition of idxpart for values from (0) to (100);
+create table idxpart2 partition of idxpart for values from (100) to (1000)
+ partition by range (a);
+create table idxpart21 partition of idxpart2 for values from (100) to (200);
+create table idxpart22 partition of idxpart2 for values from (200) to (300);
+create index on idxpart22 (a);
+create index on only idxpart2 (a);
+alter index idxpart2_a_idx attach partition idxpart22_a_idx;
+create index on idxpart (a);
diff --git a/src/test/regress/parallel_schedule b/src/test/regress/parallel_schedule
index e224977791..ad9434fb87 100644
--- a/src/test/regress/parallel_schedule
+++ b/src/test/regress/parallel_schedule
@@ -116,7 +116,7 @@ test: plancache limit plpgsql copy2 temp domain rangefuncs prepare without_oid c
# ----------
# Another group of parallel tests
# ----------
-test: identity partition_join partition_prune reloptions hash_part
+test: identity partition_join partition_prune reloptions hash_part indexing
# event triggers cannot run concurrently with any test that runs DDL
test: event_trigger
diff --git a/src/test/regress/serial_schedule b/src/test/regress/serial_schedule
index 9fc5f1a268..27cd49845e 100644
--- a/src/test/regress/serial_schedule
+++ b/src/test/regress/serial_schedule
@@ -184,5 +184,6 @@ test: partition_join
test: partition_prune
test: reloptions
test: hash_part
+test: indexing
test: event_trigger
test: stats
diff --git a/src/test/regress/sql/alter_table.sql b/src/test/regress/sql/alter_table.sql
index 02a33ca7c4..af25ee9e77 100644
--- a/src/test/regress/sql/alter_table.sql
+++ b/src/test/regress/sql/alter_table.sql
@@ -1330,6 +1330,22 @@ create table tab1 (a int, b text);
create table tab2 (x int, y tab1);
alter table tab1 alter column b type varchar; -- fails
+-- Alter column type that's part of a partitioned index
+create table at_partitioned (a int, b text) partition by range (a);
+create table at_part_1 partition of at_partitioned for values from (0) to (1000);
+insert into at_partitioned values (512, '0.123');
+create table at_part_2 (b text, a int);
+insert into at_part_2 values ('1.234', 1024);
+create index on at_partitioned (b);
+create index on at_partitioned (a);
+\d at_part_1
+\d at_part_2
+alter table at_partitioned attach partition at_part_2 for values from (1000) to (2000);
+\d at_part_2
+alter table at_partitioned alter column b type numeric using b::numeric;
+\d at_part_1
+\d at_part_2
+
-- disallow recursive containment of row types
create temp table recur1 (f1 int);
alter table recur1 add column f2 recur1; -- fails
diff --git a/src/test/regress/sql/indexing.sql b/src/test/regress/sql/indexing.sql
new file mode 100644
index 0000000000..7cee9599b6
--- /dev/null
+++ b/src/test/regress/sql/indexing.sql
@@ -0,0 +1,238 @@
+-- Creating an index on a partitioned table makes the partitions
+-- automatically get the index
+create table idxpart (a int, b int, c text) partition by range (a);
+create table idxpart1 partition of idxpart for values from (0) to (10);
+create table idxpart2 partition of idxpart for values from (10) to (100)
+ partition by range (b);
+create table idxpart21 partition of idxpart2 for values from (0) to (100);
+create index on idxpart (a);
+select relname, relkind, indparentidx::regclass
+ from pg_class left join pg_index on (indexrelid = oid)
+ where relname like 'idxpart%' order by relname;
+drop table idxpart;
+
+-- Some unsupported features
+create table idxpart (a int, b int, c text) partition by range (a);
+create table idxpart1 partition of idxpart for values from (0) to (10);
+create unique index on idxpart (a);
+create index concurrently on idxpart (a);
+drop table idxpart;
+
+-- If a table without index is attached as partition to a table with
+-- an index, the index is automatically created
+create table idxpart (a int, b int, c text) partition by range (a);
+create index idxparti on idxpart (a);
+create index idxparti2 on idxpart (b, c);
+create table idxpart1 (like idxpart);
+\d idxpart1
+alter table idxpart attach partition idxpart1 for values from (0) to (10);
+\d idxpart1
+drop table idxpart;
+
+-- If a partition already has an index, don't create a duplicative one
+create table idxpart (a int, b int) partition by range (a, b);
+create table idxpart1 partition of idxpart for values from (0, 0) to (10, 10);
+create index on idxpart1 (a, b);
+create index on idxpart (a, b);
+\d idxpart1
+select indexrelid::regclass, indrelid::regclass, indparentidx::regclass
+ from pg_index where indexrelid::regclass::text like 'idxpart%'
+ order by indexrelid::regclass;
+drop table idxpart;
+
+-- DROP behavior for partitioned indexes
+create table idxpart (a int) partition by range (a);
+create index on idxpart (a);
+create table idxpart1 partition of idxpart for values from (0) to (10);
+drop index idxpart1_a_idx; -- no way
+drop index idxpart_a_idx; -- both indexes go away
+select relname, relkind from pg_class
+ where relname like 'idxpart%' order by relname;
+create index on idxpart (a);
+drop table idxpart1; -- the index on partition goes away too
+select relname, relkind from pg_class
+ where relname like 'idxpart%' order by relname;
+drop table idxpart;
+
+-- ALTER INDEX .. ATTACH, error cases
+create table idxpart (a int, b int) partition by range (a, b);
+create table idxpart1 partition of idxpart for values from (0, 0) to (10, 10);
+create index idxpart_a_b_idx on only idxpart (a, b);
+create index idxpart1_a_b_idx on idxpart1 (a, b);
+create index idxpart1_tst1 on idxpart1 (b, a);
+create index idxpart1_tst2 on idxpart1 using hash (a);
+create index idxpart1_tst3 on idxpart1 (a, b) where a > 10;
+
+alter index idxpart attach partition idxpart1;
+alter index idxpart_a_b_idx attach partition idxpart1;
+alter index idxpart_a_b_idx attach partition idxpart_a_b_idx;
+alter index idxpart_a_b_idx attach partition idxpart1_b_idx;
+alter index idxpart_a_b_idx attach partition idxpart1_tst1;
+alter index idxpart_a_b_idx attach partition idxpart1_tst2;
+alter index idxpart_a_b_idx attach partition idxpart1_tst3;
+-- OK
+alter index idxpart_a_b_idx attach partition idxpart1_a_b_idx;
+alter index idxpart_a_b_idx attach partition idxpart1_a_b_idx; -- quiet
+
+-- reject dupe
+create index idxpart1_2_a_b on idxpart1 (a, b);
+alter index idxpart_a_b_idx attach partition idxpart1_2_a_b;
+drop table idxpart;
+-- make sure everything's gone
+select indexrelid::regclass, indrelid::regclass, indparentidx::regclass
+ from pg_index where indexrelid::regclass::text like 'idxpart%';
+
+-- Don't auto-attach incompatible indexes
+create table idxpart (a int, b int) partition by range (a);
+create table idxpart1 (a int, b int);
+create index on idxpart1 using hash (a);
+create index on idxpart1 (a) where b > 1;
+create index on idxpart1 ((a + 0));
+create index on idxpart1 (a, a);
+create index on idxpart (a);
+alter table idxpart attach partition idxpart1 for values from (0) to (1000);
+\d idxpart1
+drop table idxpart;
+
+-- If CREATE INDEX ONLY, don't create indexes on partitions; and existing
+-- indexes on partitions don't change parent. ALTER INDEX ATTACH can change
+-- the parent after the fact.
+create table idxpart (a int) partition by range (a);
+create table idxpart1 partition of idxpart for values from (0) to (100);
+create table idxpart2 partition of idxpart for values from (100) to (1000)
+ partition by range (a);
+create table idxpart21 partition of idxpart2 for values from (100) to (200);
+create table idxpart22 partition of idxpart2 for values from (200) to (300);
+create index on idxpart22 (a);
+create index on only idxpart2 (a);
+create index on idxpart (a);
+-- Here we expect that idxpart1 and idxpart2 have a new index, but idxpart21
+-- does not; also, idxpart22 is not attached.
+\d idxpart1
+\d idxpart2
+\d idxpart21
+select indexrelid::regclass, indrelid::regclass, indparentidx::regclass
+ from pg_index where indexrelid::regclass::text like 'idxpart%'
+ order by indrelid::regclass;
+alter index idxpart2_a_idx attach partition idxpart22_a_idx;
+select indexrelid::regclass, indrelid::regclass, indparentidx::regclass
+ from pg_index where indexrelid::regclass::text like 'idxpart%'
+ order by indrelid::regclass;
+-- attaching idxpart22 is not enough to set idxpart22_a_idx valid ...
+alter index idxpart2_a_idx attach partition idxpart22_a_idx;
+\d idxpart2
+-- ... but this one is.
+create index on idxpart21 (a);
+alter index idxpart2_a_idx attach partition idxpart21_a_idx;
+\d idxpart2
+drop table idxpart;
+
+-- When a table is attached a partition and it already has an index, a
+-- duplicate index should not get created, but rather the index becomes
+-- attached to the parent's index.
+create table idxpart (a int, b int, c text) partition by range (a);
+create index idxparti on idxpart (a);
+create index idxparti2 on idxpart (b, c);
+create table idxpart1 (like idxpart including indexes);
+\d idxpart1
+select relname, relkind, indparentidx::regclass
+ from pg_class left join pg_index on (indexrelid = oid)
+ where relname like 'idxpart%' order by relname;
+alter table idxpart attach partition idxpart1 for values from (0) to (10);
+\d idxpart1
+select relname, relkind, indparentidx::regclass
+ from pg_class left join pg_index on (indexrelid = oid)
+ where relname like 'idxpart%' order by relname;
+drop table idxpart;
+
+-- verify dependency handling during ALTER TABLE DETACH PARTITION
+create table idxpart (a int) partition by range (a);
+create table idxpart1 (like idxpart);
+create index on idxpart1 (a);
+create index on idxpart (a);
+create table idxpart2 (like idxpart);
+alter table idxpart attach partition idxpart1 for values from (0000) to (1000);
+alter table idxpart attach partition idxpart2 for values from (1000) to (2000);
+create table idxpart3 partition of idxpart for values from (2000) to (3000);
+select relname, relkind from pg_class where relname like 'idxpart%';
+-- a) after detaching partitions, the indexes can be dropped independently
+alter table idxpart detach partition idxpart1;
+alter table idxpart detach partition idxpart2;
+alter table idxpart detach partition idxpart3;
+drop index idxpart1_a_idx;
+drop index idxpart2_a_idx;
+drop index idxpart3_a_idx;
+select relname, relkind from pg_class where relname like 'idxpart%';
+drop table idxpart, idxpart1, idxpart2, idxpart3;
+select relname, relkind from pg_class where relname like 'idxpart%';
+
+create table idxpart (a int) partition by range (a);
+create table idxpart1 (like idxpart);
+create index on idxpart1 (a);
+create index on idxpart (a);
+create table idxpart2 (like idxpart);
+alter table idxpart attach partition idxpart1 for values from (0000) to (1000);
+alter table idxpart attach partition idxpart2 for values from (1000) to (2000);
+create table idxpart3 partition of idxpart for values from (2000) to (3000);
+-- b) after detaching, dropping the index on parent does not remove the others
+select relname, relkind from pg_class where relname like 'idxpart%';
+alter table idxpart detach partition idxpart1;
+alter table idxpart detach partition idxpart2;
+alter table idxpart detach partition idxpart3;
+drop index idxpart_a_idx;
+select relname, relkind from pg_class where relname like 'idxpart%';
+drop table idxpart, idxpart1, idxpart2, idxpart3;
+select relname, relkind from pg_class where relname like 'idxpart%';
+
+-- Make sure the partition columns are mapped correctly
+create table idxpart (a int, b int, c text) partition by range (a);
+create index idxparti on idxpart (a);
+create index idxparti2 on idxpart (c, b);
+create table idxpart1 (c text, a int, b int);
+alter table idxpart attach partition idxpart1 for values from (0) to (10);
+\d idxpart1
+drop table idxpart;
+
+-- Column number mapping: dropped columns in the partition
+create table idxpart1 (drop_1 int, drop_2 int, col_keep int, drop_3 int);
+alter table idxpart1 drop column drop_1;
+alter table idxpart1 drop column drop_2;
+alter table idxpart1 drop column drop_3;
+create index on idxpart1 (col_keep);
+create table idxpart (col_keep int) partition by range (col_keep);
+create index on idxpart (col_keep);
+alter table idxpart attach partition idxpart1 for values from (0) to (1000);
+\d idxpart
+\d idxpart1
+select attrelid::regclass, attname, attnum from pg_attribute
+ where attrelid::regclass::text like 'idxpart%' and attnum > 0
+ order by attrelid::regclass, attnum;
+drop table idxpart;
+
+-- Column number mapping: dropped columns in the parent table
+create table idxpart(drop_1 int, drop_2 int, col_keep int, drop_3 int) partition by range (col_keep);
+alter table idxpart drop column drop_1;
+alter table idxpart drop column drop_2;
+alter table idxpart drop column drop_3;
+create table idxpart1 (col_keep int);
+create index on idxpart1 (col_keep);
+create index on idxpart (col_keep);
+alter table idxpart attach partition idxpart1 for values from (0) to (1000);
+\d idxpart
+\d idxpart1
+select attrelid::regclass, attname, attnum from pg_attribute
+ where attrelid::regclass::text like 'idxpart%' and attnum > 0
+ order by attrelid::regclass, attnum;
+drop table idxpart;
+
+-- intentionally leave some objects around
+create table idxpart (a int) partition by range (a);
+create table idxpart1 partition of idxpart for values from (0) to (100);
+create table idxpart2 partition of idxpart for values from (100) to (1000)
+ partition by range (a);
+create table idxpart21 partition of idxpart2 for values from (100) to (200);
+create table idxpart22 partition of idxpart2 for values from (200) to (300);
+create index on idxpart22 (a);
+create index on only idxpart2 (a);
+alter index idxpart2_a_idx attach partition idxpart22_a_idx;
+create index on idxpart (a);
--
2.11.0
v2-0002-allow-indexes-on-partitioned-tables-to-be-unique.patchtext/plain; charset=us-asciiDownload
From b92ccc654f92f068b90a0aed2478fed5679c91a6 Mon Sep 17 00:00:00 2001
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
Date: Mon, 6 Nov 2017 17:04:55 +0100
Subject: [PATCH v2 2/2] allow indexes on partitioned tables to be unique
---
doc/src/sgml/ref/alter_table.sgml | 9 +-
doc/src/sgml/ref/create_table.sgml | 16 ++-
src/backend/bootstrap/bootparse.y | 2 +
src/backend/catalog/index.c | 28 ++++-
src/backend/catalog/toasting.c | 4 +-
src/backend/commands/indexcmds.c | 80 +++++++++++++--
src/backend/commands/tablecmds.c | 12 ++-
src/backend/parser/analyze.c | 7 ++
src/backend/parser/parse_utilcmd.c | 31 +-----
src/backend/tcop/utility.c | 1 +
src/include/catalog/index.h | 5 +-
src/include/commands/defrem.h | 1 +
src/include/parser/parse_utilcmd.h | 3 +-
src/test/regress/expected/alter_table.out | 8 --
src/test/regress/expected/create_table.out | 12 ---
src/test/regress/expected/indexing.out | 142 +++++++++++++++++++++++++-
src/test/regress/expected/insert_conflict.out | 2 +-
src/test/regress/sql/alter_table.sql | 2 -
src/test/regress/sql/create_table.sql | 8 --
src/test/regress/sql/indexing.sql | 73 ++++++++++++-
20 files changed, 363 insertions(+), 83 deletions(-)
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index 0a2f3e3646..ee6a45c9ad 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -782,8 +782,9 @@ ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
This form attaches an existing table (which might itself be partitioned)
as a partition of the target table. The table can be attached
as a partition for specific values using <literal>FOR VALUES
- </literal> or as a default partition by using <literal>DEFAULT
- </literal>. For each index in the target table, a corresponding
+ </literal> or as a default partition by using
+ <literal>DEFAULT</literal>.
+ For each index in the target table, a corresponding
one will be created in the attached table; or, if an equivalent
index already exists, will be attached to the target table's index,
as if <command>ALTER INDEX ATTACH</command> had been executed.
@@ -798,8 +799,10 @@ ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
as the target table and no more; moreover, the column types must also
match. Also, it must have all the <literal>NOT NULL</literal> and
<literal>CHECK</literal> constraints of the target table. Currently
- <literal>UNIQUE</literal>, <literal>PRIMARY KEY</literal>, and
<literal>FOREIGN KEY</literal> constraints are not considered.
+ <literal>UNIQUE</literal> and <literal>PRIMARY KEY</literal> constraints
+ from the parent table will be created in the partition, if they don't
+ already exist.
If any of the <literal>CHECK</literal> constraints of the table being
attached is marked <literal>NO INHERIT</literal>, the command will fail;
such a constraint must be recreated without the <literal>NO INHERIT</literal>
diff --git a/doc/src/sgml/ref/create_table.sgml b/doc/src/sgml/ref/create_table.sgml
index a0c9a6d257..4c56df8960 100644
--- a/doc/src/sgml/ref/create_table.sgml
+++ b/doc/src/sgml/ref/create_table.sgml
@@ -546,8 +546,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</para>
<para>
- Partitioned tables do not support <literal>UNIQUE</literal>,
- <literal>PRIMARY KEY</literal>, <literal>EXCLUDE</literal>, or
+ Partitioned tables do not support
+ <literal>EXCLUDE</literal>, or
<literal>FOREIGN KEY</literal> constraints; however, you can define
these constraints on individual partitions.
</para>
@@ -786,6 +786,11 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
primary key constraint defined for the table. (Otherwise it
would just be the same constraint listed twice.)
</para>
+
+ <para>
+ When used on partitioned tables, <literal>UNIQUE</literal> constraints
+ must include all the columns of the partition key.
+ </para>
</listitem>
</varlistentry>
@@ -814,6 +819,13 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
about the design of the schema, since a primary key implies that other
tables can rely on this set of columns as a unique identifier for rows.
</para>
+
+ <para>
+ <literal>PRIMARY KEY</literal> constraints share the restrictions that
+ <literal>UNIQUE</literal> constraints have when placed on partitioned
+ tables.
+ </para>
+
</listitem>
</varlistentry>
diff --git a/src/backend/bootstrap/bootparse.y b/src/backend/bootstrap/bootparse.y
index dfd53fa054..9e81f9514d 100644
--- a/src/backend/bootstrap/bootparse.y
+++ b/src/backend/bootstrap/bootparse.y
@@ -322,6 +322,7 @@ Boot_DeclareIndexStmt:
stmt,
$4,
InvalidOid,
+ InvalidOid,
false,
false,
false,
@@ -367,6 +368,7 @@ Boot_DeclareUniqueIndexStmt:
stmt,
$5,
InvalidOid,
+ InvalidOid,
false,
false,
false,
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index f34127c81b..047988c980 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -677,6 +677,8 @@ UpdateIndexRelation(Oid indexoid,
* nonzero to specify a preselected OID.
* parentIndexRelid: if creating an index partition, the OID of the
* parent index; otherwise InvalidOid.
+ * parentConstraintId: if creating a constraint on a partition, the OID
+ * of the constraint in the parent; otherwise InvalidOid.
* relFileNode: normally, pass InvalidOid to get new storage. May be
* nonzero to attach an existing valid build.
* indexInfo: same info executor uses to insert into the index
@@ -708,6 +710,7 @@ UpdateIndexRelation(Oid indexoid,
* (only if INDEX_CREATE_ADD_CONSTRAINT is set)
* allow_system_table_mods: allow table to be a system catalog
* is_internal: if true, post creation hook for new index
+ * constraintId: if not NULL, receives OID of created constraint
*
* Returns the OID of the created index.
*/
@@ -716,6 +719,7 @@ index_create(Relation heapRelation,
const char *indexRelationName,
Oid indexRelationId,
Oid parentIndexRelid,
+ Oid parentConstraintId,
Oid relFileNode,
IndexInfo *indexInfo,
List *indexColNames,
@@ -728,7 +732,8 @@ index_create(Relation heapRelation,
bits16 flags,
bits16 constr_flags,
bool allow_system_table_mods,
- bool is_internal)
+ bool is_internal,
+ Oid *constraintId)
{
Oid heapRelationId = RelationGetRelid(heapRelation);
Relation pg_class;
@@ -971,6 +976,7 @@ index_create(Relation heapRelation,
if ((flags & INDEX_CREATE_ADD_CONSTRAINT) != 0)
{
char constraintType;
+ ObjectAddress localaddr;
if (isprimary)
constraintType = CONSTRAINT_PRIMARY;
@@ -984,14 +990,17 @@ index_create(Relation heapRelation,
constraintType = 0; /* keep compiler quiet */
}
- index_constraint_create(heapRelation,
+ localaddr = index_constraint_create(heapRelation,
indexRelationId,
+ parentConstraintId,
indexInfo,
indexRelationName,
constraintType,
constr_flags,
allow_system_table_mods,
is_internal);
+ if (constraintId)
+ *constraintId = localaddr.objectId;
}
else
{
@@ -1163,6 +1172,8 @@ index_create(Relation heapRelation,
*
* heapRelation: table owning the index (must be suitably locked by caller)
* indexRelationId: OID of the index
+ * parentConstraintId: if constraint is on a partition, the OID of the
+ * constraint in the parent.
* indexInfo: same info executor uses to insert into the index
* constraintName: what it say (generally, should match name of index)
* constraintType: one of CONSTRAINT_PRIMARY, CONSTRAINT_UNIQUE, or
@@ -1180,6 +1191,7 @@ index_create(Relation heapRelation,
ObjectAddress
index_constraint_create(Relation heapRelation,
Oid indexRelationId,
+ Oid parentConstraintId,
IndexInfo *indexInfo,
const char *constraintName,
char constraintType,
@@ -1277,6 +1289,18 @@ index_constraint_create(Relation heapRelation,
recordDependencyOn(&myself, &referenced, DEPENDENCY_INTERNAL);
/*
+ * Also, if this is a constraint on a partition, mark it as depending
+ * on the constraint in the parent.
+ */
+ if (OidIsValid(parentConstraintId))
+ {
+ ObjectAddress third;
+
+ ObjectAddressSet(third, ConstraintRelationId, parentConstraintId);
+ recordDependencyOn(&referenced, &third, DEPENDENCY_INTERNAL);
+ }
+
+ /*
* If the constraint is deferrable, create the deferred uniqueness
* checking trigger. (The trigger will be given an internal dependency on
* the constraint by CreateTrigger.)
diff --git a/src/backend/catalog/toasting.c b/src/backend/catalog/toasting.c
index cf37011b73..f4e7b83fee 100644
--- a/src/backend/catalog/toasting.c
+++ b/src/backend/catalog/toasting.c
@@ -329,13 +329,13 @@ create_toast_table(Relation rel, Oid toastOid, Oid toastIndexOid,
coloptions[1] = 0;
index_create(toast_rel, toast_idxname, toastIndexOid, InvalidOid,
- InvalidOid,
+ InvalidOid, InvalidOid,
indexInfo,
list_make2("chunk_id", "chunk_seq"),
BTREE_AM_OID,
rel->rd_rel->reltablespace,
collationObjectId, classObjectId, coloptions, (Datum) 0,
- INDEX_CREATE_IS_PRIMARY, 0, true, true);
+ INDEX_CREATE_IS_PRIMARY, 0, true, true, NULL);
heap_close(toast_rel, NoLock);
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index c1cf83b045..0713e12954 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -299,6 +299,8 @@ CheckIndexCompatible(Oid oldId,
* nonzero to specify a preselected OID for the index.
* 'parentIndexId': the OID of the parent index; InvalidOid if not the child
* of a partitioned index.
+ * 'parentConstraintId': the OID of the parent constraint; InvalidOid if not
+ * the child of a constraint (only used when recursing)
* 'is_alter_table': this is due to an ALTER rather than a CREATE operation.
* 'check_rights': check for CREATE rights in namespace and tablespace. (This
* should be true except when ALTER is deleting/recreating an index.)
@@ -315,6 +317,7 @@ DefineIndex(Oid relationId,
IndexStmt *stmt,
Oid indexRelationId,
Oid parentIndexId,
+ Oid parentConstraintId,
bool is_alter_table,
bool check_rights,
bool check_not_in_use,
@@ -329,6 +332,7 @@ DefineIndex(Oid relationId,
Oid accessMethodId;
Oid namespaceId;
Oid tablespaceId;
+ Oid createdConstraintId;
List *indexColNames;
Relation rel;
Relation indexRelation;
@@ -426,20 +430,11 @@ DefineIndex(Oid relationId,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot create index on partitioned table \"%s\" concurrently",
RelationGetRelationName(rel))));
- if (stmt->unique)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot create unique index on partitioned table \"%s\"",
- RelationGetRelationName(rel))));
if (stmt->excludeOpNames)
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot create exclusion constraints on partitioned table \"%s\"",
RelationGetRelationName(rel))));
- if (stmt->primary || stmt->isconstraint)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot create constraints on partitioned tables")));
}
/*
@@ -637,6 +632,68 @@ DefineIndex(Oid relationId,
index_check_primary_key(rel, indexInfo, is_alter_table);
/*
+ * If this table is partitioned and we're creating a unique index or a
+ * primary key, make sure that the indexed columns are part of the
+ * partition key. Otherwise it would be possible to violate uniqueness by
+ * putting values that ought to be unique in different partitions.
+ *
+ * We could lift this limitation if we had global indexes, but those have
+ * their own problems, so this is a useful feature combination.
+ */
+ if (partitioned && (stmt->unique || stmt->primary))
+ {
+ PartitionKey key = rel->rd_partkey;
+ int i;
+
+ /*
+ * A partitioned table can have unique indexes, as long as all the
+ * columns in the partition key appear in the unique key. A
+ * partition-local index can enforce global uniqueness iff the PK
+ * value completely determines the partition that a row is in.
+ *
+ * Thus, verify that all the columns in the partition key appear
+ * in the unique key definition.
+ */
+ for (i = 0; i < key->partnatts; i++)
+ {
+ bool found = false;
+ int j;
+
+ /*
+ * It may be possible to support UNIQUE constraints when partition
+ * keys are expressions, but is it worth it? Give up for now.
+ */
+ if (key->partattrs[i] == 0)
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ /* XXX reformulate error message? */
+ errmsg("UNIQUE constraints are not supported on partitioned tables using expressions as partition keys")));
+
+ for (j = 0; j < indexInfo->ii_NumIndexAttrs; j++)
+ {
+ if (key->partattrs[i] == indexInfo->ii_KeyAttrNumbers[j])
+ {
+ found = true;
+ break;
+ }
+ }
+ if (!found)
+ {
+ Form_pg_attribute att;
+
+ att = TupleDescAttr(RelationGetDescr(rel), key->partattrs[i] - 1);
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("insufficient columns in UNIQUE constraint definition"),
+ errdetail("UNIQUE constraint on table \"%s\" does not include column \"%s\" which is part of the partition key.",
+ RelationGetRelationName(rel),
+ NameStr(att->attname))));
+ }
+ }
+ }
+
+
+ /*
* We disallow indexes on system columns other than OID. They would not
* necessarily get updated correctly, and they don't seem useful anyway.
*/
@@ -733,12 +790,14 @@ DefineIndex(Oid relationId,
indexRelationId =
index_create(rel, indexRelationName, indexRelationId, parentIndexId,
+ parentConstraintId,
stmt->oldNode, indexInfo, indexColNames,
accessMethodId, tablespaceId,
collationObjectId, classObjectId,
coloptions, reloptions,
flags, constr_flags,
- allowSystemTableMods, !check_rights);
+ allowSystemTableMods, !check_rights,
+ &createdConstraintId);
ObjectAddressSet(address, RelationRelationId, indexRelationId);
@@ -850,6 +909,7 @@ DefineIndex(Oid relationId,
DefineIndex(childRelid, childStmt,
InvalidOid, /* no predefined OID */
indexRelationId, /* this is our child */
+ createdConstraintId,
false, check_rights, check_not_in_use,
false, quiet);
}
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 4588d07fda..3dd91d6555 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -938,17 +938,20 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
Relation idxRel = index_open(lfirst_oid(cell), AccessShareLock);
AttrNumber *attmap;
IndexStmt *idxstmt;
+ Oid constraintOid;
attmap = convert_tuples_by_name_map(RelationGetDescr(rel),
RelationGetDescr(parent),
gettext_noop("could not convert row type"));
idxstmt =
generateClonedIndexStmt(NULL, RelationGetRelid(rel), idxRel,
- attmap, RelationGetDescr(rel)->natts);
+ attmap, RelationGetDescr(rel)->natts,
+ &constraintOid);
DefineIndex(RelationGetRelid(rel),
idxstmt,
InvalidOid,
RelationGetRelid(idxRel),
+ constraintOid,
false, false, false, false, false);
index_close(idxRel, AccessShareLock);
@@ -6877,6 +6880,7 @@ ATExecAddIndex(AlteredTableInfo *tab, Relation rel,
stmt,
InvalidOid, /* no predefined OID */
InvalidOid, /* no parent index */
+ InvalidOid, /* no parent constraint */
true, /* is_alter_table */
check_rights,
false, /* check_not_in_use - we did it already */
@@ -6969,6 +6973,7 @@ ATExecAddIndexConstraint(AlteredTableInfo *tab, Relation rel,
address = index_constraint_create(rel,
index_oid,
+ InvalidOid,
indexInfo,
constraintName,
constraintType,
@@ -14170,12 +14175,15 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
if (!found)
{
IndexStmt *stmt;
+ Oid constraintOid;
stmt = generateClonedIndexStmt(NULL, RelationGetRelid(attachrel),
idxRel, attmap,
- RelationGetDescr(rel)->natts);
+ RelationGetDescr(rel)->natts,
+ &constraintOid);
DefineIndex(RelationGetRelid(attachrel), stmt, InvalidOid,
RelationGetRelid(idxRel),
+ constraintOid,
false, false, false, false, false);
}
diff --git a/src/backend/parser/analyze.c b/src/backend/parser/analyze.c
index e7b2bc7e73..5b3a610cf9 100644
--- a/src/backend/parser/analyze.c
+++ b/src/backend/parser/analyze.c
@@ -1017,6 +1017,13 @@ transformOnConflictClause(ParseState *pstate,
TargetEntry *te;
int attno;
+ if (targetrel->rd_partdesc)
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("%s cannot be applied to partitioned table \"%s\"",
+ "ON CONFLICT DO UPDATE",
+ RelationGetRelationName(targetrel))));
+
/*
* All INSERT expressions have been parsed, get ready for potentially
* existing SET statements that need to be processed like an UPDATE.
diff --git a/src/backend/parser/parse_utilcmd.c b/src/backend/parser/parse_utilcmd.c
index 90bb356df8..e1ded02ecc 100644
--- a/src/backend/parser/parse_utilcmd.c
+++ b/src/backend/parser/parse_utilcmd.c
@@ -704,12 +704,6 @@ transformColumnDefinition(CreateStmtContext *cxt, ColumnDef *column)
errmsg("primary key constraints are not supported on foreign tables"),
parser_errposition(cxt->pstate,
constraint->location)));
- if (cxt->ispartitioned)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("primary key constraints are not supported on partitioned tables"),
- parser_errposition(cxt->pstate,
- constraint->location)));
/* FALL THRU */
case CONSTR_UNIQUE:
@@ -719,12 +713,6 @@ transformColumnDefinition(CreateStmtContext *cxt, ColumnDef *column)
errmsg("unique constraints are not supported on foreign tables"),
parser_errposition(cxt->pstate,
constraint->location)));
- if (cxt->ispartitioned)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("unique constraints are not supported on partitioned tables"),
- parser_errposition(cxt->pstate,
- constraint->location)));
if (constraint->keys == NIL)
constraint->keys = list_make1(makeString(column->colname));
cxt->ixconstraints = lappend(cxt->ixconstraints, constraint);
@@ -821,12 +809,6 @@ transformTableConstraint(CreateStmtContext *cxt, Constraint *constraint)
errmsg("primary key constraints are not supported on foreign tables"),
parser_errposition(cxt->pstate,
constraint->location)));
- if (cxt->ispartitioned)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("primary key constraints are not supported on partitioned tables"),
- parser_errposition(cxt->pstate,
- constraint->location)));
cxt->ixconstraints = lappend(cxt->ixconstraints, constraint);
break;
@@ -837,12 +819,6 @@ transformTableConstraint(CreateStmtContext *cxt, Constraint *constraint)
errmsg("unique constraints are not supported on foreign tables"),
parser_errposition(cxt->pstate,
constraint->location)));
- if (cxt->ispartitioned)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("unique constraints are not supported on partitioned tables"),
- parser_errposition(cxt->pstate,
- constraint->location)));
cxt->ixconstraints = lappend(cxt->ixconstraints, constraint);
break;
@@ -1184,7 +1160,7 @@ transformTableLikeClause(CreateStmtContext *cxt, TableLikeClause *table_like_cla
/* Build CREATE INDEX statement to recreate the parent_index */
index_stmt = generateClonedIndexStmt(cxt->relation, InvalidOid,
parent_index,
- attmap, tupleDesc->natts);
+ attmap, tupleDesc->natts, NULL);
/* Copy comment on index, if requested */
if (table_like_clause->options & CREATE_TABLE_LIKE_COMMENTS)
@@ -1267,7 +1243,7 @@ transformOfType(CreateStmtContext *cxt, TypeName *ofTypename)
*/
IndexStmt *
generateClonedIndexStmt(RangeVar *heapRel, Oid heapRelid, Relation source_idx,
- const AttrNumber *attmap, int attmap_length)
+ const AttrNumber *attmap, int attmap_length, Oid *constraintOid)
{
Oid source_relid = RelationGetRelid(source_idx);
HeapTuple ht_idxrel;
@@ -1365,6 +1341,9 @@ generateClonedIndexStmt(RangeVar *heapRel, Oid heapRelid, Relation source_idx,
HeapTuple ht_constr;
Form_pg_constraint conrec;
+ if (constraintOid)
+ *constraintOid = constraintId;
+
ht_constr = SearchSysCache1(CONSTROID,
ObjectIdGetDatum(constraintId));
if (!HeapTupleIsValid(ht_constr))
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 16c4f8fad4..07d32ed90d 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1337,6 +1337,7 @@ ProcessUtilitySlow(ParseState *pstate,
stmt,
InvalidOid, /* no predefined OID */
InvalidOid, /* no parent index */
+ InvalidOid, /* no parent constraint */
false, /* is_alter_table */
true, /* check_rights */
true, /* check_not_in_use */
diff --git a/src/include/catalog/index.h b/src/include/catalog/index.h
index 139365c3b3..1e920c5c27 100644
--- a/src/include/catalog/index.h
+++ b/src/include/catalog/index.h
@@ -54,6 +54,7 @@ extern Oid index_create(Relation heapRelation,
const char *indexRelationName,
Oid indexRelationId,
Oid parentIndexRelid,
+ Oid parentConstraintId,
Oid relFileNode,
IndexInfo *indexInfo,
List *indexColNames,
@@ -66,7 +67,8 @@ extern Oid index_create(Relation heapRelation,
bits16 flags,
bits16 constr_flags,
bool allow_system_table_mods,
- bool is_internal);
+ bool is_internal,
+ Oid *constraintId);
#define INDEX_CONSTR_CREATE_MARK_AS_PRIMARY (1 << 0)
#define INDEX_CONSTR_CREATE_DEFERRABLE (1 << 1)
@@ -76,6 +78,7 @@ extern Oid index_create(Relation heapRelation,
extern ObjectAddress index_constraint_create(Relation heapRelation,
Oid indexRelationId,
+ Oid parentConstraintId,
IndexInfo *indexInfo,
const char *constraintName,
char constraintType,
diff --git a/src/include/commands/defrem.h b/src/include/commands/defrem.h
index 41007162aa..d1876d0b08 100644
--- a/src/include/commands/defrem.h
+++ b/src/include/commands/defrem.h
@@ -26,6 +26,7 @@ extern ObjectAddress DefineIndex(Oid relationId,
IndexStmt *stmt,
Oid indexRelationId,
Oid parentIndexId,
+ Oid parentConstraintId,
bool is_alter_table,
bool check_rights,
bool check_not_in_use,
diff --git a/src/include/parser/parse_utilcmd.h b/src/include/parser/parse_utilcmd.h
index 64aa8234e5..35ac97940a 100644
--- a/src/include/parser/parse_utilcmd.h
+++ b/src/include/parser/parse_utilcmd.h
@@ -29,6 +29,7 @@ extern PartitionBoundSpec *transformPartitionBound(ParseState *pstate, Relation
PartitionBoundSpec *spec);
extern IndexStmt *generateClonedIndexStmt(RangeVar *heapRel, Oid heapOid,
Relation source_idx,
- const AttrNumber *attmap, int attmap_length);
+ const AttrNumber *attmap, int attmap_length,
+ Oid *constraintOid);
#endif /* PARSE_UTILCMD_H */
diff --git a/src/test/regress/expected/alter_table.out b/src/test/regress/expected/alter_table.out
index 517fb080bd..2caf930242 100644
--- a/src/test/regress/expected/alter_table.out
+++ b/src/test/regress/expected/alter_table.out
@@ -3290,14 +3290,6 @@ CREATE TABLE partitioned (
a int,
b int
) PARTITION BY RANGE (a, (a+b+1));
-ALTER TABLE partitioned ADD UNIQUE (a);
-ERROR: unique constraints are not supported on partitioned tables
-LINE 1: ALTER TABLE partitioned ADD UNIQUE (a);
- ^
-ALTER TABLE partitioned ADD PRIMARY KEY (a);
-ERROR: primary key constraints are not supported on partitioned tables
-LINE 1: ALTER TABLE partitioned ADD PRIMARY KEY (a);
- ^
ALTER TABLE partitioned ADD FOREIGN KEY (a) REFERENCES blah;
ERROR: foreign key constraints are not supported on partitioned tables
LINE 1: ALTER TABLE partitioned ADD FOREIGN KEY (a) REFERENCES blah;
diff --git a/src/test/regress/expected/create_table.out b/src/test/regress/expected/create_table.out
index 8e745402ae..866cc99b9f 100644
--- a/src/test/regress/expected/create_table.out
+++ b/src/test/regress/expected/create_table.out
@@ -276,12 +276,6 @@ CREATE TABLE partitioned (
) PARTITION BY LIST (a1, a2); -- fail
ERROR: cannot use "list" partition strategy with more than one column
-- unsupported constraint type for partitioned tables
-CREATE TABLE partitioned (
- a int PRIMARY KEY
-) PARTITION BY RANGE (a);
-ERROR: primary key constraints are not supported on partitioned tables
-LINE 2: a int PRIMARY KEY
- ^
CREATE TABLE pkrel (
a int PRIMARY KEY
);
@@ -293,12 +287,6 @@ LINE 2: a int REFERENCES pkrel(a)
^
DROP TABLE pkrel;
CREATE TABLE partitioned (
- a int UNIQUE
-) PARTITION BY RANGE (a);
-ERROR: unique constraints are not supported on partitioned tables
-LINE 2: a int UNIQUE
- ^
-CREATE TABLE partitioned (
a int,
EXCLUDE USING gist (a WITH &&)
) PARTITION BY RANGE (a);
diff --git a/src/test/regress/expected/indexing.out b/src/test/regress/expected/indexing.out
index b64b466b11..96a1f6ebca 100644
--- a/src/test/regress/expected/indexing.out
+++ b/src/test/regress/expected/indexing.out
@@ -25,8 +25,6 @@ drop table idxpart;
-- Some unsupported features
create table idxpart (a int, b int, c text) partition by range (a);
create table idxpart1 partition of idxpart for values from (0) to (10);
-create unique index on idxpart (a);
-ERROR: cannot create unique index on partitioned table "idxpart"
create index concurrently on idxpart (a);
ERROR: cannot create index on partitioned table "idxpart" concurrently
drop table idxpart;
@@ -520,6 +518,146 @@ select attrelid::regclass, attname, attnum from pg_attribute
(7 rows)
drop table idxpart;
+--
+-- Constraint-related indexes
+--
+-- Verify that it works to add primary key / unique to partitioned tables
+create table idxpart (a int primary key, b int) partition by range (a);
+\d idxpart
+ Table "public.idxpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | not null |
+ b | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "idxpart_pkey" PRIMARY KEY, btree (a)
+Number of partitions: 0
+
+drop table idxpart;
+-- but not if you fail to use the full partition key
+create table idxpart (a int unique, b int) partition by range (a, b);
+ERROR: insufficient columns in UNIQUE constraint definition
+DETAIL: UNIQUE constraint on table "idxpart" does not include column "b" which is part of the partition key.
+create table idxpart (a int, b int unique) partition by range (a, b);
+ERROR: insufficient columns in UNIQUE constraint definition
+DETAIL: UNIQUE constraint on table "idxpart" does not include column "a" which is part of the partition key.
+create table idxpart (a int primary key, b int) partition by range (b, a);
+ERROR: insufficient columns in UNIQUE constraint definition
+DETAIL: UNIQUE constraint on table "idxpart" does not include column "b" which is part of the partition key.
+create table idxpart (a int, b int primary key) partition by range (b, a);
+ERROR: insufficient columns in UNIQUE constraint definition
+DETAIL: UNIQUE constraint on table "idxpart" does not include column "a" which is part of the partition key.
+-- OK if you use them in some other order
+create table idxpart (a int, b int, c text, primary key (a, b, c)) partition by range (b, c, a);
+drop table idxpart;
+create table idxpart (a int primary key, b int) partition by range ((b + a));
+ERROR: UNIQUE constraints are not supported on partitioned tables using expressions as partition keys
+-- not other types of index-based constraints
+create table idxpart (a int, exclude (a with = )) partition by range (a);
+ERROR: exclusion constraints are not supported on partitioned tables
+LINE 1: create table idxpart (a int, exclude (a with = )) partition ...
+ ^
+-- It works to add primary keys after the partitioned table is created
+create table idxpart (a int, b int, c text) partition by range (a, b);
+alter table idxpart add primary key (a); -- not an incomplete one tho
+ERROR: insufficient columns in UNIQUE constraint definition
+DETAIL: UNIQUE constraint on table "idxpart" does not include column "b" which is part of the partition key.
+alter table idxpart add primary key (a, b);
+\d idxpart
+ Table "public.idxpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | not null |
+ b | integer | | not null |
+ c | text | | |
+Partition key: RANGE (a, b)
+Indexes:
+ "idxpart_pkey" PRIMARY KEY, btree (a, b)
+Number of partitions: 0
+
+create table idxpart1 partition of idxpart for values from (0, 0) to (1000, 1000);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | not null |
+ b | integer | | not null |
+ c | text | | |
+Partition of: idxpart FOR VALUES FROM (0, 0) TO (1000, 1000)
+Indexes:
+ "idxpart1_pkey" PRIMARY KEY, btree (a, b)
+
+drop table idxpart;
+-- It works to add unique constraints after the partitioned table is created
+create table idxpart (a int, b int) partition by range (a, b);
+alter table idxpart add unique (a); -- ... nope
+ERROR: insufficient columns in UNIQUE constraint definition
+DETAIL: UNIQUE constraint on table "idxpart" does not include column "b" which is part of the partition key.
+alter table idxpart add unique (b, a);
+\d idxpart
+ Table "public.idxpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | integer | | |
+Partition key: RANGE (a, b)
+Indexes:
+ "idxpart_b_a_key" UNIQUE CONSTRAINT, btree (b, a)
+Number of partitions: 0
+
+drop table idxpart;
+-- Exclusion constraints cannot be added
+create table idxpart (a int, b int) partition by range (a);
+alter table idxpart add exclude (a with =);
+ERROR: exclusion constraints are not supported on partitioned tables
+LINE 1: alter table idxpart add exclude (a with =);
+ ^
+drop table idxpart;
+-- When (sub)partitions are created, they also contain the constraint
+create table idxpart (a int, b int, primary key (a, b)) partition by range (a, b);
+create table idxpart1 partition of idxpart for values from (1, 1) to (10, 10);
+create table idxpart2 partition of idxpart for values from (10, 10) to (20, 20)
+ partition by range (b);
+create table idxpart21 partition of idxpart2 for values from (10) to (15);
+create table idxpart22 partition of idxpart2 for values from (15) to (20);
+create table idxpart3 (b int not null, a int not null);
+alter table idxpart attach partition idxpart3 for values from (20, 20) to (30, 30);
+select conname, contype, conrelid::regclass, conindid::regclass, conkey
+ from pg_constraint where conrelid::regclass::text like 'idxpart%'
+ order by conname;
+ conname | contype | conrelid | conindid | conkey
+----------------+---------+-----------+----------------+--------
+ idxpart1_pkey | p | idxpart1 | idxpart1_pkey | {1,2}
+ idxpart21_pkey | p | idxpart21 | idxpart21_pkey | {1,2}
+ idxpart22_pkey | p | idxpart22 | idxpart22_pkey | {1,2}
+ idxpart2_pkey | p | idxpart2 | idxpart2_pkey | {1,2}
+ idxpart3_pkey | p | idxpart3 | idxpart3_pkey | {2,1}
+ idxpart_pkey | p | idxpart | idxpart_pkey | {1,2}
+(6 rows)
+
+drop table idxpart;
+-- multi-layer partitioning honors the prohibition. So this fails:
+create table idxpart (a int, b int, primary key (a)) partition by range (a);
+create table idxpart2 partition of idxpart for values from (0) to (1000) partition by range (b);
+ERROR: insufficient columns in UNIQUE constraint definition
+DETAIL: UNIQUE constraint on table "idxpart2" does not include column "b" which is part of the partition key.
+drop table idxpart;
+-- but this works:
+create table idxpart (a int, b int, primary key (a, b)) partition by range (a);
+create table idxpart2 partition of idxpart for values from (0) to (1000) partition by range (b);
+create table idxpart21 partition of idxpart2 for values from (0) to (1000);
+select conname, contype, conrelid::regclass, conindid::regclass, conkey
+ from pg_constraint where conrelid::regclass::text like 'idxpart%'
+ order by conname;
+ conname | contype | conrelid | conindid | conkey
+----------------+---------+-----------+----------------+--------
+ idxpart21_pkey | p | idxpart21 | idxpart21_pkey | {1,2}
+ idxpart2_pkey | p | idxpart2 | idxpart2_pkey | {1,2}
+ idxpart_pkey | p | idxpart | idxpart_pkey | {1,2}
+(3 rows)
+
+drop table idxpart;
-- intentionally leave some objects around
create table idxpart (a int) partition by range (a);
create table idxpart1 partition of idxpart for values from (0) to (100);
diff --git a/src/test/regress/expected/insert_conflict.out b/src/test/regress/expected/insert_conflict.out
index 8fd2027d6a..2650faedee 100644
--- a/src/test/regress/expected/insert_conflict.out
+++ b/src/test/regress/expected/insert_conflict.out
@@ -794,7 +794,7 @@ insert into parted_conflict_test values (1, 'a') on conflict do nothing;
insert into parted_conflict_test values (1, 'a') on conflict do nothing;
-- however, on conflict do update is not supported yet
insert into parted_conflict_test values (1) on conflict (b) do update set a = excluded.a;
-ERROR: there is no unique or exclusion constraint matching the ON CONFLICT specification
+ERROR: ON CONFLICT DO UPDATE cannot be applied to partitioned table "parted_conflict_test"
-- but it works OK if we target the partition directly
insert into parted_conflict_test_1 values (1) on conflict (b) do
update set a = excluded.a;
diff --git a/src/test/regress/sql/alter_table.sql b/src/test/regress/sql/alter_table.sql
index af25ee9e77..ed0bb7845b 100644
--- a/src/test/regress/sql/alter_table.sql
+++ b/src/test/regress/sql/alter_table.sql
@@ -2016,8 +2016,6 @@ CREATE TABLE partitioned (
a int,
b int
) PARTITION BY RANGE (a, (a+b+1));
-ALTER TABLE partitioned ADD UNIQUE (a);
-ALTER TABLE partitioned ADD PRIMARY KEY (a);
ALTER TABLE partitioned ADD FOREIGN KEY (a) REFERENCES blah;
ALTER TABLE partitioned ADD EXCLUDE USING gist (a WITH &&);
diff --git a/src/test/regress/sql/create_table.sql b/src/test/regress/sql/create_table.sql
index 8f9991ef18..fefccf21a2 100644
--- a/src/test/regress/sql/create_table.sql
+++ b/src/test/regress/sql/create_table.sql
@@ -294,10 +294,6 @@ CREATE TABLE partitioned (
) PARTITION BY LIST (a1, a2); -- fail
-- unsupported constraint type for partitioned tables
-CREATE TABLE partitioned (
- a int PRIMARY KEY
-) PARTITION BY RANGE (a);
-
CREATE TABLE pkrel (
a int PRIMARY KEY
);
@@ -307,10 +303,6 @@ CREATE TABLE partitioned (
DROP TABLE pkrel;
CREATE TABLE partitioned (
- a int UNIQUE
-) PARTITION BY RANGE (a);
-
-CREATE TABLE partitioned (
a int,
EXCLUDE USING gist (a WITH &&)
) PARTITION BY RANGE (a);
diff --git a/src/test/regress/sql/indexing.sql b/src/test/regress/sql/indexing.sql
index 7cee9599b6..69b16a1174 100644
--- a/src/test/regress/sql/indexing.sql
+++ b/src/test/regress/sql/indexing.sql
@@ -14,7 +14,6 @@ drop table idxpart;
-- Some unsupported features
create table idxpart (a int, b int, c text) partition by range (a);
create table idxpart1 partition of idxpart for values from (0) to (10);
-create unique index on idxpart (a);
create index concurrently on idxpart (a);
drop table idxpart;
@@ -225,6 +224,78 @@ select attrelid::regclass, attname, attnum from pg_attribute
order by attrelid::regclass, attnum;
drop table idxpart;
+--
+-- Constraint-related indexes
+--
+
+-- Verify that it works to add primary key / unique to partitioned tables
+create table idxpart (a int primary key, b int) partition by range (a);
+\d idxpart
+drop table idxpart;
+
+-- but not if you fail to use the full partition key
+create table idxpart (a int unique, b int) partition by range (a, b);
+create table idxpart (a int, b int unique) partition by range (a, b);
+create table idxpart (a int primary key, b int) partition by range (b, a);
+create table idxpart (a int, b int primary key) partition by range (b, a);
+
+-- OK if you use them in some other order
+create table idxpart (a int, b int, c text, primary key (a, b, c)) partition by range (b, c, a);
+drop table idxpart;
+
+create table idxpart (a int primary key, b int) partition by range ((b + a));
+-- not other types of index-based constraints
+create table idxpart (a int, exclude (a with = )) partition by range (a);
+
+-- It works to add primary keys after the partitioned table is created
+create table idxpart (a int, b int, c text) partition by range (a, b);
+alter table idxpart add primary key (a); -- not an incomplete one tho
+alter table idxpart add primary key (a, b);
+\d idxpart
+create table idxpart1 partition of idxpart for values from (0, 0) to (1000, 1000);
+\d idxpart1
+drop table idxpart;
+
+-- It works to add unique constraints after the partitioned table is created
+create table idxpart (a int, b int) partition by range (a, b);
+alter table idxpart add unique (a); -- ... nope
+alter table idxpart add unique (b, a);
+\d idxpart
+drop table idxpart;
+
+-- Exclusion constraints cannot be added
+create table idxpart (a int, b int) partition by range (a);
+alter table idxpart add exclude (a with =);
+drop table idxpart;
+
+-- When (sub)partitions are created, they also contain the constraint
+create table idxpart (a int, b int, primary key (a, b)) partition by range (a, b);
+create table idxpart1 partition of idxpart for values from (1, 1) to (10, 10);
+create table idxpart2 partition of idxpart for values from (10, 10) to (20, 20)
+ partition by range (b);
+create table idxpart21 partition of idxpart2 for values from (10) to (15);
+create table idxpart22 partition of idxpart2 for values from (15) to (20);
+create table idxpart3 (b int not null, a int not null);
+alter table idxpart attach partition idxpart3 for values from (20, 20) to (30, 30);
+select conname, contype, conrelid::regclass, conindid::regclass, conkey
+ from pg_constraint where conrelid::regclass::text like 'idxpart%'
+ order by conname;
+drop table idxpart;
+
+-- multi-layer partitioning honors the prohibition. So this fails:
+create table idxpart (a int, b int, primary key (a)) partition by range (a);
+create table idxpart2 partition of idxpart for values from (0) to (1000) partition by range (b);
+drop table idxpart;
+
+-- but this works:
+create table idxpart (a int, b int, primary key (a, b)) partition by range (a);
+create table idxpart2 partition of idxpart for values from (0) to (1000) partition by range (b);
+create table idxpart21 partition of idxpart2 for values from (0) to (1000);
+select conname, contype, conrelid::regclass, conindid::regclass, conkey
+ from pg_constraint where conrelid::regclass::text like 'idxpart%'
+ order by conname;
+drop table idxpart;
+
-- intentionally leave some objects around
create table idxpart (a int) partition by range (a);
create table idxpart1 partition of idxpart for values from (0) to (100);
--
2.11.0
Re: unique indexes on partitioned tables
On 29 December 2017 at 23:06, Alvaro Herrera <alvherre@2ndquadrant.com> wrote:
This is the patch series for UNIQUE / PRIMARY KEY indexes on partitioned
tables. This is on top of the patch in
/messages/by-id/20171229175930.3aew7lzwd5w6m2x6@alvherre.pgsql
but I included it here as 0001 for simplicity. (Don't review that patch
in this thread please). This is essentially the same patch I posted
elsewhere in that thread.
This is a very simple patch, so not much to object to. Most of the
code is about passing the details thru APIs.
Looks good to go.
The comments are slightly better than the explanation in the docs.
I included Amit's support for ON CONFLICT DO UPDATE, but as I mentioned
in the other thread, it has a small bug. In principle we could push
0002 together with 0003, but I'd rather fix 0004 first and push it all
as one commit.
I agree we want 0004 also, but it would be simpler to just push 0002
and 0003 now and come back later for 0004. That would also avoid any
confusion over patch credits.
This serves as basis to build foreign keys on top; I'll post that
separately.
--
Simon Riggs http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
Re: unique indexes on partitioned tables
Alvaro Herrera wrote:
This is the patch series for UNIQUE / PRIMARY KEY indexes on partitioned
tables.
Rebased.
--
�lvaro Herrera https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
Attachments:
v3-0001-Local-partitioned-indexes.patchtext/plain; charset=us-asciiDownload
From 7f5ca5249a26260a2e9a142c6d4c1b6894b92b1a Mon Sep 17 00:00:00 2001
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
Date: Mon, 23 Oct 2017 10:18:38 +0200
Subject: [PATCH v3 1/2] Local partitioned indexes
When CREATE INDEX is run on a partitioned table, create catalog entries
for an index on the partitioned table (which is just a placeholder since
the table proper has no data of its own), and recurse to create actual
indexes on the existing partitions; create them in future partitions
also.
As a convenience gadget, if the new index definition matches some
existing index in partitions, these are picked up and used instead of
creating new ones. Whichever way these indexes come about, they become
attached to the index on the parent table and are dropped alongside it,
and cannot be dropped on isolation unless they are detached first.
To support pg_dump'ing these indexes, add commands
CREATE INDEX ON ONLY <table>
(which creates the index on the parent partitioned table, without
recursing) and
ALTER INDEX ATTACH PARTITION
(which is used after the indexes have been created individually on each
partition, to attach them to the parent index). These reconstruct prior
database state exactly.
Reviewed-by: Robert Haas, Amit Langote, Jesper Pedersen, Simon Riggs, David Rowley
Discussion: https://postgr.es/m/20171113170646.gzweigyrgg6pwsg4@alvherre.pgsql
---
doc/src/sgml/catalogs.sgml | 23 ++
doc/src/sgml/ref/alter_index.sgml | 14 +
doc/src/sgml/ref/alter_table.sgml | 8 +-
doc/src/sgml/ref/create_index.sgml | 33 +-
doc/src/sgml/ref/reindex.sgml | 5 +
src/backend/access/common/reloptions.c | 1 +
src/backend/access/heap/heapam.c | 9 +-
src/backend/access/index/indexam.c | 3 +-
src/backend/bootstrap/bootparse.y | 2 +
src/backend/catalog/aclchk.c | 9 +-
src/backend/catalog/dependency.c | 14 +-
src/backend/catalog/heap.c | 1 +
src/backend/catalog/index.c | 170 +++++++-
src/backend/catalog/objectaddress.c | 5 +-
src/backend/catalog/pg_depend.c | 13 +-
src/backend/catalog/pg_inherits.c | 80 ++++
src/backend/catalog/toasting.c | 2 +
src/backend/commands/indexcmds.c | 381 +++++++++++++++++-
src/backend/commands/tablecmds.c | 640 +++++++++++++++++++++++++++---
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/nodes/outfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 10 +-
src/backend/parser/gram.y | 33 +-
src/backend/parser/parse_utilcmd.c | 65 ++-
src/backend/tcop/utility.c | 8 +
src/backend/utils/adt/amutils.c | 3 +-
src/backend/utils/adt/ruleutils.c | 17 +-
src/backend/utils/cache/relcache.c | 39 +-
src/bin/pg_dump/common.c | 102 +++++
src/bin/pg_dump/pg_dump.c | 90 ++++-
src/bin/pg_dump/pg_dump.h | 11 +
src/bin/pg_dump/pg_dump_sort.c | 56 ++-
src/bin/psql/describe.c | 20 +-
src/bin/psql/tab-complete.c | 34 +-
src/include/catalog/dependency.h | 15 +
src/include/catalog/index.h | 8 +
src/include/catalog/pg_class.h | 1 +
src/include/catalog/pg_inherits_fn.h | 3 +
src/include/commands/defrem.h | 3 +-
src/include/nodes/execnodes.h | 1 +
src/include/nodes/parsenodes.h | 7 +-
src/include/parser/parse_utilcmd.h | 3 +
src/test/regress/expected/alter_table.out | 65 ++-
src/test/regress/expected/indexing.out | 604 ++++++++++++++++++++++++++++
src/test/regress/parallel_schedule | 2 +-
src/test/regress/serial_schedule | 1 +
src/test/regress/sql/alter_table.sql | 16 +
src/test/regress/sql/indexing.sql | 283 +++++++++++++
49 files changed, 2735 insertions(+), 181 deletions(-)
create mode 100644 src/test/regress/expected/indexing.out
create mode 100644 src/test/regress/sql/indexing.sql
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3f02202caf..71e20f2740 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -2996,6 +2996,29 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
</varlistentry>
<varlistentry>
+ <term><symbol>DEPENDENCY_INTERNAL_AUTO</symbol> (<literal>I</literal>)</term>
+ <listitem>
+ <para>
+ The dependent object was created as part of creation of the
+ referenced object, and is really just a part of its internal
+ implementation. A <command>DROP</command> of the dependent object
+ will be disallowed outright (we'll tell the user to issue a
+ <command>DROP</command> against the referenced object, instead).
+ While a regular internal dependency will prevent
+ the dependent object from being dropped while any such dependencies
+ remain, <literal>DEPENDENCY_INTERNAL_AUTO</literal> will allow such
+ a drop as long as the object can be found by following any of such
+ dependencies.
+ Example: an index on a partition is made internal-auto-dependent on
+ both the partition itself as well as on the index on the parent
+ partitioned table; so the partition index is dropped together with
+ either the partition it indexes, or with the parent index it is
+ attached to.
+ </para>
+ </listitem>
+ </varlistentry>
+
+ <varlistentry>
<term><symbol>DEPENDENCY_EXTENSION</symbol> (<literal>e</literal>)</term>
<listitem>
<para>
diff --git a/doc/src/sgml/ref/alter_index.sgml b/doc/src/sgml/ref/alter_index.sgml
index e54237272c..c0606689f0 100644
--- a/doc/src/sgml/ref/alter_index.sgml
+++ b/doc/src/sgml/ref/alter_index.sgml
@@ -23,6 +23,7 @@ PostgreSQL documentation
<synopsis>
ALTER INDEX [ IF EXISTS ] <replaceable class="parameter">name</replaceable> RENAME TO <replaceable class="parameter">new_name</replaceable>
ALTER INDEX [ IF EXISTS ] <replaceable class="parameter">name</replaceable> SET TABLESPACE <replaceable class="parameter">tablespace_name</replaceable>
+ALTER INDEX <replaceable class="parameter">name</replaceable> ATTACH PARTITION <replaceable class="parameter">index_name</replaceable>
ALTER INDEX <replaceable class="parameter">name</replaceable> DEPENDS ON EXTENSION <replaceable class="parameter">extension_name</replaceable>
ALTER INDEX [ IF EXISTS ] <replaceable class="parameter">name</replaceable> SET ( <replaceable class="parameter">storage_parameter</replaceable> = <replaceable class="parameter">value</replaceable> [, ... ] )
ALTER INDEX [ IF EXISTS ] <replaceable class="parameter">name</replaceable> RESET ( <replaceable class="parameter">storage_parameter</replaceable> [, ... ] )
@@ -76,6 +77,19 @@ ALTER INDEX ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
</varlistentry>
<varlistentry>
+ <term><literal>ATTACH PARTITION</literal></term>
+ <listitem>
+ <para>
+ Causes the named index to become attached to the altered index.
+ The named index must be on a partition of the table containing the
+ index being altered, and have an equivalent definition. An attached
+ index cannot be dropped by itself, and will automatically be dropped
+ if its parent index is dropped.
+ </para>
+ </listitem>
+ </varlistentry>
+
+ <varlistentry>
<term><literal>DEPENDS ON EXTENSION</literal></term>
<listitem>
<para>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index 7bcf242846..2c41f2107e 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -783,7 +783,10 @@ ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
as a partition of the target table. The table can be attached
as a partition for specific values using <literal>FOR VALUES
</literal> or as a default partition by using <literal>DEFAULT
- </literal>.
+ </literal>. For each index in the target table, a corresponding
+ one will be created in the attached table; or, if an equivalent
+ index already exists, will be attached to the target table's index,
+ as if <command>ALTER INDEX ATTACH PARTITION</command> had been executed.
</para>
<para>
@@ -844,7 +847,8 @@ ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
<para>
This form detaches specified partition of the target table. The detached
partition continues to exist as a standalone table, but no longer has any
- ties to the table from which it was detached.
+ ties to the table from which it was detached. Any indexes that were
+ attached to the target table's indexes are detached.
</para>
</listitem>
</varlistentry>
diff --git a/doc/src/sgml/ref/create_index.sgml b/doc/src/sgml/ref/create_index.sgml
index 025537575b..5137fe6383 100644
--- a/doc/src/sgml/ref/create_index.sgml
+++ b/doc/src/sgml/ref/create_index.sgml
@@ -21,7 +21,7 @@ PostgreSQL documentation
<refsynopsisdiv>
<synopsis>
-CREATE [ UNIQUE ] INDEX [ CONCURRENTLY ] [ [ IF NOT EXISTS ] <replaceable class="parameter">name</replaceable> ] ON <replaceable class="parameter">table_name</replaceable> [ USING <replaceable class="parameter">method</replaceable> ]
+CREATE [ UNIQUE ] INDEX [ CONCURRENTLY ] [ [ IF NOT EXISTS ] <replaceable class="parameter">name</replaceable> ] ON [ ONLY ] <replaceable class="parameter">table_name</replaceable> [ USING <replaceable class="parameter">method</replaceable> ]
( { <replaceable class="parameter">column_name</replaceable> | ( <replaceable class="parameter">expression</replaceable> ) } [ COLLATE <replaceable class="parameter">collation</replaceable> ] [ <replaceable class="parameter">opclass</replaceable> ] [ ASC | DESC ] [ NULLS { FIRST | LAST } ] [, ...] )
[ WITH ( <replaceable class="parameter">storage_parameter</replaceable> = <replaceable class="parameter">value</replaceable> [, ... ] ) ]
[ TABLESPACE <replaceable class="parameter">tablespace_name</replaceable> ]
@@ -152,6 +152,16 @@ CREATE [ UNIQUE ] INDEX [ CONCURRENTLY ] [ [ IF NOT EXISTS ] <replaceable class=
</varlistentry>
<varlistentry>
+ <term><literal>ONLY</literal></term>
+ <listitem>
+ <para>
+ Indicates not to recurse creating indexes on partitions, if the
+ table is partitioned. The default is to recurse.
+ </para>
+ </listitem>
+ </varlistentry>
+
+ <varlistentry>
<term><replaceable class="parameter">table_name</replaceable></term>
<listitem>
<para>
@@ -546,6 +556,27 @@ Indexes:
</para>
<para>
+ When <literal>CREATE INDEX</literal> is invoked on a partitioned
+ table, the default behavior is to recurse to all partitions to ensure
+ they all have matching indexes.
+ Each partition is first checked to determine whether an equivalent
+ index already exists, and if so, that index will become attached as a
+ partition index to the index being created, which will become its
+ parent index.
+ If no matching index exists, a new index will be created and
+ automatically attached; the name of the new index in each partition
+ will be determined as if no index name had been specified in the
+ command.
+ If the <literal>ONLY</literal> option is specified, no recursion
+ is done, and the index is marked invalid
+ (<command>ALTER INDEX ... ATTACH PARTITION</command> turns the index
+ valid, once all partitions acquire the index.) Note, however, that
+ any partition that is created in the future using
+ <command>CREATE TABLE ... PARTITION OF</command> will automatically
+ contain the index regardless of whether this option was specified.
+ </para>
+
+ <para>
For index methods that support ordered scans (currently, only B-tree),
the optional clauses <literal>ASC</literal>, <literal>DESC</literal>, <literal>NULLS
FIRST</literal>, and/or <literal>NULLS LAST</literal> can be specified to modify
diff --git a/doc/src/sgml/ref/reindex.sgml b/doc/src/sgml/ref/reindex.sgml
index 79f6931c6a..1c21fafb80 100644
--- a/doc/src/sgml/ref/reindex.sgml
+++ b/doc/src/sgml/ref/reindex.sgml
@@ -231,6 +231,11 @@ REINDEX [ ( VERBOSE ) ] { INDEX | TABLE | SCHEMA | DATABASE | SYSTEM } <replacea
reindex anything.
</para>
+ <para>
+ Reindexing partitioned tables or partitioned indexes is not supported.
+ Each individual partition can be reindexed separately instead.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/access/common/reloptions.c b/src/backend/access/common/reloptions.c
index 425bc5d06e..274f7aa8e9 100644
--- a/src/backend/access/common/reloptions.c
+++ b/src/backend/access/common/reloptions.c
@@ -993,6 +993,7 @@ extractRelOptions(HeapTuple tuple, TupleDesc tupdesc,
options = view_reloptions(datum, false);
break;
case RELKIND_INDEX:
+ case RELKIND_PARTITIONED_INDEX:
options = index_reloptions(amoptions, datum, false);
break;
case RELKIND_FOREIGN_TABLE:
diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c
index dbc8f2d6c7..be263850cd 100644
--- a/src/backend/access/heap/heapam.c
+++ b/src/backend/access/heap/heapam.c
@@ -1293,7 +1293,8 @@ heap_open(Oid relationId, LOCKMODE lockmode)
r = relation_open(relationId, lockmode);
- if (r->rd_rel->relkind == RELKIND_INDEX)
+ if (r->rd_rel->relkind == RELKIND_INDEX ||
+ r->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is an index",
@@ -1321,7 +1322,8 @@ heap_openrv(const RangeVar *relation, LOCKMODE lockmode)
r = relation_openrv(relation, lockmode);
- if (r->rd_rel->relkind == RELKIND_INDEX)
+ if (r->rd_rel->relkind == RELKIND_INDEX ||
+ r->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is an index",
@@ -1353,7 +1355,8 @@ heap_openrv_extended(const RangeVar *relation, LOCKMODE lockmode,
if (r)
{
- if (r->rd_rel->relkind == RELKIND_INDEX)
+ if (r->rd_rel->relkind == RELKIND_INDEX ||
+ r->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is an index",
diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c
index 1b61cd9515..91247f0fa5 100644
--- a/src/backend/access/index/indexam.c
+++ b/src/backend/access/index/indexam.c
@@ -154,7 +154,8 @@ index_open(Oid relationId, LOCKMODE lockmode)
r = relation_open(relationId, lockmode);
- if (r->rd_rel->relkind != RELKIND_INDEX)
+ if (r->rd_rel->relkind != RELKIND_INDEX &&
+ r->rd_rel->relkind != RELKIND_PARTITIONED_INDEX)
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is not an index",
diff --git a/src/backend/bootstrap/bootparse.y b/src/backend/bootstrap/bootparse.y
index 8c52846a92..dfd53fa054 100644
--- a/src/backend/bootstrap/bootparse.y
+++ b/src/backend/bootstrap/bootparse.y
@@ -321,6 +321,7 @@ Boot_DeclareIndexStmt:
DefineIndex(relationId,
stmt,
$4,
+ InvalidOid,
false,
false,
false,
@@ -365,6 +366,7 @@ Boot_DeclareUniqueIndexStmt:
DefineIndex(relationId,
stmt,
$5,
+ InvalidOid,
false,
false,
false,
diff --git a/src/backend/catalog/aclchk.c b/src/backend/catalog/aclchk.c
index fac80612b8..50a2e2681b 100644
--- a/src/backend/catalog/aclchk.c
+++ b/src/backend/catalog/aclchk.c
@@ -1824,7 +1824,8 @@ ExecGrant_Relation(InternalGrant *istmt)
pg_class_tuple = (Form_pg_class) GETSTRUCT(tuple);
/* Not sensible to grant on an index */
- if (pg_class_tuple->relkind == RELKIND_INDEX)
+ if (pg_class_tuple->relkind == RELKIND_INDEX ||
+ pg_class_tuple->relkind == RELKIND_PARTITIONED_INDEX)
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is an index",
@@ -5405,7 +5406,8 @@ recordExtObjInitPriv(Oid objoid, Oid classoid)
pg_class_tuple = (Form_pg_class) GETSTRUCT(tuple);
/* Indexes don't have permissions */
- if (pg_class_tuple->relkind == RELKIND_INDEX)
+ if (pg_class_tuple->relkind == RELKIND_INDEX ||
+ pg_class_tuple->relkind == RELKIND_PARTITIONED_INDEX)
return;
/* Composite types don't have permissions either */
@@ -5690,7 +5692,8 @@ removeExtObjInitPriv(Oid objoid, Oid classoid)
pg_class_tuple = (Form_pg_class) GETSTRUCT(tuple);
/* Indexes don't have permissions */
- if (pg_class_tuple->relkind == RELKIND_INDEX)
+ if (pg_class_tuple->relkind == RELKIND_INDEX ||
+ pg_class_tuple->relkind == RELKIND_PARTITIONED_INDEX)
return;
/* Composite types don't have permissions either */
diff --git a/src/backend/catalog/dependency.c b/src/backend/catalog/dependency.c
index 269111b4c1..be60270ea5 100644
--- a/src/backend/catalog/dependency.c
+++ b/src/backend/catalog/dependency.c
@@ -582,6 +582,7 @@ findDependentObjects(const ObjectAddress *object,
/* FALL THRU */
case DEPENDENCY_INTERNAL:
+ case DEPENDENCY_INTERNAL_AUTO:
/*
* This object is part of the internal implementation of
@@ -633,6 +634,14 @@ findDependentObjects(const ObjectAddress *object,
* transform this deletion request into a delete of this
* owning object.
*
+ * For INTERNAL_AUTO dependencies, we don't enforce this;
+ * in other words, we don't follow the links back to the
+ * owning object.
+ */
+ if (foundDep->deptype == DEPENDENCY_INTERNAL_AUTO)
+ break;
+
+ /*
* First, release caller's lock on this object and get
* deletion lock on the owning object. (We must release
* caller's lock to avoid deadlock against a concurrent
@@ -675,6 +684,7 @@ findDependentObjects(const ObjectAddress *object,
/* And we're done here. */
systable_endscan(scan);
return;
+
case DEPENDENCY_PIN:
/*
@@ -762,6 +772,7 @@ findDependentObjects(const ObjectAddress *object,
case DEPENDENCY_AUTO_EXTENSION:
subflags = DEPFLAG_AUTO;
break;
+ case DEPENDENCY_INTERNAL_AUTO:
case DEPENDENCY_INTERNAL:
subflags = DEPFLAG_INTERNAL;
break;
@@ -1109,7 +1120,8 @@ doDeletion(const ObjectAddress *object, int flags)
{
char relKind = get_rel_relkind(object->objectId);
- if (relKind == RELKIND_INDEX)
+ if (relKind == RELKIND_INDEX ||
+ relKind == RELKIND_PARTITIONED_INDEX)
{
bool concurrent = ((flags & PERFORM_DELETION_CONCURRENTLY) != 0);
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 089b7965f2..99f4d59863 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -294,6 +294,7 @@ heap_create(const char *relname,
case RELKIND_COMPOSITE_TYPE:
case RELKIND_FOREIGN_TABLE:
case RELKIND_PARTITIONED_TABLE:
+ case RELKIND_PARTITIONED_INDEX:
create_storage = false;
/*
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 330488b96f..6a7fa91e25 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -41,6 +41,8 @@
#include "catalog/pg_collation.h"
#include "catalog/pg_constraint.h"
#include "catalog/pg_constraint_fn.h"
+#include "catalog/pg_depend.h"
+#include "catalog/pg_inherits_fn.h"
#include "catalog/pg_operator.h"
#include "catalog/pg_opclass.h"
#include "catalog/pg_tablespace.h"
@@ -55,6 +57,7 @@
#include "nodes/nodeFuncs.h"
#include "optimizer/clauses.h"
#include "parser/parser.h"
+#include "rewrite/rewriteManip.h"
#include "storage/bufmgr.h"
#include "storage/lmgr.h"
#include "storage/predicate.h"
@@ -98,6 +101,7 @@ static void InitializeAttributeOids(Relation indexRelation,
int numatts, Oid indexoid);
static void AppendAttributeTuples(Relation indexRelation, int numatts);
static void UpdateIndexRelation(Oid indexoid, Oid heapoid,
+ Oid parentIndexId,
IndexInfo *indexInfo,
Oid *collationOids,
Oid *classOids,
@@ -105,7 +109,8 @@ static void UpdateIndexRelation(Oid indexoid, Oid heapoid,
bool primary,
bool isexclusion,
bool immediate,
- bool isvalid);
+ bool isvalid,
+ bool isready);
static void index_update_stats(Relation rel,
bool hasindex, bool isprimary,
double reltuples);
@@ -551,6 +556,7 @@ AppendAttributeTuples(Relation indexRelation, int numatts)
static void
UpdateIndexRelation(Oid indexoid,
Oid heapoid,
+ Oid parentIndexOid,
IndexInfo *indexInfo,
Oid *collationOids,
Oid *classOids,
@@ -558,7 +564,8 @@ UpdateIndexRelation(Oid indexoid,
bool primary,
bool isexclusion,
bool immediate,
- bool isvalid)
+ bool isvalid,
+ bool isready)
{
int2vector *indkey;
oidvector *indcollation;
@@ -632,8 +639,7 @@ UpdateIndexRelation(Oid indexoid,
values[Anum_pg_index_indisclustered - 1] = BoolGetDatum(false);
values[Anum_pg_index_indisvalid - 1] = BoolGetDatum(isvalid);
values[Anum_pg_index_indcheckxmin - 1] = BoolGetDatum(false);
- /* we set isvalid and isready the same way */
- values[Anum_pg_index_indisready - 1] = BoolGetDatum(isvalid);
+ values[Anum_pg_index_indisready - 1] = BoolGetDatum(isready);
values[Anum_pg_index_indislive - 1] = BoolGetDatum(true);
values[Anum_pg_index_indisreplident - 1] = BoolGetDatum(false);
values[Anum_pg_index_indkey - 1] = PointerGetDatum(indkey);
@@ -670,6 +676,8 @@ UpdateIndexRelation(Oid indexoid,
* indexRelationId: normally, pass InvalidOid to let this routine
* generate an OID for the index. During bootstrap this may be
* nonzero to specify a preselected OID.
+ * parentIndexRelid: if creating an index partition, the OID of the
+ * parent index; otherwise InvalidOid.
* relFileNode: normally, pass InvalidOid to get new storage. May be
* nonzero to attach an existing valid build.
* indexInfo: same info executor uses to insert into the index
@@ -695,6 +703,8 @@ UpdateIndexRelation(Oid indexoid,
* INDEX_CREATE_IF_NOT_EXISTS:
* do not throw an error if a relation with the same name
* already exists.
+ * INDEX_CREATE_PARTITIONED:
+ * create a partitioned index (table must be partitioned)
* constr_flags: flags passed to index_constraint_create
* (only if INDEX_CREATE_ADD_CONSTRAINT is set)
* allow_system_table_mods: allow table to be a system catalog
@@ -706,6 +716,7 @@ Oid
index_create(Relation heapRelation,
const char *indexRelationName,
Oid indexRelationId,
+ Oid parentIndexRelid,
Oid relFileNode,
IndexInfo *indexInfo,
List *indexColNames,
@@ -731,12 +742,18 @@ index_create(Relation heapRelation,
int i;
char relpersistence;
bool isprimary = (flags & INDEX_CREATE_IS_PRIMARY) != 0;
+ bool invalid = (flags & INDEX_CREATE_INVALID) != 0;
bool concurrent = (flags & INDEX_CREATE_CONCURRENT) != 0;
+ bool partitioned = (flags & INDEX_CREATE_PARTITIONED) != 0;
+ char relkind;
/* constraint flags can only be set when a constraint is requested */
Assert((constr_flags == 0) ||
((flags & INDEX_CREATE_ADD_CONSTRAINT) != 0));
+ /* partitioned indexes must never be "built" by themselves */
+ Assert(!partitioned || (flags & INDEX_CREATE_SKIP_BUILD));
+ relkind = partitioned ? RELKIND_PARTITIONED_INDEX : RELKIND_INDEX;
is_exclusion = (indexInfo->ii_ExclusionOps != NULL);
pg_class = heap_open(RelationRelationId, RowExclusiveLock);
@@ -854,9 +871,9 @@ index_create(Relation heapRelation,
}
/*
- * create the index relation's relcache entry and physical disk file. (If
- * we fail further down, it's the smgr's responsibility to remove the disk
- * file again.)
+ * create the index relation's relcache entry and, if necessary, the
+ * physical disk file. (If we fail further down, it's the smgr's
+ * responsibility to remove the disk file again, if any.)
*/
indexRelation = heap_create(indexRelationName,
namespaceId,
@@ -864,7 +881,7 @@ index_create(Relation heapRelation,
indexRelationId,
relFileNode,
indexTupDesc,
- RELKIND_INDEX,
+ relkind,
relpersistence,
shared_relation,
mapped_relation,
@@ -921,12 +938,18 @@ index_create(Relation heapRelation,
* (Or, could define a rule to maintain the predicate) --Nels, Feb '92
* ----------------
*/
- UpdateIndexRelation(indexRelationId, heapRelationId, indexInfo,
+ UpdateIndexRelation(indexRelationId, heapRelationId, parentIndexRelid,
+ indexInfo,
collationObjectId, classObjectId, coloptions,
isprimary, is_exclusion,
(constr_flags & INDEX_CONSTR_CREATE_DEFERRABLE) == 0,
+ !concurrent && !invalid,
!concurrent);
+ /* update pg_inherits, if needed */
+ if (OidIsValid(parentIndexRelid))
+ StoreSingleInheritance(indexRelationId, parentIndexRelid, 1);
+
/*
* Register constraint and dependencies for the index.
*
@@ -978,6 +1001,9 @@ index_create(Relation heapRelation,
else
{
bool have_simple_col = false;
+ DependencyType deptype;
+
+ deptype = OidIsValid(parentIndexRelid) ? DEPENDENCY_INTERNAL_AUTO : DEPENDENCY_AUTO;
/* Create auto dependencies on simply-referenced columns */
for (i = 0; i < indexInfo->ii_NumIndexAttrs; i++)
@@ -988,7 +1014,7 @@ index_create(Relation heapRelation,
referenced.objectId = heapRelationId;
referenced.objectSubId = indexInfo->ii_KeyAttrNumbers[i];
- recordDependencyOn(&myself, &referenced, DEPENDENCY_AUTO);
+ recordDependencyOn(&myself, &referenced, deptype);
have_simple_col = true;
}
@@ -1006,10 +1032,20 @@ index_create(Relation heapRelation,
referenced.objectId = heapRelationId;
referenced.objectSubId = 0;
- recordDependencyOn(&myself, &referenced, DEPENDENCY_AUTO);
+ recordDependencyOn(&myself, &referenced, deptype);
}
}
+ /* Store dependency on parent index, if any */
+ if (OidIsValid(parentIndexRelid))
+ {
+ referenced.classId = RelationRelationId;
+ referenced.objectId = parentIndexRelid;
+ referenced.objectSubId = 0;
+
+ recordDependencyOn(&myself, &referenced, DEPENDENCY_INTERNAL_AUTO);
+ }
+
/* Store dependency on collations */
/* The default collation is pinned, so don't bother recording it */
for (i = 0; i < indexInfo->ii_NumIndexAttrs; i++)
@@ -1555,9 +1591,10 @@ index_drop(Oid indexId, bool concurrent)
}
/*
- * Schedule physical removal of the files
+ * Schedule physical removal of the files (if any)
*/
- RelationDropStorage(userIndexRelation);
+ if (userIndexRelation->rd_rel->relkind != RELKIND_PARTITIONED_INDEX)
+ RelationDropStorage(userIndexRelation);
/*
* Close and flush the index's relcache entry, to ensure relcache doesn't
@@ -1602,6 +1639,11 @@ index_drop(Oid indexId, bool concurrent)
DeleteRelationTuple(indexId);
/*
+ * fix INHERITS relation
+ */
+ DeleteInheritsTuple(indexId, InvalidOid);
+
+ /*
* We are presently too lazy to attempt to compute the new correct value
* of relhasindex (the next VACUUM will fix it if necessary). So there is
* no need to update the pg_class tuple for the owning relation. But we
@@ -1694,12 +1736,87 @@ BuildIndexInfo(Relation index)
ii->ii_BrokenHotChain = false;
/* set up for possible use by index AM */
+ ii->ii_Am = index->rd_rel->relam;
ii->ii_AmCache = NULL;
ii->ii_Context = CurrentMemoryContext;
return ii;
}
+/*
+ * CompareIndexInfo
+ * Compare two IndexInfos, and return true if they are similar enough that
+ * an index built with one can pass as an index built with the other.
+ *
+ * attmap is an attribute map where info2 is input and info1 is output.
+ */
+bool
+CompareIndexInfo(IndexInfo *info1, IndexInfo *info2, AttrNumber *attmap, int maplen)
+{
+ int i;
+
+ /* indexes are only equivalent if they have the same access method */
+ if (info1->ii_Am != info2->ii_Am)
+ return false;
+
+ /* and same number of attributes */
+ if (info1->ii_NumIndexAttrs != info2->ii_NumIndexAttrs)
+ return false;
+
+ /*
+ * and columns match through the attribute map (actual attribute numbers
+ * might differ!) Note that this implies that index columns that are
+ * expressions appear in the same positions. We will next compare the
+ * expressions themselves.
+ */
+ for (i = 0; i < info1->ii_NumIndexAttrs; i++)
+ {
+ if (attmap[info1->ii_KeyAttrNumbers[i] - 1] !=
+ info2->ii_KeyAttrNumbers[i])
+ return false;
+ }
+
+ /*
+ * For expression indexes: either both are expression indexes, or neither
+ * is; if they are, make sure the expressions match.
+ */
+ if ((info1->ii_Expressions != NIL) != (info2->ii_Expressions != NIL))
+ return false;
+ else
+ {
+ bool found_whole_row;
+ Node *mapped;
+
+ mapped = map_variable_attnos((Node *) info2->ii_Expressions,
+ 1, 0, attmap, maplen,
+ InvalidOid, &found_whole_row);
+ if (found_whole_row)
+ {
+ /*
+ * we could throw an error here, but seems out of scope for this
+ * routine.
+ */
+ return false;
+ }
+
+ if (!equal(info1->ii_Expressions, mapped))
+ return false;
+ }
+
+ /* Index predicates must be identical */
+ if (!equal(info1->ii_Predicate, info2->ii_Predicate))
+ return false;
+
+ /* No support currently for comparing exclusion indexes. */
+ if (info1->ii_ExclusionOps != NULL || info2->ii_ExclusionOps != NULL)
+ return false;
+
+ if (info1->ii_Unique != info2->ii_Unique)
+ return false;
+
+ return true;
+}
+
/* ----------------
* BuildSpeculativeIndexInfo
* Add extra state to IndexInfo record
@@ -1922,6 +2039,9 @@ index_update_stats(Relation rel,
elog(ERROR, "could not find tuple for relation %u", relid);
rd_rel = (Form_pg_class) GETSTRUCT(tuple);
+ /* Should this be a more comprehensive test? */
+ Assert(rd_rel->relkind != RELKIND_PARTITIONED_INDEX);
+
/* Apply required updates, if any, to copied tuple */
dirty = false;
@@ -3332,6 +3452,14 @@ reindex_index(Oid indexId, bool skip_constraint_checks, char persistence,
iRel = index_open(indexId, AccessExclusiveLock);
/*
+ * The case of reindexing partitioned tables and indexes is handled
+ * differently by upper layers, so this case shouldn't arise.
+ */
+ if (iRel->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
+ elog(ERROR, "unsupported relation kind for index \"%s\"",
+ RelationGetRelationName(iRel));
+
+ /*
* Don't allow reindex on temp tables of other backends ... their local
* buffer manager is not going to cope.
*/
@@ -3530,6 +3658,22 @@ reindex_relation(Oid relid, int flags, int options)
*/
rel = heap_open(relid, ShareLock);
+ /*
+ * This may be useful when implemented someday; but that day is not today.
+ * For now, avoid erroring out when called in a multi-table context
+ * (REINDEX SCHEMA) and happen to come across a partitioned table. The
+ * partitions may be reindexed on their own anyway.
+ */
+ if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
+ {
+ ereport(WARNING,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("REINDEX of partitioned tables is not yet implemented, skipping \"%s\"",
+ RelationGetRelationName(rel))));
+ heap_close(rel, ShareLock);
+ return false;
+ }
+
toast_relid = rel->rd_rel->reltoastrelid;
/*
diff --git a/src/backend/catalog/objectaddress.c b/src/backend/catalog/objectaddress.c
index bc999ca3c4..7576606c1b 100644
--- a/src/backend/catalog/objectaddress.c
+++ b/src/backend/catalog/objectaddress.c
@@ -1217,7 +1217,8 @@ get_relation_by_qualified_name(ObjectType objtype, List *object,
switch (objtype)
{
case OBJECT_INDEX:
- if (relation->rd_rel->relkind != RELKIND_INDEX)
+ if (relation->rd_rel->relkind != RELKIND_INDEX &&
+ relation->rd_rel->relkind != RELKIND_PARTITIONED_INDEX)
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is not an index",
@@ -3483,6 +3484,7 @@ getRelationDescription(StringInfo buffer, Oid relid)
relname);
break;
case RELKIND_INDEX:
+ case RELKIND_PARTITIONED_INDEX:
appendStringInfo(buffer, _("index %s"),
relname);
break;
@@ -3957,6 +3959,7 @@ getRelationTypeDescription(StringInfo buffer, Oid relid, int32 objectSubId)
appendStringInfoString(buffer, "table");
break;
case RELKIND_INDEX:
+ case RELKIND_PARTITIONED_INDEX:
appendStringInfoString(buffer, "index");
break;
case RELKIND_SEQUENCE:
diff --git a/src/backend/catalog/pg_depend.c b/src/backend/catalog/pg_depend.c
index 9dfbe123b5..2ea05f350b 100644
--- a/src/backend/catalog/pg_depend.c
+++ b/src/backend/catalog/pg_depend.c
@@ -656,14 +656,19 @@ get_constraint_index(Oid constraintId)
/*
* We assume any internal dependency of an index on the constraint
- * must be what we are looking for. (The relkind test is just
- * paranoia; there shouldn't be any such dependencies otherwise.)
+ * must be what we are looking for.
*/
if (deprec->classid == RelationRelationId &&
deprec->objsubid == 0 &&
- deprec->deptype == DEPENDENCY_INTERNAL &&
- get_rel_relkind(deprec->objid) == RELKIND_INDEX)
+ deprec->deptype == DEPENDENCY_INTERNAL)
{
+ char relkind = get_rel_relkind(deprec->objid);
+
+ /* This is pure paranoia; there shouldn't be any such */
+ if (relkind != RELKIND_INDEX &&
+ relkind != RELKIND_PARTITIONED_INDEX)
+ break;
+
indexId = deprec->objid;
break;
}
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index b32d677347..5a5beb9273 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -405,3 +405,83 @@ typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId)
return result;
}
+
+/*
+ * Create a single pg_inherits row with the given data
+ */
+void
+StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
+{
+ Datum values[Natts_pg_inherits];
+ bool nulls[Natts_pg_inherits];
+ HeapTuple tuple;
+ Relation inhRelation;
+
+ inhRelation = heap_open(InheritsRelationId, RowExclusiveLock);
+
+ /*
+ * Make the pg_inherits entry
+ */
+ values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
+ values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
+ values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+
+ memset(nulls, 0, sizeof(nulls));
+
+ tuple = heap_form_tuple(RelationGetDescr(inhRelation), values, nulls);
+
+ CatalogTupleInsert(inhRelation, tuple);
+
+ heap_freetuple(tuple);
+
+ heap_close(inhRelation, RowExclusiveLock);
+}
+
+/*
+ * DeleteInheritsTuple
+ *
+ * Delete pg_inherits tuples with the given inhrelid. inhparent may be given
+ * as InvalidOid, in which case all tuples matching inhrelid are deleted;
+ * otherwise only delete tuples with the specified inhparent.
+ *
+ * Returns whether at least one row was deleted.
+ */
+bool
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+{
+ bool found = false;
+ Relation catalogRelation;
+ ScanKeyData key;
+ SysScanDesc scan;
+ HeapTuple inheritsTuple;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = heap_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(inhrelid));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ Oid parent;
+
+ /* Compare inhparent if it was given, and do the actual deletion. */
+ parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
+ if (!OidIsValid(inhparent) || parent == inhparent)
+ {
+ CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
+ found = true;
+ }
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ heap_close(catalogRelation, RowExclusiveLock);
+
+ return found;
+}
diff --git a/src/backend/catalog/toasting.c b/src/backend/catalog/toasting.c
index 0b4b5631a1..cf37011b73 100644
--- a/src/backend/catalog/toasting.c
+++ b/src/backend/catalog/toasting.c
@@ -315,6 +315,7 @@ create_toast_table(Relation rel, Oid toastOid, Oid toastIndexOid,
indexInfo->ii_ReadyForInserts = true;
indexInfo->ii_Concurrent = false;
indexInfo->ii_BrokenHotChain = false;
+ indexInfo->ii_Am = BTREE_AM_OID;
indexInfo->ii_AmCache = NULL;
indexInfo->ii_Context = CurrentMemoryContext;
@@ -328,6 +329,7 @@ create_toast_table(Relation rel, Oid toastOid, Oid toastIndexOid,
coloptions[1] = 0;
index_create(toast_rel, toast_idxname, toastIndexOid, InvalidOid,
+ InvalidOid,
indexInfo,
list_make2("chunk_id", "chunk_seq"),
BTREE_AM_OID,
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 9e6ba92008..566a27cc1d 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -23,7 +23,10 @@
#include "catalog/catalog.h"
#include "catalog/index.h"
#include "catalog/indexing.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
+#include "catalog/pg_inherits_fn.h"
#include "catalog/pg_opclass.h"
#include "catalog/pg_opfamily.h"
#include "catalog/pg_tablespace.h"
@@ -35,6 +38,7 @@
#include "commands/tablespace.h"
#include "mb/pg_wchar.h"
#include "miscadmin.h"
+#include "nodes/makefuncs.h"
#include "nodes/nodeFuncs.h"
#include "optimizer/clauses.h"
#include "optimizer/planner.h"
@@ -77,6 +81,7 @@ static char *ChooseIndexNameAddition(List *colnames);
static List *ChooseIndexColumnNames(List *indexElems);
static void RangeVarCallbackForReindexIndex(const RangeVar *relation,
Oid relId, Oid oldRelId, void *arg);
+static void ReindexPartitionedIndex(Relation parentIdx);
/*
* CheckIndexCompatible
@@ -183,6 +188,7 @@ CheckIndexCompatible(Oid oldId,
indexInfo->ii_ExclusionOps = NULL;
indexInfo->ii_ExclusionProcs = NULL;
indexInfo->ii_ExclusionStrats = NULL;
+ indexInfo->ii_Am = accessMethodId;
indexInfo->ii_AmCache = NULL;
indexInfo->ii_Context = CurrentMemoryContext;
typeObjectId = (Oid *) palloc(numberOfAttributes * sizeof(Oid));
@@ -292,14 +298,15 @@ CheckIndexCompatible(Oid oldId,
* 'stmt': IndexStmt describing the properties of the new index.
* 'indexRelationId': normally InvalidOid, but during bootstrap can be
* nonzero to specify a preselected OID for the index.
+ * 'parentIndexId': the OID of the parent index; InvalidOid if not the child
+ * of a partitioned index.
* 'is_alter_table': this is due to an ALTER rather than a CREATE operation.
* 'check_rights': check for CREATE rights in namespace and tablespace. (This
* should be true except when ALTER is deleting/recreating an index.)
* 'check_not_in_use': check for table not already in use in current session.
* This should be true unless caller is holding the table open, in which
* case the caller had better have checked it earlier.
- * 'skip_build': make the catalog entries but leave the index file empty;
- * it will be filled later.
+ * 'skip_build': make the catalog entries but don't create the index files
* 'quiet': suppress the NOTICE chatter ordinarily provided for constraints.
*
* Returns the object address of the created index.
@@ -308,6 +315,7 @@ ObjectAddress
DefineIndex(Oid relationId,
IndexStmt *stmt,
Oid indexRelationId,
+ Oid parentIndexId,
bool is_alter_table,
bool check_rights,
bool check_not_in_use,
@@ -330,6 +338,7 @@ DefineIndex(Oid relationId,
IndexAmRoutine *amRoutine;
bool amcanorder;
amoptions_function amoptions;
+ bool partitioned;
Datum reloptions;
int16 *coloptions;
IndexInfo *indexInfo;
@@ -382,23 +391,56 @@ DefineIndex(Oid relationId,
{
case RELKIND_RELATION:
case RELKIND_MATVIEW:
+ case RELKIND_PARTITIONED_TABLE:
/* OK */
break;
case RELKIND_FOREIGN_TABLE:
+ /*
+ * Custom error message for FOREIGN TABLE since the term is close
+ * to a regular table and can confuse the user.
+ */
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("cannot create index on foreign table \"%s\"",
RelationGetRelationName(rel))));
- case RELKIND_PARTITIONED_TABLE:
- ereport(ERROR,
- (errcode(ERRCODE_WRONG_OBJECT_TYPE),
- errmsg("cannot create index on partitioned table \"%s\"",
- RelationGetRelationName(rel))));
default:
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is not a table or materialized view",
RelationGetRelationName(rel))));
+ break;
+ }
+
+ /*
+ * Establish behavior for partitioned tables, and verify sanity of
+ * parameters.
+ *
+ * We do not build an actual index in this case; we only create a few
+ * catalog entries. The actual indexes are built by recursing for each
+ * partition.
+ */
+ partitioned = rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE;
+ if (partitioned)
+ {
+ if (stmt->concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("cannot create index on partitioned table \"%s\" concurrently",
+ RelationGetRelationName(rel))));
+ if (stmt->unique)
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("cannot create unique index on partitioned table \"%s\"",
+ RelationGetRelationName(rel))));
+ if (stmt->excludeOpNames)
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("cannot create exclusion constraints on partitioned table \"%s\"",
+ RelationGetRelationName(rel))));
+ if (stmt->primary || stmt->isconstraint)
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("cannot create constraints on partitioned tables")));
}
/*
@@ -574,6 +616,7 @@ DefineIndex(Oid relationId,
indexInfo->ii_ReadyForInserts = !stmt->concurrent;
indexInfo->ii_Concurrent = stmt->concurrent;
indexInfo->ii_BrokenHotChain = false;
+ indexInfo->ii_Am = accessMethodId;
indexInfo->ii_AmCache = NULL;
indexInfo->ii_Context = CurrentMemoryContext;
@@ -665,19 +708,24 @@ DefineIndex(Oid relationId,
/*
* Make the catalog entries for the index, including constraints. This
* step also actually builds the index, except if caller requested not to
- * or in concurrent mode, in which case it'll be done later.
+ * or in concurrent mode, in which case it'll be done later, or
+ * doing a partitioned index (because those don't have storage).
*/
flags = constr_flags = 0;
if (stmt->isconstraint)
flags |= INDEX_CREATE_ADD_CONSTRAINT;
- if (skip_build || stmt->concurrent)
+ if (skip_build || stmt->concurrent || partitioned)
flags |= INDEX_CREATE_SKIP_BUILD;
if (stmt->if_not_exists)
flags |= INDEX_CREATE_IF_NOT_EXISTS;
if (stmt->concurrent)
flags |= INDEX_CREATE_CONCURRENT;
+ if (partitioned)
+ flags |= INDEX_CREATE_PARTITIONED;
if (stmt->primary)
flags |= INDEX_CREATE_IS_PRIMARY;
+ if (partitioned && stmt->relation && !stmt->relation->inh)
+ flags |= INDEX_CREATE_INVALID;
if (stmt->deferrable)
constr_flags |= INDEX_CONSTR_CREATE_DEFERRABLE;
@@ -685,8 +733,8 @@ DefineIndex(Oid relationId,
constr_flags |= INDEX_CONSTR_CREATE_INIT_DEFERRED;
indexRelationId =
- index_create(rel, indexRelationName, indexRelationId, stmt->oldNode,
- indexInfo, indexColNames,
+ index_create(rel, indexRelationName, indexRelationId, parentIndexId,
+ stmt->oldNode, indexInfo, indexColNames,
accessMethodId, tablespaceId,
collationObjectId, classObjectId,
coloptions, reloptions,
@@ -706,6 +754,144 @@ DefineIndex(Oid relationId,
CreateComments(indexRelationId, RelationRelationId, 0,
stmt->idxcomment);
+ if (partitioned)
+ {
+ /*
+ * Unless caller specified to skip this step (via ONLY), process
+ * each partition to make sure they all contain a corresponding index.
+ *
+ * If we're called internally (no stmt->relation), recurse always.
+ */
+ if (!stmt->relation || stmt->relation->inh)
+ {
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ int nparts = partdesc->nparts;
+ Oid *part_oids;
+ TupleDesc parentDesc;
+ bool invalidate_parent = false;
+
+ nparts = partdesc->nparts;
+ part_oids = palloc(sizeof(Oid) * nparts);
+ memcpy(part_oids, partdesc->oids, sizeof(Oid) * nparts);
+
+ parentDesc = CreateTupleDescCopy(RelationGetDescr(rel));
+
+ heap_close(rel, NoLock);
+
+ /*
+ * For each partition, scan all existing indexes; if one matches
+ * our index definition and is not already attached to some other
+ * parent index, attach it to the one we just created.
+ *
+ * If none matches, build a new index by calling ourselves
+ * recursively with the same options (except for the index name).
+ */
+ for (i = 0; i < nparts; i++)
+ {
+ Oid childRelid = part_oids[i];
+ Relation childrel;
+ List *childidxs;
+ ListCell *cell;
+ AttrNumber *attmap = NULL;
+ bool found = false;
+
+ childrel = heap_open(childRelid, lockmode);
+ childidxs = RelationGetIndexList(childrel);
+
+ foreach(cell, childidxs)
+ {
+ Oid cldidxid = lfirst_oid(cell);
+ Relation cldidx;
+ IndexInfo *cldIdxInfo;
+
+ /* this index is already partition of another one */
+ if (has_superclass(cldidxid))
+ continue;
+
+ cldidx = index_open(cldidxid, lockmode);
+
+ cldIdxInfo = BuildIndexInfo(cldidx);
+ if (attmap == NULL)
+ attmap =
+ convert_tuples_by_name_map(RelationGetDescr(childrel),
+ parentDesc,
+ gettext_noop("could not convert row type"));
+
+ if (CompareIndexInfo(cldIdxInfo, indexInfo, attmap, parentDesc->natts))
+ {
+ /*
+ * Found a match. Attach index to parent and we're
+ * done, but keep lock till commit.
+ */
+ IndexSetParentIndex(cldidx, indexRelationId);
+
+ if (!IndexIsValid(cldidx->rd_index))
+ invalidate_parent = true;
+
+ found = true;
+ index_close(cldidx, NoLock);
+ break;
+ }
+
+ index_close(cldidx, lockmode);
+ }
+
+ list_free(childidxs);
+ heap_close(childrel, NoLock);
+ if (attmap)
+ pfree(attmap);
+
+ /*
+ * If no matching index was found, create our own.
+ */
+ if (!found)
+ {
+ IndexStmt *childStmt = copyObject(stmt);
+
+ childStmt->idxname = NULL;
+ childStmt->relationId = childRelid;
+ DefineIndex(childRelid, childStmt,
+ InvalidOid, /* no predefined OID */
+ indexRelationId, /* this is our child */
+ false, check_rights, check_not_in_use,
+ false, quiet);
+ }
+ }
+
+ /*
+ * The pg_index row we inserted for this index was marked
+ * indisvalid=true. But if we attached an existing index that
+ * is invalid, this is incorrect, so update our row to
+ * invalid too.
+ */
+ if (invalidate_parent)
+ {
+ Relation pg_index = heap_open(IndexRelationId, RowExclusiveLock);
+ HeapTuple tup,
+ newtup;
+
+ tup = SearchSysCache1(INDEXRELID,
+ ObjectIdGetDatum(indexRelationId));
+ if (!tup)
+ elog(ERROR, "cache lookup failed for index %u",
+ indexRelationId);
+ newtup = heap_copytuple(tup);
+ ((Form_pg_index) GETSTRUCT(newtup))->indisvalid = false;
+ CatalogTupleUpdate(pg_index, &tup->t_self, newtup);
+ ReleaseSysCache(tup);
+ heap_close(pg_index, RowExclusiveLock);
+ }
+ }
+ else
+ heap_close(rel, NoLock);
+
+ /*
+ * Indexes on partitioned tables are not themselves built, so we're
+ * done here.
+ */
+ return address;
+ }
+
if (!stmt->concurrent)
{
/* Close the heap and we're done, in the non-concurrent case */
@@ -1765,7 +1951,7 @@ ChooseIndexColumnNames(List *indexElems)
* ReindexIndex
* Recreate a specific index.
*/
-Oid
+void
ReindexIndex(RangeVar *indexRelation, int options)
{
Oid indOid;
@@ -1788,12 +1974,17 @@ ReindexIndex(RangeVar *indexRelation, int options)
* lock on the index.
*/
irel = index_open(indOid, NoLock);
+
+ if (irel->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
+ {
+ ReindexPartitionedIndex(irel);
+ return;
+ }
+
persistence = irel->rd_rel->relpersistence;
index_close(irel, NoLock);
reindex_index(indOid, false, persistence, options);
-
- return indOid;
}
/*
@@ -1832,7 +2023,8 @@ RangeVarCallbackForReindexIndex(const RangeVar *relation,
relkind = get_rel_relkind(relId);
if (!relkind)
return;
- if (relkind != RELKIND_INDEX)
+ if (relkind != RELKIND_INDEX &&
+ relkind != RELKIND_PARTITIONED_INDEX)
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is not an index", relation->relname)));
@@ -1976,6 +2168,12 @@ ReindexMultipleTables(const char *objectName, ReindexObjectType objectKind,
/*
* Only regular tables and matviews can have indexes, so ignore any
* other kind of relation.
+ *
+ * It is tempting to also consider partitioned tables here, but that
+ * has the problem that if the children are in the same schema, they
+ * would be processed twice. Maybe we could have a separate list of
+ * partitioned tables, and expand that afterwards into relids,
+ * ignoring any duplicates.
*/
if (classtuple->relkind != RELKIND_RELATION &&
classtuple->relkind != RELKIND_MATVIEW)
@@ -2038,3 +2236,156 @@ ReindexMultipleTables(const char *objectName, ReindexObjectType objectKind,
MemoryContextDelete(private_context);
}
+
+/*
+ * Reindex each child of a partitioned index.
+ *
+ * The parent index is given, locked in AccessExclusive mode; this routine
+ * obtains the list of children and releases the lock on parent before
+ * applying reindex on each child.
+ */
+static void
+ReindexPartitionedIndex(Relation parentIdx)
+{
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("REINDEX is not yet implemented for partitioned indexes")));
+}
+
+/*
+ * Insert or delete an appropriate pg_inherits tuple to make the given index
+ * be a partition of the indicated parent index.
+ *
+ * This also corrects the pg_depend information for the affected index.
+ */
+void
+IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
+{
+ Relation pg_inherits;
+ ScanKeyData key[2];
+ SysScanDesc scan;
+ Oid partRelid = RelationGetRelid(partitionIdx);
+ HeapTuple tuple;
+ bool fix_dependencies;
+
+ /* Make sure this is an index */
+ Assert(partitionIdx->rd_rel->relkind == RELKIND_INDEX ||
+ partitionIdx->rd_rel->relkind == RELKIND_PARTITIONED_INDEX);
+
+ /*
+ * Scan pg_inherits for rows linking our index to some parent.
+ */
+ pg_inherits = relation_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key[0],
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(partRelid));
+ ScanKeyInit(&key[1],
+ Anum_pg_inherits_inhseqno,
+ BTEqualStrategyNumber, F_INT4EQ,
+ Int32GetDatum(1));
+ scan = systable_beginscan(pg_inherits, InheritsRelidSeqnoIndexId, true,
+ NULL, 2, key);
+ tuple = systable_getnext(scan);
+
+ if (!HeapTupleIsValid(tuple))
+ {
+ if (parentOid == InvalidOid)
+ {
+ /*
+ * No pg_inherits row, and no parent wanted: nothing to do in
+ * this case.
+ */
+ fix_dependencies = false;
+ }
+ else
+ {
+ Datum values[Natts_pg_inherits];
+ bool isnull[Natts_pg_inherits];
+
+ /*
+ * No pg_inherits row exists, and we want a parent for this index,
+ * so insert it.
+ */
+ values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(partRelid);
+ values[Anum_pg_inherits_inhparent - 1] =
+ ObjectIdGetDatum(parentOid);
+ values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ memset(isnull, false, sizeof(isnull));
+
+ tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
+ values, isnull);
+ CatalogTupleInsert(pg_inherits, tuple);
+
+ fix_dependencies = true;
+ }
+ }
+ else
+ {
+ Form_pg_inherits inhForm = (Form_pg_inherits) GETSTRUCT(tuple);
+
+ if (parentOid == InvalidOid)
+ {
+ /*
+ * There exists a pg_inherits row, which we want to clear; do so.
+ */
+ CatalogTupleDelete(pg_inherits, &tuple->t_self);
+ fix_dependencies = true;
+ }
+ else
+ {
+ /*
+ * A pg_inherits row exists. If it's the same we want, then we're
+ * good; if it differs, that amounts to a corrupt catalog and
+ * should not happen.
+ */
+ if (inhForm->inhparent != parentOid)
+ {
+ /* unexpected: we should not get called in this case */
+ elog(ERROR, "bogus pg_inherit row: inhrelid %u inhparent %u",
+ inhForm->inhrelid, inhForm->inhparent);
+ }
+
+ /* already in the right state */
+ fix_dependencies = false;
+ }
+ }
+
+ /* done with pg_inherits */
+ systable_endscan(scan);
+ relation_close(pg_inherits, RowExclusiveLock);
+
+ if (fix_dependencies)
+ {
+ ObjectAddress partIdx;
+
+ /*
+ * Insert/delete pg_depend rows. If setting a parent, add an
+ * INTERNAL_AUTO dependency to the parent index; if making standalone,
+ * remove all existing rows and put back the regular dependency on the
+ * table.
+ */
+ ObjectAddressSet(partIdx, RelationRelationId, partRelid);
+
+ if (OidIsValid(parentOid))
+ {
+ ObjectAddress parentIdx;
+
+ ObjectAddressSet(parentIdx, RelationRelationId, parentOid);
+ recordDependencyOn(&partIdx, &parentIdx, DEPENDENCY_INTERNAL_AUTO);
+ }
+ else
+ {
+ ObjectAddress partitionTbl;
+
+ ObjectAddressSet(partitionTbl, RelationRelationId,
+ partitionIdx->rd_index->indrelid);
+
+ deleteDependencyRecordsForClass(RelationRelationId, partRelid,
+ RelationRelationId,
+ DEPENDENCY_INTERNAL_AUTO);
+
+ recordDependencyOn(&partIdx, &partitionTbl, DEPENDENCY_AUTO);
+ }
+ }
+}
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index f2a928b823..c8575ea17b 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -266,6 +266,12 @@ static const struct dropmsgstrings dropmsgstringarray[] = {
gettext_noop("table \"%s\" does not exist, skipping"),
gettext_noop("\"%s\" is not a table"),
gettext_noop("Use DROP TABLE to remove a table.")},
+ {RELKIND_PARTITIONED_INDEX,
+ ERRCODE_UNDEFINED_OBJECT,
+ gettext_noop("index \"%s\" does not exist"),
+ gettext_noop("index \"%s\" does not exist, skipping"),
+ gettext_noop("\"%s\" is not an index"),
+ gettext_noop("Use DROP INDEX to remove an index.")},
{'\0', 0, NULL, NULL, NULL, NULL}
};
@@ -475,11 +481,17 @@ static void CreateInheritance(Relation child_rel, Relation parent_rel);
static void RemoveInheritance(Relation child_rel, Relation parent_rel);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd);
+static void AttachPartitionEnsureIndexes(Relation rel, Relation attachrel);
static void ValidatePartitionConstraints(List **wqueue, Relation scanrel,
List *scanrel_children,
List *partConstraint,
bool validate_default);
static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
+ RangeVar *name);
+static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
+static void refuseDupeIndexAttach(Relation parentIdx, Relation partIdx,
+ Relation partitionTbl);
/* ----------------------------------------------------------------
@@ -897,6 +909,53 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
StorePartitionKey(rel, strategy, partnatts, partattrs, partexprs,
partopclass, partcollation);
+
+ /* make it all visible */
+ CommandCounterIncrement();
+ }
+
+ /*
+ * If we're creating a partition, create now all the indexes defined in
+ * the parent. We can't do it earlier, because DefineIndex wants to know
+ * the partition key which we just stored.
+ */
+ if (stmt->partbound)
+ {
+ Oid parentId = linitial_oid(inheritOids);
+ Relation parent;
+ List *idxlist;
+ ListCell *cell;
+
+ /* Already have strong enough lock on the parent */
+ parent = heap_open(parentId, NoLock);
+ idxlist = RelationGetIndexList(parent);
+
+ /*
+ * For each index in the parent table, create one in the partition
+ */
+ foreach(cell, idxlist)
+ {
+ Relation idxRel = index_open(lfirst_oid(cell), AccessShareLock);
+ AttrNumber *attmap;
+ IndexStmt *idxstmt;
+
+ attmap = convert_tuples_by_name_map(RelationGetDescr(rel),
+ RelationGetDescr(parent),
+ gettext_noop("could not convert row type"));
+ idxstmt =
+ generateClonedIndexStmt(NULL, RelationGetRelid(rel), idxRel,
+ attmap, RelationGetDescr(rel)->natts);
+ DefineIndex(RelationGetRelid(rel),
+ idxstmt,
+ InvalidOid,
+ RelationGetRelid(idxRel),
+ false, false, false, false, false);
+
+ index_close(idxRel, AccessShareLock);
+ }
+
+ list_free(idxlist);
+ heap_close(parent, NoLock);
}
/*
@@ -1179,10 +1238,13 @@ RangeVarCallbackForDropRelation(const RangeVar *rel, Oid relOid, Oid oldRelOid,
* but RemoveRelations() can only pass one relkind for a given relation.
* It chooses RELKIND_RELATION for both regular and partitioned tables.
* That means we must be careful before giving the wrong type error when
- * the relation is RELKIND_PARTITIONED_TABLE.
+ * the relation is RELKIND_PARTITIONED_TABLE. An equivalent problem
+ * exists with indexes.
*/
if (classform->relkind == RELKIND_PARTITIONED_TABLE)
expected_relkind = RELKIND_RELATION;
+ else if (classform->relkind == RELKIND_PARTITIONED_INDEX)
+ expected_relkind = RELKIND_INDEX;
else
expected_relkind = classform->relkind;
@@ -1210,7 +1272,8 @@ RangeVarCallbackForDropRelation(const RangeVar *rel, Oid relOid, Oid oldRelOid,
* we do it the other way around. No error if we don't find a pg_index
* entry, though --- the relation may have been dropped.
*/
- if (relkind == RELKIND_INDEX && relOid != oldRelOid)
+ if ((relkind == RELKIND_INDEX || relkind == RELKIND_PARTITIONED_INDEX) &&
+ relOid != oldRelOid)
{
state->heapOid = IndexGetRelation(relOid, true);
if (OidIsValid(state->heapOid))
@@ -2396,27 +2459,11 @@ StoreCatalogInheritance1(Oid relationId, Oid parentOid,
int16 seqNumber, Relation inhRelation,
bool child_is_partition)
{
- TupleDesc desc = RelationGetDescr(inhRelation);
- Datum values[Natts_pg_inherits];
- bool nulls[Natts_pg_inherits];
ObjectAddress childobject,
parentobject;
- HeapTuple tuple;
- /*
- * Make the pg_inherits entry
- */
- values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
- values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
- values[Anum_pg_inherits_inhseqno - 1] = Int16GetDatum(seqNumber);
-
- memset(nulls, 0, sizeof(nulls));
-
- tuple = heap_form_tuple(desc, values, nulls);
-
- CatalogTupleInsert(inhRelation, tuple);
-
- heap_freetuple(tuple);
+ /* store the pg_inherits row */
+ StoreSingleInheritance(relationId, parentOid, seqNumber);
/*
* Store a dependency too
@@ -2540,6 +2587,7 @@ renameatt_check(Oid myrelid, Form_pg_class classform, bool recursing)
relkind != RELKIND_MATVIEW &&
relkind != RELKIND_COMPOSITE_TYPE &&
relkind != RELKIND_INDEX &&
+ relkind != RELKIND_PARTITIONED_INDEX &&
relkind != RELKIND_FOREIGN_TABLE &&
relkind != RELKIND_PARTITIONED_TABLE)
ereport(ERROR,
@@ -3019,7 +3067,8 @@ RenameRelationInternal(Oid myrelid, const char *newrelname, bool is_internal)
/*
* Also rename the associated constraint, if any.
*/
- if (targetrelation->rd_rel->relkind == RELKIND_INDEX)
+ if (targetrelation->rd_rel->relkind == RELKIND_INDEX ||
+ targetrelation->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
{
Oid constraintId = get_index_constraint(myrelid);
@@ -3073,6 +3122,7 @@ CheckTableNotInUse(Relation rel, const char *stmt)
stmt, RelationGetRelationName(rel))));
if (rel->rd_rel->relkind != RELKIND_INDEX &&
+ rel->rd_rel->relkind != RELKIND_PARTITIONED_INDEX &&
AfterTriggerPendingOnRel(RelationGetRelid(rel)))
ereport(ERROR,
(errcode(ERRCODE_OBJECT_IN_USE),
@@ -3764,6 +3814,10 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
pass = AT_PASS_MISC;
break;
case AT_AttachPartition:
+ ATSimplePermissions(rel, ATT_TABLE | ATT_INDEX);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
case AT_DetachPartition:
ATSimplePermissions(rel, ATT_TABLE);
/* No command-specific prep needed */
@@ -4112,9 +4166,14 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab, Relation rel,
ATExecGenericOptions(rel, (List *) cmd->def);
break;
case AT_AttachPartition:
- ATExecAttachPartition(wqueue, rel, (PartitionCmd *) cmd->def);
+ if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
+ ATExecAttachPartition(wqueue, rel, (PartitionCmd *) cmd->def);
+ else
+ ATExecAttachPartitionIdx(wqueue, rel,
+ ((PartitionCmd *) cmd->def)->name);
break;
case AT_DetachPartition:
+ Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
@@ -4148,9 +4207,13 @@ ATRewriteTables(AlterTableStmt *parsetree, List **wqueue, LOCKMODE lockmode)
{
AlteredTableInfo *tab = (AlteredTableInfo *) lfirst(ltab);
- /* Foreign tables have no storage, nor do partitioned tables. */
+ /*
+ * Foreign tables have no storage, nor do partitioned tables and
+ * indexes.
+ */
if (tab->relkind == RELKIND_FOREIGN_TABLE ||
- tab->relkind == RELKIND_PARTITIONED_TABLE)
+ tab->relkind == RELKIND_PARTITIONED_TABLE ||
+ tab->relkind == RELKIND_PARTITIONED_INDEX)
continue;
/*
@@ -4750,6 +4813,7 @@ ATSimplePermissions(Relation rel, int allowed_targets)
actual_target = ATT_MATVIEW;
break;
case RELKIND_INDEX:
+ case RELKIND_PARTITIONED_INDEX:
actual_target = ATT_INDEX;
break;
case RELKIND_COMPOSITE_TYPE:
@@ -6194,6 +6258,7 @@ ATPrepSetStatistics(Relation rel, const char *colName, int16 colNum, Node *newVa
if (rel->rd_rel->relkind != RELKIND_RELATION &&
rel->rd_rel->relkind != RELKIND_MATVIEW &&
rel->rd_rel->relkind != RELKIND_INDEX &&
+ rel->rd_rel->relkind != RELKIND_PARTITIONED_INDEX &&
rel->rd_rel->relkind != RELKIND_FOREIGN_TABLE &&
rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
ereport(ERROR,
@@ -6205,7 +6270,9 @@ ATPrepSetStatistics(Relation rel, const char *colName, int16 colNum, Node *newVa
* We allow referencing columns by numbers only for indexes, since table
* column numbers could contain gaps if columns are later dropped.
*/
- if (rel->rd_rel->relkind != RELKIND_INDEX && !colName)
+ if (rel->rd_rel->relkind != RELKIND_INDEX &&
+ rel->rd_rel->relkind != RELKIND_PARTITIONED_INDEX &&
+ !colName)
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot refer to non-index column by number")));
@@ -6283,7 +6350,8 @@ ATExecSetStatistics(Relation rel, const char *colName, int16 colNum, Node *newVa
errmsg("cannot alter system column \"%s\"",
colName)));
- if (rel->rd_rel->relkind == RELKIND_INDEX &&
+ if ((rel->rd_rel->relkind == RELKIND_INDEX ||
+ rel->rd_rel->relkind == RELKIND_PARTITIONED_INDEX) &&
rel->rd_index->indkey.values[attnum - 1] != 0)
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
@@ -6736,6 +6804,7 @@ ATExecAddIndex(AlteredTableInfo *tab, Relation rel,
address = DefineIndex(RelationGetRelid(rel),
stmt,
InvalidOid, /* no predefined OID */
+ InvalidOid, /* no parent index */
true, /* is_alter_table */
check_rights,
false, /* check_not_in_use - we did it already */
@@ -9139,7 +9208,8 @@ ATExecAlterColumnType(AlteredTableInfo *tab, Relation rel,
{
char relKind = get_rel_relkind(foundObject.objectId);
- if (relKind == RELKIND_INDEX)
+ if (relKind == RELKIND_INDEX ||
+ relKind == RELKIND_PARTITIONED_INDEX)
{
Assert(foundObject.objectSubId == 0);
if (!list_member_oid(tab->changedIndexOids, foundObject.objectId))
@@ -9982,6 +10052,15 @@ ATExecChangeOwner(Oid relationOid, Oid newOwnerId, bool recursing, LOCKMODE lock
newOwnerId = tuple_class->relowner;
}
break;
+ case RELKIND_PARTITIONED_INDEX:
+ if (recursing)
+ break;
+ ereport(ERROR,
+ (errcode(ERRCODE_WRONG_OBJECT_TYPE),
+ errmsg("cannot change owner of index \"%s\"",
+ NameStr(tuple_class->relname)),
+ errhint("Change the ownership of the index's table, instead.")));
+ break;
case RELKIND_SEQUENCE:
if (!recursing &&
tuple_class->relowner != newOwnerId)
@@ -10103,6 +10182,7 @@ ATExecChangeOwner(Oid relationOid, Oid newOwnerId, bool recursing, LOCKMODE lock
*/
if (tuple_class->relkind != RELKIND_COMPOSITE_TYPE &&
tuple_class->relkind != RELKIND_INDEX &&
+ tuple_class->relkind != RELKIND_PARTITIONED_INDEX &&
tuple_class->relkind != RELKIND_TOASTVALUE)
changeDependencyOnOwner(RelationRelationId, relationOid,
newOwnerId);
@@ -10110,7 +10190,8 @@ ATExecChangeOwner(Oid relationOid, Oid newOwnerId, bool recursing, LOCKMODE lock
/*
* Also change the ownership of the table's row type, if it has one
*/
- if (tuple_class->relkind != RELKIND_INDEX)
+ if (tuple_class->relkind != RELKIND_INDEX &&
+ tuple_class->relkind != RELKIND_PARTITIONED_INDEX)
AlterTypeOwnerInternal(tuple_class->reltype, newOwnerId);
/*
@@ -10119,6 +10200,7 @@ ATExecChangeOwner(Oid relationOid, Oid newOwnerId, bool recursing, LOCKMODE lock
* relation, as well as its toast table (if it has one).
*/
if (tuple_class->relkind == RELKIND_RELATION ||
+ tuple_class->relkind == RELKIND_PARTITIONED_TABLE ||
tuple_class->relkind == RELKIND_MATVIEW ||
tuple_class->relkind == RELKIND_TOASTVALUE)
{
@@ -10427,6 +10509,7 @@ ATExecSetRelOptions(Relation rel, List *defList, AlterTableType operation,
(void) view_reloptions(newOptions, true);
break;
case RELKIND_INDEX:
+ case RELKIND_PARTITIONED_INDEX:
(void) index_reloptions(rel->rd_amroutine->amoptions, newOptions, true);
break;
default:
@@ -10839,7 +10922,8 @@ AlterTableMoveAll(AlterTableMoveAllStmt *stmt)
relForm->relkind != RELKIND_RELATION &&
relForm->relkind != RELKIND_PARTITIONED_TABLE) ||
(stmt->objtype == OBJECT_INDEX &&
- relForm->relkind != RELKIND_INDEX) ||
+ relForm->relkind != RELKIND_INDEX &&
+ relForm->relkind != RELKIND_PARTITIONED_INDEX) ||
(stmt->objtype == OBJECT_MATVIEW &&
relForm->relkind != RELKIND_MATVIEW))
continue;
@@ -11633,45 +11717,18 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
Relation catalogRelation;
SysScanDesc scan;
ScanKeyData key[3];
- HeapTuple inheritsTuple,
- attributeTuple,
+ HeapTuple attributeTuple,
constraintTuple;
List *connames;
- bool found = false;
+ bool found;
bool child_is_partition = false;
/* If parent_rel is a partitioned table, child_rel must be a partition */
if (parent_rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
child_is_partition = true;
- /*
- * Find and destroy the pg_inherits entry linking the two, or error out if
- * there is none.
- */
- catalogRelation = heap_open(InheritsRelationId, RowExclusiveLock);
- ScanKeyInit(&key[0],
- Anum_pg_inherits_inhrelid,
- BTEqualStrategyNumber, F_OIDEQ,
- ObjectIdGetDatum(RelationGetRelid(child_rel)));
- scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
- true, NULL, 1, key);
-
- while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
- {
- Oid inhparent;
-
- inhparent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
- if (inhparent == RelationGetRelid(parent_rel))
- {
- CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
- found = true;
- break;
- }
- }
-
- systable_endscan(scan);
- heap_close(catalogRelation, RowExclusiveLock);
-
+ found = DeleteInheritsTuple(RelationGetRelid(child_rel),
+ RelationGetRelid(parent_rel));
if (!found)
{
if (child_is_partition)
@@ -13226,7 +13283,8 @@ RangeVarCallbackForAlterRelation(const RangeVar *rv, Oid relid, Oid oldrelid,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
errmsg("\"%s\" is not a composite type", rv->relname)));
- if (reltype == OBJECT_INDEX && relkind != RELKIND_INDEX
+ if (reltype == OBJECT_INDEX && relkind != RELKIND_INDEX &&
+ relkind != RELKIND_PARTITIONED_INDEX
&& !IsA(stmt, RenameStmt))
ereport(ERROR,
(errcode(ERRCODE_WRONG_OBJECT_TYPE),
@@ -13946,6 +14004,9 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
/* Update the pg_class entry. */
StorePartitionBound(attachrel, rel, cmd->bound);
+ /* Ensure there exists a correct set of indexes in the partition. */
+ AttachPartitionEnsureIndexes(rel, attachrel);
+
/*
* Generate partition constraint from the partition bound specification.
* If the parent itself is a partition, make sure to include its
@@ -14016,6 +14077,123 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
}
/*
+ * AttachPartitionEnsureIndexes
+ *
+ * Enforce the indexing rule for partitioned tables during ALTER TABLE / ATTACH
+ * PARTITION: every partition must have an index attached to each index on the
+ * partitioned table.
+ */
+static void
+AttachPartitionEnsureIndexes(Relation rel, Relation attachrel)
+{
+ AttrNumber *attmap = NULL;
+ List *idxes;
+ List *attachRelIdxs;
+ Relation *attachrelIdxRels;
+ IndexInfo **attachInfos;
+ int i;
+ ListCell *cell;
+ MemoryContext cxt;
+ MemoryContext oldcxt;
+
+ cxt = AllocSetContextCreate(CurrentMemoryContext,
+ "AttachPartitionEnsureIndexes",
+ ALLOCSET_DEFAULT_SIZES);
+ oldcxt = MemoryContextSwitchTo(cxt);
+
+ idxes = RelationGetIndexList(rel);
+ attachRelIdxs = RelationGetIndexList(attachrel);
+ attachrelIdxRels = palloc(sizeof(Relation) * list_length(attachRelIdxs));
+ attachInfos = palloc(sizeof(IndexInfo *) * list_length(attachRelIdxs));
+
+ /* Build arrays of all existing indexes and their IndexInfos */
+ i = 0;
+ foreach(cell, attachRelIdxs)
+ {
+ Oid cldIdxId = lfirst_oid(cell);
+
+ attachrelIdxRels[i] = index_open(cldIdxId, AccessShareLock);
+ attachInfos[i] = BuildIndexInfo(attachrelIdxRels[i]);
+ i++;
+ }
+
+ /*
+ * For each index on the partitioned table, find a matching one in the
+ * partition-to-be; if one is not found, create one.
+ */
+ foreach(cell, idxes)
+ {
+ Oid idx = lfirst_oid(cell);
+ Relation idxRel = index_open(idx, AccessShareLock);
+ IndexInfo *info;
+ bool found = false;
+
+ /*
+ * Ignore indexes in the partitioned table other than partitioned
+ * indexes.
+ */
+ if (idxRel->rd_rel->relkind != RELKIND_PARTITIONED_INDEX)
+ {
+ index_close(idxRel, AccessShareLock);
+ continue;
+ }
+
+ /* construct an indexinfo to compare existing indexes against */
+ info = BuildIndexInfo(idxRel);
+ if (attmap == NULL)
+ attmap =
+ convert_tuples_by_name_map(RelationGetDescr(attachrel),
+ RelationGetDescr(rel),
+ gettext_noop("could not convert row type"));
+
+ /*
+ * Scan the list of existing indexes in the partition-to-be, and mark
+ * the first matching, unattached one we find, if any, as partition of
+ * the parent index. If we find one, we're done.
+ */
+ for (i = 0; i < list_length(attachRelIdxs); i++)
+ {
+ /* does this index have a parent? if so, can't use it */
+ if (has_superclass(RelationGetRelid(attachrelIdxRels[i])))
+ continue;
+
+ if (CompareIndexInfo(info, attachInfos[i], attmap,
+ RelationGetDescr(rel)->natts))
+ {
+ /* bingo. */
+ IndexSetParentIndex(attachrelIdxRels[i], idx);
+ found = true;
+ break;
+ }
+ }
+
+ /*
+ * If no suitable index was found in the partition-to-be, create one
+ * now.
+ */
+ if (!found)
+ {
+ IndexStmt *stmt;
+
+ stmt = generateClonedIndexStmt(NULL, RelationGetRelid(attachrel),
+ idxRel, attmap,
+ RelationGetDescr(rel)->natts);
+ DefineIndex(RelationGetRelid(attachrel), stmt, InvalidOid,
+ RelationGetRelid(idxRel),
+ false, false, false, false, false);
+ }
+
+ index_close(idxRel, AccessShareLock);
+ }
+
+ /* Clean up. */
+ for (i = 0; i < list_length(attachRelIdxs); i++)
+ index_close(attachrelIdxRels[i], AccessShareLock);
+ MemoryContextSwitchTo(oldcxt);
+ MemoryContextDelete(cxt);
+}
+
+/*
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
@@ -14033,6 +14211,8 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
new_repl[Natts_pg_class];
ObjectAddress address;
Oid defaultPartOid;
+ List *indexes;
+ ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
@@ -14094,6 +14274,24 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
}
}
+ /* detach indexes too */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+ relation_close(idx, AccessExclusiveLock);
+ }
+
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
@@ -14107,3 +14305,323 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
return address;
}
+
+/*
+ * Before acquiring lock on an index, acquire the same lock on the owning
+ * table.
+ */
+struct AttachIndexCallbackState
+{
+ Oid partitionOid;
+ Oid parentTblOid;
+ bool lockedParentTbl;
+};
+
+static void
+RangeVarCallbackForAttachIndex(const RangeVar *rv, Oid relOid, Oid oldRelOid,
+ void *arg)
+{
+ struct AttachIndexCallbackState *state;
+ Form_pg_class classform;
+ HeapTuple tuple;
+
+ state = (struct AttachIndexCallbackState *) arg;
+
+ if (!state->lockedParentTbl)
+ {
+ LockRelationOid(state->parentTblOid, AccessShareLock);
+ state->lockedParentTbl = true;
+ }
+
+ /*
+ * If we previously locked some other heap, and the name we're looking up
+ * no longer refers to an index on that relation, release the now-useless
+ * lock. XXX maybe we should do *after* we verify whether the index does
+ * not actually belong to the same relation ...
+ */
+ if (relOid != oldRelOid && OidIsValid(state->partitionOid))
+ {
+ UnlockRelationOid(state->partitionOid, AccessShareLock);
+ state->partitionOid = InvalidOid;
+ }
+
+ /* Didn't find a relation, so no need for locking or permission checks. */
+ if (!OidIsValid(relOid))
+ return;
+
+ tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(relOid));
+ if (!HeapTupleIsValid(tuple))
+ return; /* concurrently dropped, so nothing to do */
+ classform = (Form_pg_class) GETSTRUCT(tuple);
+ if (classform->relkind != RELKIND_PARTITIONED_INDEX &&
+ classform->relkind != RELKIND_INDEX)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_OBJECT_DEFINITION),
+ errmsg("\"%s\" is not an index", rv->relname)));
+ ReleaseSysCache(tuple);
+
+ /*
+ * Since we need only examine the heap's tupledesc, an access share lock
+ * on it (preventing any DDL) is sufficient.
+ */
+ state->partitionOid = IndexGetRelation(relOid, false);
+ LockRelationOid(state->partitionOid, AccessShareLock);
+}
+
+/*
+ * ALTER INDEX i1 ATTACH PARTITION i2
+ */
+static ObjectAddress
+ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
+{
+ Relation partIdx;
+ Relation partTbl;
+ Relation parentTbl;
+ ObjectAddress address;
+ Oid partIdxId;
+ Oid currParent;
+ struct AttachIndexCallbackState state;
+
+ /*
+ * We need to obtain lock on the index 'name' to modify it, but we also
+ * need to read its owning table's tuple descriptor -- so we need to lock
+ * both. To avoid deadlocks, obtain lock on the table before doing so on
+ * the index. Furthermore, we need to examine the parent table of the
+ * partition, so lock that one too.
+ */
+ state.partitionOid = InvalidOid;
+ state.parentTblOid = parentIdx->rd_index->indrelid;
+ state.lockedParentTbl = false;
+ partIdxId =
+ RangeVarGetRelidExtended(name, AccessExclusiveLock, false, false,
+ RangeVarCallbackForAttachIndex,
+ (void *) &state);
+ /* Not there? */
+ if (!OidIsValid(partIdxId))
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_OBJECT),
+ errmsg("index \"%s\" does not exist", name->relname)));
+
+ /* no deadlock risk: RangeVarGetRelidExtended already acquired the lock */
+ partIdx = relation_open(partIdxId, AccessExclusiveLock);
+
+ /* we already hold locks on both tables, so this is safe: */
+ parentTbl = relation_open(parentIdx->rd_index->indrelid, AccessShareLock);
+ partTbl = relation_open(partIdx->rd_index->indrelid, NoLock);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partIdx));
+
+ /* Silently do nothing if already in the right state */
+ currParent = !has_superclass(partIdxId) ? InvalidOid :
+ get_partition_parent(partIdxId);
+ if (currParent != RelationGetRelid(parentIdx))
+ {
+ IndexInfo *childInfo;
+ IndexInfo *parentInfo;
+ AttrNumber *attmap;
+ bool found;
+ int i;
+ PartitionDesc partDesc;
+
+ /*
+ * If this partition already has an index attached, refuse the operation.
+ */
+ refuseDupeIndexAttach(parentIdx, partIdx, partTbl);
+
+ if (OidIsValid(currParent))
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot attach index \"%s\" as a partition of index \"%s\"",
+ RelationGetRelationName(partIdx),
+ RelationGetRelationName(parentIdx)),
+ errdetail("Index \"%s\" is already attached to another index.",
+ RelationGetRelationName(partIdx))));
+
+ /* Make sure it indexes a partition of the other index's table */
+ partDesc = RelationGetPartitionDesc(parentTbl);
+ found = false;
+ for (i = 0; i < partDesc->nparts; i++)
+ {
+ if (partDesc->oids[i] == state.partitionOid)
+ {
+ found = true;
+ break;
+ }
+ }
+ if (!found)
+ ereport(ERROR,
+ (errmsg("cannot attach index \"%s\" as a partition of index \"%s\"",
+ RelationGetRelationName(partIdx),
+ RelationGetRelationName(parentIdx)),
+ errdetail("Index \"%s\" is not on a partition of table \"%s\".",
+ RelationGetRelationName(partIdx),
+ RelationGetRelationName(parentTbl))));
+
+ /* Ensure the indexes are compatible */
+ childInfo = BuildIndexInfo(partIdx);
+ parentInfo = BuildIndexInfo(parentIdx);
+ attmap = convert_tuples_by_name_map(RelationGetDescr(parentTbl),
+ RelationGetDescr(partTbl),
+ gettext_noop("could not convert row type"));
+ if (!CompareIndexInfo(parentInfo, childInfo, attmap,
+ RelationGetDescr(partTbl)->natts))
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_OBJECT_DEFINITION),
+ errmsg("cannot attach index \"%s\" as a partition of index \"%s\"",
+ RelationGetRelationName(partIdx),
+ RelationGetRelationName(parentIdx)),
+ errdetail("The index definitions do not match.")));
+
+ /* All good -- do it */
+ IndexSetParentIndex(partIdx, RelationGetRelid(parentIdx));
+ pfree(attmap);
+
+ CommandCounterIncrement();
+
+ validatePartitionedIndex(parentIdx, parentTbl);
+ }
+
+ relation_close(parentTbl, AccessShareLock);
+ /* keep these locks till commit */
+ relation_close(partTbl, NoLock);
+ relation_close(partIdx, NoLock);
+
+ return address;
+}
+
+/*
+ * Verify whether the given partition already contains an index attached
+ * to the given partitioned index. If so, raise an error.
+ */
+static void
+refuseDupeIndexAttach(Relation parentIdx, Relation partIdx, Relation partitionTbl)
+{
+ Relation pg_inherits;
+ ScanKeyData key;
+ HeapTuple tuple;
+ SysScanDesc scan;
+
+ pg_inherits = heap_open(InheritsRelationId, AccessShareLock);
+ ScanKeyInit(&key, Anum_pg_inherits_inhparent,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(parentIdx)));
+ scan = systable_beginscan(pg_inherits, InheritsParentIndexId, true,
+ NULL, 1, &key);
+ while (HeapTupleIsValid(tuple = systable_getnext(scan)))
+ {
+ Form_pg_inherits inhForm;
+ Oid tab;
+
+ inhForm = (Form_pg_inherits) GETSTRUCT(tuple);
+ tab = IndexGetRelation(inhForm->inhrelid, false);
+ if (tab == RelationGetRelid(partitionTbl))
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot attach index \"%s\" as a partition of index \"%s\"",
+ RelationGetRelationName(partIdx),
+ RelationGetRelationName(parentIdx)),
+ errdetail("Another index is already attached for partition \"%s\".",
+ RelationGetRelationName(partitionTbl))));
+ }
+
+ systable_endscan(scan);
+ heap_close(pg_inherits, AccessShareLock);
+}
+
+/*
+ * Verify whether the set of attached partition indexes to a parent index on
+ * a partitioned table is complete. If it is, mark the parent index valid.
+ *
+ * This should be called each time a partition index is attached.
+ */
+static void
+validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
+{
+ Relation inheritsRel;
+ SysScanDesc scan;
+ ScanKeyData key;
+ int tuples = 0;
+ HeapTuple inhTup;
+ bool updated = false;
+
+ Assert(partedIdx->rd_rel->relkind == RELKIND_PARTITIONED_INDEX);
+
+ /*
+ * Scan pg_inherits for this parent index. Count each valid index we find
+ * (verifying the pg_index entry for each), and if we reach the total
+ * amount we expect, we can mark this parent index as valid.
+ */
+ inheritsRel = heap_open(InheritsRelationId, AccessShareLock);
+ ScanKeyInit(&key, Anum_pg_inherits_inhparent,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(partedIdx)));
+ scan = systable_beginscan(inheritsRel, InheritsParentIndexId, true,
+ NULL, 1, &key);
+ while ((inhTup = systable_getnext(scan)) != NULL)
+ {
+ Form_pg_inherits inhForm = (Form_pg_inherits) GETSTRUCT(inhTup);
+ HeapTuple indTup;
+ Form_pg_index indexForm;
+
+ indTup = SearchSysCache1(INDEXRELID,
+ ObjectIdGetDatum(inhForm->inhrelid));
+ if (!indTup)
+ elog(ERROR, "cache lookup failed for index %u",
+ inhForm->inhrelid);
+ indexForm = (Form_pg_index) GETSTRUCT(indTup);
+ if (IndexIsValid(indexForm))
+ tuples += 1;
+ ReleaseSysCache(indTup);
+ }
+
+ /* Done with pg_inherits */
+ systable_endscan(scan);
+ heap_close(inheritsRel, AccessShareLock);
+
+ /*
+ * If we found as many inherited indexes as the partitioned table has
+ * partitions, we're good; update pg_index to set indisvalid.
+ */
+ if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ {
+ Relation idxRel;
+ HeapTuple newtup;
+
+ idxRel = heap_open(IndexRelationId, RowExclusiveLock);
+
+ newtup = heap_copytuple(partedIdx->rd_indextuple);
+ ((Form_pg_index) GETSTRUCT(newtup))->indisvalid = true;
+ updated = true;
+
+ CatalogTupleUpdate(idxRel, &partedIdx->rd_indextuple->t_self, newtup);
+
+ heap_close(idxRel, RowExclusiveLock);
+ }
+
+ /*
+ * If this index is in turn a partition of a larger index, validating it
+ * might cause the parent to become valid also. Try that.
+ */
+ if (updated &&
+ has_superclass(RelationGetRelid(partedIdx)))
+ {
+ Oid parentIdxId,
+ parentTblId;
+ Relation parentIdx,
+ parentTbl;
+
+ /* make sure we see the validation we just did */
+ CommandCounterIncrement();
+
+ parentIdxId = get_partition_parent(RelationGetRelid(partedIdx));
+ parentTblId = get_partition_parent(RelationGetRelid(partedTbl));
+ parentIdx = relation_open(parentIdxId, AccessExclusiveLock);
+ parentTbl = relation_open(parentTblId, AccessExclusiveLock);
+ Assert(!parentIdx->rd_index->indisvalid);
+
+ validatePartitionedIndex(parentIdx, parentTbl);
+
+ relation_close(parentIdx, AccessExclusiveLock);
+ relation_close(parentTbl, AccessExclusiveLock);
+ }
+}
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index ddbbc79823..65d8c77d7a 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -3379,6 +3379,7 @@ _copyIndexStmt(const IndexStmt *from)
COPY_STRING_FIELD(idxname);
COPY_NODE_FIELD(relation);
+ COPY_SCALAR_FIELD(relationId);
COPY_STRING_FIELD(accessMethod);
COPY_STRING_FIELD(tableSpace);
COPY_NODE_FIELD(indexParams);
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index 30ccc9c5ae..0bd12e862e 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -1332,6 +1332,7 @@ _equalIndexStmt(const IndexStmt *a, const IndexStmt *b)
{
COMPARE_STRING_FIELD(idxname);
COMPARE_NODE_FIELD(relation);
+ COMPARE_SCALAR_FIELD(relationId);
COMPARE_STRING_FIELD(accessMethod);
COMPARE_STRING_FIELD(tableSpace);
COMPARE_NODE_FIELD(indexParams);
diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index 5e72df137e..b1cdfc36a6 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -2650,6 +2650,7 @@ _outIndexStmt(StringInfo str, const IndexStmt *node)
WRITE_STRING_FIELD(idxname);
WRITE_NODE_FIELD(relation);
+ WRITE_OID_FIELD(relationId);
WRITE_STRING_FIELD(accessMethod);
WRITE_STRING_FIELD(tableSpace);
WRITE_NODE_FIELD(indexParams);
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 8c60b35068..5f263395c5 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -372,7 +372,12 @@ get_relation_info(PlannerInfo *root, Oid relationObjectId, bool inhparent,
* a table, except we can be sure that the index is not larger
* than the table.
*/
- if (info->indpred == NIL)
+ if (indexRelation->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
+ {
+ info->pages = 0;
+ info->tuples = 0;
+ }
+ else if (info->indpred == NIL)
{
info->pages = RelationGetNumberOfBlocks(indexRelation);
info->tuples = rel->tuples;
@@ -387,7 +392,8 @@ get_relation_info(PlannerInfo *root, Oid relationObjectId, bool inhparent,
info->tuples = rel->tuples;
}
- if (info->relam == BTREE_AM_OID)
+ if (indexRelation->rd_rel->relkind == RELKIND_INDEX &&
+ info->relam == BTREE_AM_OID)
{
/* For btrees, get tree height while we have the index open */
info->tree_height = _bt_getrootheight(indexRelation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index e42b7caff6..93e67e8adc 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -290,7 +290,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
%type <ival> add_drop opt_asc_desc opt_nulls_order
%type <node> alter_table_cmd alter_type_cmd opt_collate_clause
- replica_identity partition_cmd
+ replica_identity partition_cmd index_partition_cmd
%type <list> alter_table_cmds alter_type_cmds
%type <list> alter_identity_column_option_list
%type <defelt> alter_identity_column_option
@@ -1891,6 +1891,15 @@ AlterTableStmt:
n->missing_ok = true;
$$ = (Node *)n;
}
+ | ALTER INDEX qualified_name index_partition_cmd
+ {
+ AlterTableStmt *n = makeNode(AlterTableStmt);
+ n->relation = $3;
+ n->cmds = list_make1($4);
+ n->relkind = OBJECT_INDEX;
+ n->missing_ok = false;
+ $$ = (Node *)n;
+ }
| ALTER INDEX ALL IN_P TABLESPACE name SET TABLESPACE name opt_nowait
{
AlterTableMoveAllStmt *n =
@@ -2025,6 +2034,22 @@ partition_cmd:
}
;
+index_partition_cmd:
+ /* ALTER INDEX <name> ATTACH PARTITION <index_name> */
+ ATTACH PARTITION qualified_name
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_AttachPartition;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ n->def = (Node *) cmd;
+
+ $$ = (Node *) n;
+ }
+ ;
+
alter_table_cmd:
/* ALTER TABLE <name> ADD <coldef> */
ADD_P columnDef
@@ -7330,7 +7355,7 @@ defacl_privilege_target:
*****************************************************************************/
IndexStmt: CREATE opt_unique INDEX opt_concurrently opt_index_name
- ON qualified_name access_method_clause '(' index_params ')'
+ ON relation_expr access_method_clause '(' index_params ')'
opt_reloptions OptTableSpace where_clause
{
IndexStmt *n = makeNode(IndexStmt);
@@ -7338,6 +7363,7 @@ IndexStmt: CREATE opt_unique INDEX opt_concurrently opt_index_name
n->concurrent = $4;
n->idxname = $5;
n->relation = $7;
+ n->relationId = InvalidOid;
n->accessMethod = $8;
n->indexParams = $10;
n->options = $12;
@@ -7356,7 +7382,7 @@ IndexStmt: CREATE opt_unique INDEX opt_concurrently opt_index_name
$$ = (Node *)n;
}
| CREATE opt_unique INDEX opt_concurrently IF_P NOT EXISTS index_name
- ON qualified_name access_method_clause '(' index_params ')'
+ ON relation_expr access_method_clause '(' index_params ')'
opt_reloptions OptTableSpace where_clause
{
IndexStmt *n = makeNode(IndexStmt);
@@ -7364,6 +7390,7 @@ IndexStmt: CREATE opt_unique INDEX opt_concurrently opt_index_name
n->concurrent = $4;
n->idxname = $8;
n->relation = $10;
+ n->relationId = InvalidOid;
n->accessMethod = $11;
n->indexParams = $13;
n->options = $15;
diff --git a/src/backend/parser/parse_utilcmd.c b/src/backend/parser/parse_utilcmd.c
index 128f1679c6..90bb356df8 100644
--- a/src/backend/parser/parse_utilcmd.c
+++ b/src/backend/parser/parse_utilcmd.c
@@ -118,9 +118,6 @@ static void transformTableLikeClause(CreateStmtContext *cxt,
TableLikeClause *table_like_clause);
static void transformOfType(CreateStmtContext *cxt,
TypeName *ofTypename);
-static IndexStmt *generateClonedIndexStmt(CreateStmtContext *cxt,
- Relation source_idx,
- const AttrNumber *attmap, int attmap_length);
static List *get_collation(Oid collation, Oid actual_datatype);
static List *get_opclass(Oid opclass, Oid actual_datatype);
static void transformIndexConstraints(CreateStmtContext *cxt);
@@ -1185,7 +1182,8 @@ transformTableLikeClause(CreateStmtContext *cxt, TableLikeClause *table_like_cla
parent_index = index_open(parent_index_oid, AccessShareLock);
/* Build CREATE INDEX statement to recreate the parent_index */
- index_stmt = generateClonedIndexStmt(cxt, parent_index,
+ index_stmt = generateClonedIndexStmt(cxt->relation, InvalidOid,
+ parent_index,
attmap, tupleDesc->natts);
/* Copy comment on index, if requested */
@@ -1263,10 +1261,12 @@ transformOfType(CreateStmtContext *cxt, TypeName *ofTypename)
/*
* Generate an IndexStmt node using information from an already existing index
- * "source_idx". Attribute numbers should be adjusted according to attmap.
+ * "source_idx", for the rel identified either by heapRel or heapRelid.
+ *
+ * Attribute numbers should be adjusted according to attmap.
*/
-static IndexStmt *
-generateClonedIndexStmt(CreateStmtContext *cxt, Relation source_idx,
+IndexStmt *
+generateClonedIndexStmt(RangeVar *heapRel, Oid heapRelid, Relation source_idx,
const AttrNumber *attmap, int attmap_length)
{
Oid source_relid = RelationGetRelid(source_idx);
@@ -1287,6 +1287,9 @@ generateClonedIndexStmt(CreateStmtContext *cxt, Relation source_idx,
Datum datum;
bool isnull;
+ Assert((heapRel == NULL && OidIsValid(heapRelid)) ||
+ (heapRel != NULL && !OidIsValid(heapRelid)));
+
/*
* Fetch pg_class tuple of source index. We can't use the copy in the
* relcache entry because it doesn't include optional fields.
@@ -1322,7 +1325,8 @@ generateClonedIndexStmt(CreateStmtContext *cxt, Relation source_idx,
/* Begin building the IndexStmt */
index = makeNode(IndexStmt);
- index->relation = cxt->relation;
+ index->relation = heapRel;
+ index->relationId = heapRelid;
index->accessMethod = pstrdup(NameStr(amrec->amname));
if (OidIsValid(idxrelrec->reltablespace))
index->tableSpace = get_tablespace_name(idxrelrec->reltablespace);
@@ -3289,18 +3293,39 @@ transformPartitionCmd(CreateStmtContext *cxt, PartitionCmd *cmd)
{
Relation parentRel = cxt->rel;
- /* the table must be partitioned */
- if (parentRel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_OBJECT_DEFINITION),
- errmsg("\"%s\" is not partitioned",
- RelationGetRelationName(parentRel))));
-
- /* transform the partition bound, if any */
- Assert(RelationGetPartitionKey(parentRel) != NULL);
- if (cmd->bound != NULL)
- cxt->partbound = transformPartitionBound(cxt->pstate, parentRel,
- cmd->bound);
+ switch (parentRel->rd_rel->relkind)
+ {
+ case RELKIND_PARTITIONED_TABLE:
+ /* transform the partition bound, if any */
+ Assert(RelationGetPartitionKey(parentRel) != NULL);
+ if (cmd->bound != NULL)
+ cxt->partbound = transformPartitionBound(cxt->pstate, parentRel,
+ cmd->bound);
+ break;
+ case RELKIND_PARTITIONED_INDEX:
+ /* nothing to check */
+ Assert(cmd->bound == NULL);
+ break;
+ case RELKIND_RELATION:
+ /* the table must be partitioned */
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_OBJECT_DEFINITION),
+ errmsg("table \"%s\" is not partitioned",
+ RelationGetRelationName(parentRel))));
+ break;
+ case RELKIND_INDEX:
+ /* the index must be partitioned */
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_OBJECT_DEFINITION),
+ errmsg("index \"%s\" is not partitioned",
+ RelationGetRelationName(parentRel))));
+ break;
+ default:
+ /* parser shouldn't let this case through */
+ elog(ERROR, "\"%s\" is not a partitioned table or index",
+ RelationGetRelationName(parentRel));
+ break;
+ }
}
/*
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index ec98a612ec..1a29dda783 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -23,6 +23,7 @@
#include "access/xlog.h"
#include "catalog/catalog.h"
#include "catalog/namespace.h"
+#include "catalog/pg_inherits_fn.h"
#include "catalog/toasting.h"
#include "commands/alter.h"
#include "commands/async.h"
@@ -1300,6 +1301,7 @@ ProcessUtilitySlow(ParseState *pstate,
IndexStmt *stmt = (IndexStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ List *inheritors = NIL;
if (stmt->concurrent)
PreventTransactionChain(isTopLevel,
@@ -1321,6 +1323,9 @@ ProcessUtilitySlow(ParseState *pstate,
false, false,
RangeVarCallbackOwnsRelation,
NULL);
+ /* Also, lock any descendant tables if recursive */
+ if (stmt->relation->inh)
+ inheritors = find_all_inheritors(relid, lockmode, NULL);
/* Run parse analysis ... */
stmt = transformIndexStmt(relid, stmt, queryString);
@@ -1331,6 +1336,7 @@ ProcessUtilitySlow(ParseState *pstate,
DefineIndex(relid, /* OID of heap relation */
stmt,
InvalidOid, /* no predefined OID */
+ InvalidOid, /* no parent index */
false, /* is_alter_table */
true, /* check_rights */
true, /* check_not_in_use */
@@ -1346,6 +1352,8 @@ ProcessUtilitySlow(ParseState *pstate,
parsetree);
commandCollected = true;
EventTriggerAlterTableEnd();
+
+ list_free(inheritors);
}
break;
diff --git a/src/backend/utils/adt/amutils.c b/src/backend/utils/adt/amutils.c
index a6d8feea5b..0f7ceb62eb 100644
--- a/src/backend/utils/adt/amutils.c
+++ b/src/backend/utils/adt/amutils.c
@@ -183,7 +183,8 @@ indexam_property(FunctionCallInfo fcinfo,
if (!HeapTupleIsValid(tuple))
PG_RETURN_NULL();
rd_rel = (Form_pg_class) GETSTRUCT(tuple);
- if (rd_rel->relkind != RELKIND_INDEX)
+ if (rd_rel->relkind != RELKIND_INDEX &&
+ rd_rel->relkind != RELKIND_PARTITIONED_INDEX)
{
ReleaseSysCache(tuple);
PG_RETURN_NULL();
diff --git a/src/backend/utils/adt/ruleutils.c b/src/backend/utils/adt/ruleutils.c
index 9cdbb06add..c5f5a1ca3f 100644
--- a/src/backend/utils/adt/ruleutils.c
+++ b/src/backend/utils/adt/ruleutils.c
@@ -317,7 +317,7 @@ static void decompile_column_index_array(Datum column_index_array, Oid relId,
static char *pg_get_ruledef_worker(Oid ruleoid, int prettyFlags);
static char *pg_get_indexdef_worker(Oid indexrelid, int colno,
const Oid *excludeOps,
- bool attrsOnly, bool showTblSpc,
+ bool attrsOnly, bool showTblSpc, bool inherits,
int prettyFlags, bool missing_ok);
static char *pg_get_statisticsobj_worker(Oid statextid, bool missing_ok);
static char *pg_get_partkeydef_worker(Oid relid, int prettyFlags,
@@ -1086,7 +1086,7 @@ pg_get_indexdef(PG_FUNCTION_ARGS)
prettyFlags = PRETTYFLAG_INDENT;
- res = pg_get_indexdef_worker(indexrelid, 0, NULL, false, false,
+ res = pg_get_indexdef_worker(indexrelid, 0, NULL, false, false, false,
prettyFlags, true);
if (res == NULL)
@@ -1107,7 +1107,7 @@ pg_get_indexdef_ext(PG_FUNCTION_ARGS)
prettyFlags = pretty ? PRETTYFLAG_PAREN | PRETTYFLAG_INDENT : PRETTYFLAG_INDENT;
res = pg_get_indexdef_worker(indexrelid, colno, NULL, colno != 0, false,
- prettyFlags, true);
+ false, prettyFlags, true);
if (res == NULL)
PG_RETURN_NULL();
@@ -1123,7 +1123,7 @@ pg_get_indexdef_ext(PG_FUNCTION_ARGS)
char *
pg_get_indexdef_string(Oid indexrelid)
{
- return pg_get_indexdef_worker(indexrelid, 0, NULL, false, true, 0, false);
+ return pg_get_indexdef_worker(indexrelid, 0, NULL, false, true, true, 0, false);
}
/* Internal version that just reports the column definitions */
@@ -1133,7 +1133,7 @@ pg_get_indexdef_columns(Oid indexrelid, bool pretty)
int prettyFlags;
prettyFlags = pretty ? PRETTYFLAG_PAREN | PRETTYFLAG_INDENT : PRETTYFLAG_INDENT;
- return pg_get_indexdef_worker(indexrelid, 0, NULL, true, false,
+ return pg_get_indexdef_worker(indexrelid, 0, NULL, true, false, false,
prettyFlags, false);
}
@@ -1146,7 +1146,7 @@ pg_get_indexdef_columns(Oid indexrelid, bool pretty)
static char *
pg_get_indexdef_worker(Oid indexrelid, int colno,
const Oid *excludeOps,
- bool attrsOnly, bool showTblSpc,
+ bool attrsOnly, bool showTblSpc, bool inherits,
int prettyFlags, bool missing_ok)
{
/* might want a separate isConstraint parameter later */
@@ -1259,9 +1259,11 @@ pg_get_indexdef_worker(Oid indexrelid, int colno,
if (!attrsOnly)
{
if (!isConstraint)
- appendStringInfo(&buf, "CREATE %sINDEX %s ON %s USING %s (",
+ appendStringInfo(&buf, "CREATE %sINDEX %s ON %s%s USING %s (",
idxrec->indisunique ? "UNIQUE " : "",
quote_identifier(NameStr(idxrelrec->relname)),
+ idxrelrec->relkind == RELKIND_PARTITIONED_INDEX
+ && !inherits ? "ONLY " : "",
generate_relation_name(indrelid, NIL),
quote_identifier(NameStr(amrec->amname)));
else /* currently, must be EXCLUDE constraint */
@@ -2148,6 +2150,7 @@ pg_get_constraintdef_worker(Oid constraintId, bool fullCommand,
operators,
false,
false,
+ false,
prettyFlags,
false));
break;
diff --git a/src/backend/utils/cache/relcache.c b/src/backend/utils/cache/relcache.c
index 28a4483434..3b4ed7ddd9 100644
--- a/src/backend/utils/cache/relcache.c
+++ b/src/backend/utils/cache/relcache.c
@@ -430,18 +430,26 @@ static void
RelationParseRelOptions(Relation relation, HeapTuple tuple)
{
bytea *options;
+ amoptions_function amoptsfn;
relation->rd_options = NULL;
- /* Fall out if relkind should not have options */
+ /*
+ * Look up any AM-specific parse function; fall out if relkind should not
+ * have options.
+ */
switch (relation->rd_rel->relkind)
{
case RELKIND_RELATION:
case RELKIND_TOASTVALUE:
- case RELKIND_INDEX:
case RELKIND_VIEW:
case RELKIND_MATVIEW:
case RELKIND_PARTITIONED_TABLE:
+ amoptsfn = NULL;
+ break;
+ case RELKIND_INDEX:
+ case RELKIND_PARTITIONED_INDEX:
+ amoptsfn = relation->rd_amroutine->amoptions;
break;
default:
return;
@@ -452,10 +460,7 @@ RelationParseRelOptions(Relation relation, HeapTuple tuple)
* we might not have any other for pg_class yet (consider executing this
* code for pg_class itself)
*/
- options = extractRelOptions(tuple,
- GetPgClassDescriptor(),
- relation->rd_rel->relkind == RELKIND_INDEX ?
- relation->rd_amroutine->amoptions : NULL);
+ options = extractRelOptions(tuple, GetPgClassDescriptor(), amoptsfn);
/*
* Copy parsed data into CacheMemoryContext. To guard against the
@@ -2053,7 +2058,8 @@ RelationIdGetRelation(Oid relationId)
* and we don't want to use the full-blown procedure because it's
* a headache for indexes that reload itself depends on.
*/
- if (rd->rd_rel->relkind == RELKIND_INDEX)
+ if (rd->rd_rel->relkind == RELKIND_INDEX ||
+ rd->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
RelationReloadIndexInfo(rd);
else
RelationClearRelation(rd, true);
@@ -2167,7 +2173,8 @@ RelationReloadIndexInfo(Relation relation)
Form_pg_class relp;
/* Should be called only for invalidated indexes */
- Assert(relation->rd_rel->relkind == RELKIND_INDEX &&
+ Assert((relation->rd_rel->relkind == RELKIND_INDEX ||
+ relation->rd_rel->relkind == RELKIND_PARTITIONED_INDEX) &&
!relation->rd_isvalid);
/* Ensure it's closed at smgr level */
@@ -2387,7 +2394,8 @@ RelationClearRelation(Relation relation, bool rebuild)
{
RelationInitPhysicalAddr(relation);
- if (relation->rd_rel->relkind == RELKIND_INDEX)
+ if (relation->rd_rel->relkind == RELKIND_INDEX ||
+ relation->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
{
relation->rd_isvalid = false; /* needs to be revalidated */
if (relation->rd_refcnt > 1 && IsTransactionState())
@@ -2403,7 +2411,8 @@ RelationClearRelation(Relation relation, bool rebuild)
* re-read the pg_class row to handle possible physical relocation of the
* index, and we check for pg_index updates too.
*/
- if (relation->rd_rel->relkind == RELKIND_INDEX &&
+ if ((relation->rd_rel->relkind == RELKIND_INDEX ||
+ relation->rd_rel->relkind == RELKIND_PARTITIONED_INDEX) &&
relation->rd_refcnt > 0 &&
relation->rd_indexcxt != NULL)
{
@@ -5461,7 +5470,10 @@ load_relcache_init_file(bool shared)
rel->rd_att->constr = constr;
}
- /* If it's an index, there's more to do */
+ /*
+ * If it's an index, there's more to do. Note we explicitly ignore
+ * partitioned indexes here.
+ */
if (rel->rd_rel->relkind == RELKIND_INDEX)
{
MemoryContext indexcxt;
@@ -5825,7 +5837,10 @@ write_relcache_init_file(bool shared)
(rel->rd_options ? VARSIZE(rel->rd_options) : 0),
fp);
- /* If it's an index, there's more to do */
+ /*
+ * If it's an index, there's more to do. Note we explicitly ignore
+ * partitioned indexes here.
+ */
if (rel->rd_rel->relkind == RELKIND_INDEX)
{
/* write the pg_index tuple */
diff --git a/src/bin/pg_dump/common.c b/src/bin/pg_dump/common.c
index 7f5f351486..9483053680 100644
--- a/src/bin/pg_dump/common.c
+++ b/src/bin/pg_dump/common.c
@@ -68,6 +68,7 @@ static int numextmembers;
static void flagInhTables(Archive *fout, TableInfo *tbinfo, int numTables,
InhInfo *inhinfo, int numInherits);
+static void flagInhIndexes(Archive *fout, TableInfo *tblinfo, int numTables);
static void flagInhAttrs(DumpOptions *dopt, TableInfo *tblinfo, int numTables);
static DumpableObject **buildIndexArray(void *objArray, int numObjs,
Size objSize);
@@ -76,6 +77,8 @@ static int ExtensionMemberIdCompare(const void *p1, const void *p2);
static void findParentsByOid(TableInfo *self,
InhInfo *inhinfo, int numInherits);
static int strInArray(const char *pattern, char **arr, int arr_size);
+static IndxInfo *findIndexByOid(Oid oid, DumpableObject **idxinfoindex,
+ int numIndexes);
/*
@@ -258,6 +261,10 @@ getSchemaData(Archive *fout, int *numTablesPtr)
getIndexes(fout, tblinfo, numTables);
if (g_verbose)
+ write_msg(NULL, "flagging indexes in partitioned tables\n");
+ flagInhIndexes(fout, tblinfo, numTables);
+
+ if (g_verbose)
write_msg(NULL, "reading extended statistics\n");
getExtendedStatistics(fout, tblinfo, numTables);
@@ -354,6 +361,89 @@ flagInhTables(Archive *fout, TableInfo *tblinfo, int numTables,
}
}
+/*
+ * flagInhIndexes -
+ * Create AttachIndexInfo objects for partitioned indexes, and add
+ * appropriate dependency links.
+ */
+static void
+flagInhIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
+{
+ int i,
+ j,
+ k;
+ DumpableObject ***parentIndexArray;
+
+ parentIndexArray = (DumpableObject ***)
+ pg_malloc0(getMaxDumpId() * sizeof(DumpableObject **));
+
+ for (i = 0; i < numTables; i++)
+ {
+ TableInfo *parenttbl;
+ IndexAttachInfo *attachinfo;
+
+ if (!tblinfo[i].ispartition || tblinfo[i].numParents == 0)
+ continue;
+
+ Assert(tblinfo[i].numParents == 1);
+ parenttbl = tblinfo[i].parents[0];
+
+ /*
+ * We need access to each parent table's index list, but there is no
+ * index to cover them outside of this function. To avoid having to
+ * sort every parent table's indexes each time we come across each of
+ * its partitions, create an indexed array for each parent the first
+ * time it is required.
+ */
+ if (parentIndexArray[parenttbl->dobj.dumpId] == NULL)
+ parentIndexArray[parenttbl->dobj.dumpId] =
+ buildIndexArray(parenttbl->indexes,
+ parenttbl->numIndexes,
+ sizeof(IndxInfo));
+
+ attachinfo = (IndexAttachInfo *)
+ pg_malloc0(tblinfo[i].numIndexes * sizeof(IndexAttachInfo));
+ for (j = 0, k = 0; j < tblinfo[i].numIndexes; j++)
+ {
+ IndxInfo *index = &(tblinfo[i].indexes[j]);
+ IndxInfo *parentidx;
+
+ if (index->parentidx == 0)
+ continue;
+
+ parentidx = findIndexByOid(index->parentidx,
+ parentIndexArray[parenttbl->dobj.dumpId],
+ parenttbl->numIndexes);
+ if (parentidx == NULL)
+ continue;
+
+ attachinfo[k].dobj.objType = DO_INDEX_ATTACH;
+ attachinfo[k].dobj.catId.tableoid = 0;
+ attachinfo[k].dobj.catId.oid = 0;
+ AssignDumpId(&attachinfo[k].dobj);
+ attachinfo[k].dobj.name = pg_strdup(index->dobj.name);
+ attachinfo[k].parentIdx = parentidx;
+ attachinfo[k].partitionIdx = index;
+
+ /*
+ * We want dependencies from parent to partition (so that the
+ * partition index is created first), and another one from
+ * attach object to parent (so that the partition index is
+ * attached once the parent index has been created).
+ */
+ addObjectDependency(&parentidx->dobj, index->dobj.dumpId);
+ addObjectDependency(&attachinfo[k].dobj, parentidx->dobj.dumpId);
+
+ k++;
+ }
+ }
+
+ for (i = 0; i < numTables; i++)
+ if (parentIndexArray[i])
+ pg_free(parentIndexArray[i]);
+ pg_free(parentIndexArray);
+}
+
/* flagInhAttrs -
* for each dumpable table in tblinfo, flag its inherited attributes
*
@@ -827,6 +917,18 @@ findExtensionByOid(Oid oid)
return (ExtensionInfo *) findObjectByOid(oid, extinfoindex, numExtensions);
}
+/*
+ * findIndexByOid
+ * find the entry of the index with the given oid
+ *
+ * This one's signature is different from the previous ones because we lack a
+ * global array of all indexes, so caller must pass their array as argument.
+ */
+static IndxInfo *
+findIndexByOid(Oid oid, DumpableObject **idxinfoindex, int numIndexes)
+{
+ return (IndxInfo *) findObjectByOid(oid, idxinfoindex, numIndexes);
+}
/*
* setExtensionMembership
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index 27628a397c..92b29e2e5f 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -193,6 +193,7 @@ static void dumpAttrDef(Archive *fout, AttrDefInfo *adinfo);
static void dumpSequence(Archive *fout, TableInfo *tbinfo);
static void dumpSequenceData(Archive *fout, TableDataInfo *tdinfo);
static void dumpIndex(Archive *fout, IndxInfo *indxinfo);
+static void dumpIndexAttach(Archive *fout, IndexAttachInfo *attachinfo);
static void dumpStatisticsExt(Archive *fout, StatsExtInfo *statsextinfo);
static void dumpConstraint(Archive *fout, ConstraintInfo *coninfo);
static void dumpTableConstraintComment(Archive *fout, ConstraintInfo *coninfo);
@@ -6509,6 +6510,7 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
int i_tableoid,
i_oid,
i_indexname,
+ i_parentidx,
i_indexdef,
i_indnkeys,
i_indkey,
@@ -6530,10 +6532,6 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
{
TableInfo *tbinfo = &tblinfo[i];
- /* Only plain tables and materialized views have indexes. */
- if (tbinfo->relkind != RELKIND_RELATION &&
- tbinfo->relkind != RELKIND_MATVIEW)
- continue;
if (!tbinfo->hasindex)
continue;
@@ -6561,7 +6559,39 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
* is not.
*/
resetPQExpBuffer(query);
- if (fout->remoteVersion >= 90400)
+ if (fout->remoteVersion >= 11000)
+ {
+ appendPQExpBuffer(query,
+ "SELECT t.tableoid, t.oid, "
+ "t.relname AS indexname, "
+ "inh.inhparent AS parentidx, "
+ "pg_catalog.pg_get_indexdef(i.indexrelid) AS indexdef, "
+ "t.relnatts AS indnkeys, "
+ "i.indkey, i.indisclustered, "
+ "i.indisreplident, t.relpages, "
+ "c.contype, c.conname, "
+ "c.condeferrable, c.condeferred, "
+ "c.tableoid AS contableoid, "
+ "c.oid AS conoid, "
+ "pg_catalog.pg_get_constraintdef(c.oid, false) AS condef, "
+ "(SELECT spcname FROM pg_catalog.pg_tablespace s WHERE s.oid = t.reltablespace) AS tablespace, "
+ "t.reloptions AS indreloptions "
+ "FROM pg_catalog.pg_index i "
+ "JOIN pg_catalog.pg_class t ON (t.oid = i.indexrelid) "
+ "JOIN pg_catalog.pg_class t2 ON (t2.oid = i.indrelid) "
+ "LEFT JOIN pg_catalog.pg_constraint c "
+ "ON (i.indrelid = c.conrelid AND "
+ "i.indexrelid = c.conindid AND "
+ "c.contype IN ('p','u','x')) "
+ "LEFT JOIN pg_catalog.pg_inherits inh "
+ "ON (inh.inhrelid = indexrelid) "
+ "WHERE i.indrelid = '%u'::pg_catalog.oid "
+ "AND (i.indisvalid OR t2.relkind = 'p') "
+ "AND i.indisready "
+ "ORDER BY indexname",
+ tbinfo->dobj.catId.oid);
+ }
+ else if (fout->remoteVersion >= 90400)
{
/*
* the test on indisready is necessary in 9.2, and harmless in
@@ -6570,6 +6600,7 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
appendPQExpBuffer(query,
"SELECT t.tableoid, t.oid, "
"t.relname AS indexname, "
+ "0 AS parentidx, "
"pg_catalog.pg_get_indexdef(i.indexrelid) AS indexdef, "
"t.relnatts AS indnkeys, "
"i.indkey, i.indisclustered, "
@@ -6601,6 +6632,7 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
appendPQExpBuffer(query,
"SELECT t.tableoid, t.oid, "
"t.relname AS indexname, "
+ "0 AS parentidx, "
"pg_catalog.pg_get_indexdef(i.indexrelid) AS indexdef, "
"t.relnatts AS indnkeys, "
"i.indkey, i.indisclustered, "
@@ -6628,6 +6660,7 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
appendPQExpBuffer(query,
"SELECT t.tableoid, t.oid, "
"t.relname AS indexname, "
+ "0 AS parentidx, "
"pg_catalog.pg_get_indexdef(i.indexrelid) AS indexdef, "
"t.relnatts AS indnkeys, "
"i.indkey, i.indisclustered, "
@@ -6658,6 +6691,7 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
appendPQExpBuffer(query,
"SELECT t.tableoid, t.oid, "
"t.relname AS indexname, "
+ "0 AS parentidx, "
"pg_catalog.pg_get_indexdef(i.indexrelid) AS indexdef, "
"t.relnatts AS indnkeys, "
"i.indkey, i.indisclustered, "
@@ -6690,6 +6724,7 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
i_tableoid = PQfnumber(res, "tableoid");
i_oid = PQfnumber(res, "oid");
i_indexname = PQfnumber(res, "indexname");
+ i_parentidx = PQfnumber(res, "parentidx");
i_indexdef = PQfnumber(res, "indexdef");
i_indnkeys = PQfnumber(res, "indnkeys");
i_indkey = PQfnumber(res, "indkey");
@@ -6706,8 +6741,10 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
i_tablespace = PQfnumber(res, "tablespace");
i_indreloptions = PQfnumber(res, "indreloptions");
- indxinfo = (IndxInfo *) pg_malloc(ntups * sizeof(IndxInfo));
+ tbinfo->indexes = indxinfo =
+ (IndxInfo *) pg_malloc(ntups * sizeof(IndxInfo));
constrinfo = (ConstraintInfo *) pg_malloc(ntups * sizeof(ConstraintInfo));
+ tbinfo->numIndexes = ntups;
for (j = 0; j < ntups; j++)
{
@@ -6729,6 +6766,7 @@ getIndexes(Archive *fout, TableInfo tblinfo[], int numTables)
indxinfo[j].indkeys, indxinfo[j].indnkeys);
indxinfo[j].indisclustered = (PQgetvalue(res, j, i_indisclustered)[0] == 't');
indxinfo[j].indisreplident = (PQgetvalue(res, j, i_indisreplident)[0] == 't');
+ indxinfo[j].parentidx = atooid(PQgetvalue(res, j, i_parentidx));
indxinfo[j].relpages = atoi(PQgetvalue(res, j, i_relpages));
contype = *(PQgetvalue(res, j, i_contype));
@@ -9512,6 +9550,9 @@ dumpDumpableObject(Archive *fout, DumpableObject *dobj)
case DO_INDEX:
dumpIndex(fout, (IndxInfo *) dobj);
break;
+ case DO_INDEX_ATTACH:
+ dumpIndexAttach(fout, (IndexAttachInfo *) dobj);
+ break;
case DO_STATSEXT:
dumpStatisticsExt(fout, (StatsExtInfo *) dobj);
break;
@@ -16173,6 +16214,42 @@ dumpIndex(Archive *fout, IndxInfo *indxinfo)
}
/*
+ * dumpIndexAttach
+ * write out to fout a partitioned-index attachment clause
+ */
+void
+dumpIndexAttach(Archive *fout, IndexAttachInfo *attachinfo)
+{
+ if (fout->dopt->dataOnly)
+ return;
+
+ if (attachinfo->partitionIdx->dobj.dump & DUMP_COMPONENT_DEFINITION)
+ {
+ PQExpBuffer q = createPQExpBuffer();
+
+ appendPQExpBuffer(q, "\nALTER INDEX %s ",
+ fmtQualifiedId(fout->remoteVersion,
+ attachinfo->parentIdx->dobj.namespace->dobj.name,
+ attachinfo->parentIdx->dobj.name));
+ appendPQExpBuffer(q, "ATTACH PARTITION %s;\n",
+ fmtQualifiedId(fout->remoteVersion,
+ attachinfo->partitionIdx->dobj.namespace->dobj.name,
+ attachinfo->partitionIdx->dobj.name));
+
+ ArchiveEntry(fout, attachinfo->dobj.catId, attachinfo->dobj.dumpId,
+ attachinfo->dobj.name,
+ NULL, NULL,
+ "",
+ false, "INDEX ATTACH", SECTION_POST_DATA,
+ q->data, "", NULL,
+ NULL, 0,
+ NULL, NULL);
+
+ destroyPQExpBuffer(q);
+ }
+}
+
+/*
* dumpStatisticsExt
* write out to fout an extended statistics object
*/
@@ -17803,6 +17880,7 @@ addBoundaryDependencies(DumpableObject **dobjs, int numObjs,
addObjectDependency(postDataBound, dobj->dumpId);
break;
case DO_INDEX:
+ case DO_INDEX_ATTACH:
case DO_STATSEXT:
case DO_REFRESH_MATVIEW:
case DO_TRIGGER:
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index 49a02b4fa8..6c18d451ef 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -56,6 +56,7 @@ typedef enum
DO_TABLE,
DO_ATTRDEF,
DO_INDEX,
+ DO_INDEX_ATTACH,
DO_STATSEXT,
DO_RULE,
DO_TRIGGER,
@@ -328,6 +329,8 @@ typedef struct _tableInfo
*/
int numParents; /* number of (immediate) parent tables */
struct _tableInfo **parents; /* TableInfos of immediate parents */
+ int numIndexes; /* number of indexes */
+ struct _indxInfo *indexes; /* indexes */
struct _tableDataInfo *dataObj; /* TableDataInfo, if dumping its data */
int numTriggers; /* number of triggers for table */
struct _triggerInfo *triggers; /* array of TriggerInfo structs */
@@ -361,11 +364,19 @@ typedef struct _indxInfo
Oid *indkeys;
bool indisclustered;
bool indisreplident;
+ Oid parentidx; /* if partitioned, parent index OID */
/* if there is an associated constraint object, its dumpId: */
DumpId indexconstraint;
int relpages; /* relpages of the underlying table */
} IndxInfo;
+typedef struct _indexAttachInfo
+{
+ DumpableObject dobj;
+ IndxInfo *parentIdx; /* link to index on partitioned table */
+ IndxInfo *partitionIdx; /* link to index on partition */
+} IndexAttachInfo;
+
typedef struct _statsExtInfo
{
DumpableObject dobj;
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index 6da1c35a42..5ce3c5d485 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -35,6 +35,10 @@ static const char *modulename = gettext_noop("sorter");
* pg_dump.c; that is, PRE_DATA objects must sort before DO_PRE_DATA_BOUNDARY,
* POST_DATA objects must sort after DO_POST_DATA_BOUNDARY, and DATA objects
* must sort between them.
+ *
+ * Note: sortDataAndIndexObjectsBySize wants to have all DO_TABLE_DATA and
+ * DO_INDEX objects in contiguous chunks, so do not reuse the values for those
+ * for other object types.
*/
static const int dbObjectTypePriority[] =
{
@@ -53,11 +57,12 @@ static const int dbObjectTypePriority[] =
18, /* DO_TABLE */
20, /* DO_ATTRDEF */
28, /* DO_INDEX */
- 29, /* DO_STATSEXT */
- 30, /* DO_RULE */
- 31, /* DO_TRIGGER */
+ 29, /* DO_INDEX_ATTACH */
+ 30, /* DO_STATSEXT */
+ 31, /* DO_RULE */
+ 32, /* DO_TRIGGER */
27, /* DO_CONSTRAINT */
- 32, /* DO_FK_CONSTRAINT */
+ 33, /* DO_FK_CONSTRAINT */
2, /* DO_PROCLANG */
10, /* DO_CAST */
23, /* DO_TABLE_DATA */
@@ -69,18 +74,18 @@ static const int dbObjectTypePriority[] =
15, /* DO_TSCONFIG */
16, /* DO_FDW */
17, /* DO_FOREIGN_SERVER */
- 32, /* DO_DEFAULT_ACL */
+ 33, /* DO_DEFAULT_ACL */
3, /* DO_TRANSFORM */
21, /* DO_BLOB */
25, /* DO_BLOB_DATA */
22, /* DO_PRE_DATA_BOUNDARY */
26, /* DO_POST_DATA_BOUNDARY */
- 33, /* DO_EVENT_TRIGGER */
- 38, /* DO_REFRESH_MATVIEW */
- 34, /* DO_POLICY */
- 35, /* DO_PUBLICATION */
- 36, /* DO_PUBLICATION_REL */
- 37 /* DO_SUBSCRIPTION */
+ 34, /* DO_EVENT_TRIGGER */
+ 39, /* DO_REFRESH_MATVIEW */
+ 35, /* DO_POLICY */
+ 36, /* DO_PUBLICATION */
+ 37, /* DO_PUBLICATION_REL */
+ 38 /* DO_SUBSCRIPTION */
};
static DumpId preDataBoundId;
@@ -937,6 +942,13 @@ repairDomainConstraintMultiLoop(DumpableObject *domainobj,
addObjectDependency(constraintobj, postDataBoundId);
}
+static void
+repairIndexLoop(DumpableObject *partedindex,
+ DumpableObject *partindex)
+{
+ removeObjectDependency(partedindex, partindex->dumpId);
+}
+
/*
* Fix a dependency loop, or die trying ...
*
@@ -1099,6 +1111,23 @@ repairDependencyLoop(DumpableObject **loop,
return;
}
+ /* index on partitioned table and corresponding index on partition */
+ if (nLoop == 2 &&
+ loop[0]->objType == DO_INDEX &&
+ loop[1]->objType == DO_INDEX)
+ {
+ if (((IndxInfo *) loop[0])->parentidx == loop[1]->catId.oid)
+ {
+ repairIndexLoop(loop[0], loop[1]);
+ return;
+ }
+ else if (((IndxInfo *) loop[1])->parentidx == loop[0]->catId.oid)
+ {
+ repairIndexLoop(loop[1], loop[0]);
+ return;
+ }
+ }
+
/* Indirect loop involving table and attribute default */
if (nLoop > 2)
{
@@ -1292,6 +1321,11 @@ describeDumpableObject(DumpableObject *obj, char *buf, int bufsize)
"INDEX %s (ID %d OID %u)",
obj->name, obj->dumpId, obj->catId.oid);
return;
+ case DO_INDEX_ATTACH:
+ snprintf(buf, bufsize,
+ "INDEX ATTACH %s (ID %d)",
+ obj->name, obj->dumpId);
+ return;
case DO_STATSEXT:
snprintf(buf, bufsize,
"STATISTICS %s (ID %d OID %u)",
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index f2e62946d8..7a0c2423ac 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -1705,7 +1705,8 @@ describeOneTableDetails(const char *schemaname,
appendPQExpBufferStr(&buf, ",\n a.attidentity");
else
appendPQExpBufferStr(&buf, ",\n ''::pg_catalog.char AS attidentity");
- if (tableinfo.relkind == RELKIND_INDEX)
+ if (tableinfo.relkind == RELKIND_INDEX ||
+ tableinfo.relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ",\n pg_catalog.pg_get_indexdef(a.attrelid, a.attnum, TRUE) AS indexdef");
else
appendPQExpBufferStr(&buf, ",\n NULL AS indexdef");
@@ -1766,6 +1767,7 @@ describeOneTableDetails(const char *schemaname,
schemaname, relationname);
break;
case RELKIND_INDEX:
+ case RELKIND_PARTITIONED_INDEX:
if (tableinfo.relpersistence == 'u')
printfPQExpBuffer(&title, _("Unlogged index \"%s.%s\""),
schemaname, relationname);
@@ -1823,7 +1825,8 @@ describeOneTableDetails(const char *schemaname,
show_column_details = true;
}
- if (tableinfo.relkind == RELKIND_INDEX)
+ if (tableinfo.relkind == RELKIND_INDEX ||
+ tableinfo.relkind == RELKIND_PARTITIONED_INDEX)
headers[cols++] = gettext_noop("Definition");
if (tableinfo.relkind == RELKIND_FOREIGN_TABLE && pset.sversion >= 90200)
@@ -1834,6 +1837,7 @@ describeOneTableDetails(const char *schemaname,
headers[cols++] = gettext_noop("Storage");
if (tableinfo.relkind == RELKIND_RELATION ||
tableinfo.relkind == RELKIND_INDEX ||
+ tableinfo.relkind == RELKIND_PARTITIONED_INDEX ||
tableinfo.relkind == RELKIND_MATVIEW ||
tableinfo.relkind == RELKIND_FOREIGN_TABLE ||
tableinfo.relkind == RELKIND_PARTITIONED_TABLE)
@@ -1906,7 +1910,8 @@ describeOneTableDetails(const char *schemaname,
}
/* Expression for index column */
- if (tableinfo.relkind == RELKIND_INDEX)
+ if (tableinfo.relkind == RELKIND_INDEX ||
+ tableinfo.relkind == RELKIND_PARTITIONED_INDEX)
printTableAddCell(&cont, PQgetvalue(res, i, 7), false, false);
/* FDW options for foreign table column, only for 9.2 or later */
@@ -1930,6 +1935,7 @@ describeOneTableDetails(const char *schemaname,
/* Statistics target, if the relkind supports this feature */
if (tableinfo.relkind == RELKIND_RELATION ||
tableinfo.relkind == RELKIND_INDEX ||
+ tableinfo.relkind == RELKIND_PARTITIONED_INDEX ||
tableinfo.relkind == RELKIND_MATVIEW ||
tableinfo.relkind == RELKIND_FOREIGN_TABLE ||
tableinfo.relkind == RELKIND_PARTITIONED_TABLE)
@@ -2021,7 +2027,8 @@ describeOneTableDetails(const char *schemaname,
PQclear(result);
}
- if (tableinfo.relkind == RELKIND_INDEX)
+ if (tableinfo.relkind == RELKIND_INDEX ||
+ tableinfo.relkind == RELKIND_PARTITIONED_INDEX)
{
/* Footer information about an index */
PGresult *result;
@@ -3397,6 +3404,7 @@ listTables(const char *tabtypes, const char *pattern, bool verbose, bool showSys
" WHEN 's' THEN '%s'"
" WHEN " CppAsString2(RELKIND_FOREIGN_TABLE) " THEN '%s'"
" WHEN " CppAsString2(RELKIND_PARTITIONED_TABLE) " THEN '%s'"
+ " WHEN " CppAsString2(RELKIND_PARTITIONED_INDEX) " THEN '%s'"
" END as \"%s\",\n"
" pg_catalog.pg_get_userbyid(c.relowner) as \"%s\"",
gettext_noop("Schema"),
@@ -3409,6 +3417,7 @@ listTables(const char *tabtypes, const char *pattern, bool verbose, bool showSys
gettext_noop("special"),
gettext_noop("foreign table"),
gettext_noop("table"), /* partitioned table */
+ gettext_noop("index"), /* partitioned index */
gettext_noop("Type"),
gettext_noop("Owner"));
@@ -3454,7 +3463,8 @@ listTables(const char *tabtypes, const char *pattern, bool verbose, bool showSys
if (showMatViews)
appendPQExpBufferStr(&buf, CppAsString2(RELKIND_MATVIEW) ",");
if (showIndexes)
- appendPQExpBufferStr(&buf, CppAsString2(RELKIND_INDEX) ",");
+ appendPQExpBufferStr(&buf, CppAsString2(RELKIND_INDEX) ","
+ CppAsString2(RELKIND_PARTITIONED_INDEX) ",");
if (showSeq)
appendPQExpBufferStr(&buf, CppAsString2(RELKIND_SEQUENCE) ",");
if (showSystem || pattern)
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index b51098deca..8bc4a194a5 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -412,7 +412,8 @@ static const SchemaQuery Query_for_list_of_indexes = {
/* catname */
"pg_catalog.pg_class c",
/* selcondition */
- "c.relkind IN (" CppAsString2(RELKIND_INDEX) ")",
+ "c.relkind IN (" CppAsString2(RELKIND_INDEX) ", "
+ CppAsString2(RELKIND_PARTITIONED_INDEX) ")",
/* viscondition */
"pg_catalog.pg_table_is_visible(c.oid)",
/* namespace */
@@ -600,6 +601,23 @@ static const SchemaQuery Query_for_list_of_tmf = {
NULL
};
+static const SchemaQuery Query_for_list_of_tpm = {
+ /* catname */
+ "pg_catalog.pg_class c",
+ /* selcondition */
+ "c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
+ CppAsString2(RELKIND_MATVIEW) ")",
+ /* viscondition */
+ "pg_catalog.pg_table_is_visible(c.oid)",
+ /* namespace */
+ "c.relnamespace",
+ /* result */
+ "pg_catalog.quote_ident(c.relname)",
+ /* qualresult */
+ NULL
+};
+
static const SchemaQuery Query_for_list_of_tm = {
/* catname */
"pg_catalog.pg_class c",
@@ -1676,7 +1694,12 @@ psql_completion(const char *text, int start, int end)
"UNION SELECT 'ALL IN TABLESPACE'");
/* ALTER INDEX <name> */
else if (Matches3("ALTER", "INDEX", MatchAny))
- COMPLETE_WITH_LIST5("ALTER COLUMN", "OWNER TO", "RENAME TO", "SET", "RESET");
+ COMPLETE_WITH_LIST6("ALTER COLUMN", "OWNER TO", "RENAME TO", "SET",
+ "RESET", "ATTACH PARTITION");
+ else if (Matches4("ALTER", "INDEX", MatchAny, "ATTACH"))
+ COMPLETE_WITH_CONST("PARTITION");
+ else if (Matches5("ALTER", "INDEX", MatchAny, "ATTACH", "PARTITION"))
+ COMPLETE_WITH_SCHEMA_QUERY(Query_for_list_of_indexes, NULL);
/* ALTER INDEX <name> ALTER COLUMN <colnum> */
else if (Matches6("ALTER", "INDEX", MatchAny, "ALTER", "COLUMN", MatchAny))
COMPLETE_WITH_CONST("SET STATISTICS");
@@ -2338,10 +2361,13 @@ psql_completion(const char *text, int start, int end)
COMPLETE_WITH_SCHEMA_QUERY(Query_for_list_of_indexes,
" UNION SELECT 'ON'"
" UNION SELECT 'CONCURRENTLY'");
- /* Complete ... INDEX|CONCURRENTLY [<name>] ON with a list of tables */
+ /*
+ * Complete ... INDEX|CONCURRENTLY [<name>] ON with a list of relations
+ * that can indexes can be created on
+ */
else if (TailMatches3("INDEX|CONCURRENTLY", MatchAny, "ON") ||
TailMatches2("INDEX|CONCURRENTLY", "ON"))
- COMPLETE_WITH_SCHEMA_QUERY(Query_for_list_of_tm, NULL);
+ COMPLETE_WITH_SCHEMA_QUERY(Query_for_list_of_tpm, NULL);
/*
* Complete CREATE|UNIQUE INDEX CONCURRENTLY with "ON" and existing
diff --git a/src/include/catalog/dependency.h b/src/include/catalog/dependency.h
index 6f290d5c6f..46c271a46c 100644
--- a/src/include/catalog/dependency.h
+++ b/src/include/catalog/dependency.h
@@ -49,6 +49,20 @@
* Example: a trigger that's created to enforce a foreign-key constraint
* is made internally dependent on the constraint's pg_constraint entry.
*
+ * DEPENDENCY_INTERNAL_AUTO ('I'): the dependent object was created as
+ * part of creation of the referenced object, and is really just a part
+ * of its internal implementation. A DROP of the dependent object will
+ * be disallowed outright (we'll tell the user to issue a DROP against the
+ * referenced object, instead). While a regular internal dependency will
+ * prevent the dependent object from being dropped while any such
+ * dependencies remain, DEPENDENCY_INTERNAL_AUTO will allow such a drop as
+ * long as the object can be found by following any of such dependencies.
+ * Example: an index on a partition is made internal-auto-dependent on
+ * both the partition itself as well as on the index on the parent
+ * partitioned table; so the partition index is dropped together with
+ * either the partition it indexes, or with the parent index it is attached
+ * to.
+
* DEPENDENCY_EXTENSION ('e'): the dependent object is a member of the
* extension that is the referenced object. The dependent object can be
* dropped only via DROP EXTENSION on the referenced object. Functionally
@@ -75,6 +89,7 @@ typedef enum DependencyType
DEPENDENCY_NORMAL = 'n',
DEPENDENCY_AUTO = 'a',
DEPENDENCY_INTERNAL = 'i',
+ DEPENDENCY_INTERNAL_AUTO = 'I',
DEPENDENCY_EXTENSION = 'e',
DEPENDENCY_AUTO_EXTENSION = 'x',
DEPENDENCY_PIN = 'p'
diff --git a/src/include/catalog/index.h b/src/include/catalog/index.h
index 12bf35567a..8ff3a0732e 100644
--- a/src/include/catalog/index.h
+++ b/src/include/catalog/index.h
@@ -47,10 +47,13 @@ extern void index_check_primary_key(Relation heapRel,
#define INDEX_CREATE_SKIP_BUILD (1 << 2)
#define INDEX_CREATE_CONCURRENT (1 << 3)
#define INDEX_CREATE_IF_NOT_EXISTS (1 << 4)
+#define INDEX_CREATE_PARTITIONED (1 << 5)
+#define INDEX_CREATE_INVALID (1 << 6)
extern Oid index_create(Relation heapRelation,
const char *indexRelationName,
Oid indexRelationId,
+ Oid parentIndexRelid,
Oid relFileNode,
IndexInfo *indexInfo,
List *indexColNames,
@@ -84,6 +87,9 @@ extern void index_drop(Oid indexId, bool concurrent);
extern IndexInfo *BuildIndexInfo(Relation index);
+extern bool CompareIndexInfo(IndexInfo *info1, IndexInfo *info2,
+ AttrNumber *attmap, int maplen);
+
extern void BuildSpeculativeIndexInfo(Relation index, IndexInfo *ii);
extern void FormIndexDatum(IndexInfo *indexInfo,
@@ -134,4 +140,6 @@ extern bool ReindexIsProcessingHeap(Oid heapOid);
extern bool ReindexIsProcessingIndex(Oid indexOid);
extern Oid IndexGetRelation(Oid indexId, bool missing_ok);
+extern void IndexSetParentIndex(Relation idx, Oid parentOid);
+
#endif /* INDEX_H */
diff --git a/src/include/catalog/pg_class.h b/src/include/catalog/pg_class.h
index e7049438eb..26b1866c69 100644
--- a/src/include/catalog/pg_class.h
+++ b/src/include/catalog/pg_class.h
@@ -166,6 +166,7 @@ DESCR("");
#define RELKIND_COMPOSITE_TYPE 'c' /* composite type */
#define RELKIND_FOREIGN_TABLE 'f' /* foreign table */
#define RELKIND_PARTITIONED_TABLE 'p' /* partitioned table */
+#define RELKIND_PARTITIONED_INDEX 'I' /* partitioned index */
#define RELPERSISTENCE_PERMANENT 'p' /* regular table */
#define RELPERSISTENCE_UNLOGGED 'u' /* unlogged permanent table */
diff --git a/src/include/catalog/pg_inherits_fn.h b/src/include/catalog/pg_inherits_fn.h
index 405af230d1..eebee977a5 100644
--- a/src/include/catalog/pg_inherits_fn.h
+++ b/src/include/catalog/pg_inherits_fn.h
@@ -23,5 +23,8 @@ extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
extern bool has_subclass(Oid relationId);
extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
+extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
+ int32 seqNumber);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
#endif /* PG_INHERITS_FN_H */
diff --git a/src/include/commands/defrem.h b/src/include/commands/defrem.h
index 1f18cad963..41007162aa 100644
--- a/src/include/commands/defrem.h
+++ b/src/include/commands/defrem.h
@@ -25,12 +25,13 @@ extern void RemoveObjects(DropStmt *stmt);
extern ObjectAddress DefineIndex(Oid relationId,
IndexStmt *stmt,
Oid indexRelationId,
+ Oid parentIndexId,
bool is_alter_table,
bool check_rights,
bool check_not_in_use,
bool skip_build,
bool quiet);
-extern Oid ReindexIndex(RangeVar *indexRelation, int options);
+extern void ReindexIndex(RangeVar *indexRelation, int options);
extern Oid ReindexTable(RangeVar *relation, int options);
extern void ReindexMultipleTables(const char *objectName, ReindexObjectType objectKind,
int options);
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 4bb5cb163d..63a75bd5ed 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -158,6 +158,7 @@ typedef struct IndexInfo
bool ii_ReadyForInserts;
bool ii_Concurrent;
bool ii_BrokenHotChain;
+ Oid ii_Am;
void *ii_AmCache;
MemoryContext ii_Context;
} IndexInfo;
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index b72178efd1..0296784726 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -839,7 +839,7 @@ typedef struct PartitionRangeDatum
} PartitionRangeDatum;
/*
- * PartitionCmd - info for ALTER TABLE ATTACH/DETACH PARTITION commands
+ * PartitionCmd - info for ALTER TABLE/INDEX ATTACH/DETACH PARTITION commands
*/
typedef struct PartitionCmd
{
@@ -2702,6 +2702,10 @@ typedef struct FetchStmt
* index, just a UNIQUE/PKEY constraint using an existing index. isconstraint
* must always be true in this case, and the fields describing the index
* properties are empty.
+ *
+ * The relation to build the index on can be represented either by name
+ * (in which case the RangeVar indicates whether to recurse or not) or by OID
+ * (in which case the command is always recursive).
* ----------------------
*/
typedef struct IndexStmt
@@ -2709,6 +2713,7 @@ typedef struct IndexStmt
NodeTag type;
char *idxname; /* name of new index, or NULL for default */
RangeVar *relation; /* relation to build index on */
+ Oid relationId; /* OID of relation to build index on */
char *accessMethod; /* name of access method (eg. btree) */
char *tableSpace; /* tablespace, or NULL for default */
List *indexParams; /* columns to index: a list of IndexElem */
diff --git a/src/include/parser/parse_utilcmd.h b/src/include/parser/parse_utilcmd.h
index a7f5e0caea..64aa8234e5 100644
--- a/src/include/parser/parse_utilcmd.h
+++ b/src/include/parser/parse_utilcmd.h
@@ -27,5 +27,8 @@ extern void transformRuleStmt(RuleStmt *stmt, const char *queryString,
extern List *transformCreateSchemaStmt(CreateSchemaStmt *stmt);
extern PartitionBoundSpec *transformPartitionBound(ParseState *pstate, Relation parent,
PartitionBoundSpec *spec);
+extern IndexStmt *generateClonedIndexStmt(RangeVar *heapRel, Oid heapOid,
+ Relation source_idx,
+ const AttrNumber *attmap, int attmap_length);
#endif /* PARSE_UTILCMD_H */
diff --git a/src/test/regress/expected/alter_table.out b/src/test/regress/expected/alter_table.out
index 11f0baa11b..517fb080bd 100644
--- a/src/test/regress/expected/alter_table.out
+++ b/src/test/regress/expected/alter_table.out
@@ -1965,6 +1965,67 @@ create table tab1 (a int, b text);
create table tab2 (x int, y tab1);
alter table tab1 alter column b type varchar; -- fails
ERROR: cannot alter table "tab1" because column "tab2.y" uses its row type
+-- Alter column type that's part of a partitioned index
+create table at_partitioned (a int, b text) partition by range (a);
+create table at_part_1 partition of at_partitioned for values from (0) to (1000);
+insert into at_partitioned values (512, '0.123');
+create table at_part_2 (b text, a int);
+insert into at_part_2 values ('1.234', 1024);
+create index on at_partitioned (b);
+create index on at_partitioned (a);
+\d at_part_1
+ Table "public.at_part_1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | text | | |
+Partition of: at_partitioned FOR VALUES FROM (0) TO (1000)
+Indexes:
+ "at_part_1_a_idx" btree (a)
+ "at_part_1_b_idx" btree (b)
+
+\d at_part_2
+ Table "public.at_part_2"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ b | text | | |
+ a | integer | | |
+
+alter table at_partitioned attach partition at_part_2 for values from (1000) to (2000);
+\d at_part_2
+ Table "public.at_part_2"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ b | text | | |
+ a | integer | | |
+Partition of: at_partitioned FOR VALUES FROM (1000) TO (2000)
+Indexes:
+ "at_part_2_a_idx" btree (a)
+ "at_part_2_b_idx" btree (b)
+
+alter table at_partitioned alter column b type numeric using b::numeric;
+\d at_part_1
+ Table "public.at_part_1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | numeric | | |
+Partition of: at_partitioned FOR VALUES FROM (0) TO (1000)
+Indexes:
+ "at_part_1_a_idx" btree (a)
+ "at_part_1_b_idx" btree (b)
+
+\d at_part_2
+ Table "public.at_part_2"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ b | numeric | | |
+ a | integer | | |
+Partition of: at_partitioned FOR VALUES FROM (1000) TO (2000)
+Indexes:
+ "at_part_2_a_idx" btree (a)
+ "at_part_2_b_idx" btree (b)
+
-- disallow recursive containment of row types
create temp table recur1 (f1 int);
alter table recur1 add column f2 recur1; -- fails
@@ -3276,7 +3337,7 @@ CREATE TABLE unparted (
);
CREATE TABLE fail_part (like unparted);
ALTER TABLE unparted ATTACH PARTITION fail_part FOR VALUES IN ('a');
-ERROR: "unparted" is not partitioned
+ERROR: table "unparted" is not partitioned
DROP TABLE unparted, fail_part;
-- check that partition bound is compatible
CREATE TABLE list_parted (
@@ -3656,7 +3717,7 @@ DROP TABLE fail_part;
-- check that the table is partitioned at all
CREATE TABLE regular_table (a int);
ALTER TABLE regular_table DETACH PARTITION any_name;
-ERROR: "regular_table" is not partitioned
+ERROR: table "regular_table" is not partitioned
DROP TABLE regular_table;
-- check that the partition being detached exists at all
ALTER TABLE list_parted2 DETACH PARTITION part_4;
diff --git a/src/test/regress/expected/indexing.out b/src/test/regress/expected/indexing.out
new file mode 100644
index 0000000000..da5f557b72
--- /dev/null
+++ b/src/test/regress/expected/indexing.out
@@ -0,0 +1,604 @@
+-- Creating an index on a partitioned table makes the partitions
+-- automatically get the index
+create table idxpart (a int, b int, c text) partition by range (a);
+create table idxpart1 partition of idxpart for values from (0) to (10);
+create table idxpart2 partition of idxpart for values from (10) to (100)
+ partition by range (b);
+create table idxpart21 partition of idxpart2 for values from (0) to (100);
+create index on idxpart (a);
+select relname, relkind, inhparent::regclass
+ from pg_class left join pg_index ix on (indexrelid = oid)
+ left join pg_inherits on (ix.indexrelid = inhrelid)
+ where relname like 'idxpart%' order by relname;
+ relname | relkind | inhparent
+-----------------+---------+----------------
+ idxpart | p |
+ idxpart1 | r |
+ idxpart1_a_idx | i | idxpart_a_idx
+ idxpart2 | p |
+ idxpart21 | r |
+ idxpart21_a_idx | i | idxpart2_a_idx
+ idxpart2_a_idx | I | idxpart_a_idx
+ idxpart_a_idx | I |
+(8 rows)
+
+drop table idxpart;
+-- Some unsupported features
+create table idxpart (a int, b int, c text) partition by range (a);
+create table idxpart1 partition of idxpart for values from (0) to (10);
+create unique index on idxpart (a);
+ERROR: cannot create unique index on partitioned table "idxpart"
+create index concurrently on idxpart (a);
+ERROR: cannot create index on partitioned table "idxpart" concurrently
+drop table idxpart;
+-- If a table without index is attached as partition to a table with
+-- an index, the index is automatically created
+create table idxpart (a int, b int, c text) partition by range (a);
+create index idxparti on idxpart (a);
+create index idxparti2 on idxpart (b, c);
+create table idxpart1 (like idxpart);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | integer | | |
+ c | text | | |
+
+alter table idxpart attach partition idxpart1 for values from (0) to (10);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | integer | | |
+ c | text | | |
+Partition of: idxpart FOR VALUES FROM (0) TO (10)
+Indexes:
+ "idxpart1_a_idx" btree (a)
+ "idxpart1_b_c_idx" btree (b, c)
+
+drop table idxpart;
+-- If a partition already has an index, don't create a duplicative one
+create table idxpart (a int, b int) partition by range (a, b);
+create table idxpart1 partition of idxpart for values from (0, 0) to (10, 10);
+create index on idxpart1 (a, b);
+create index on idxpart (a, b);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | integer | | |
+Partition of: idxpart FOR VALUES FROM (0, 0) TO (10, 10)
+Indexes:
+ "idxpart1_a_b_idx" btree (a, b)
+
+select relname, relkind, inhparent::regclass
+ from pg_class left join pg_index ix on (indexrelid = oid)
+ left join pg_inherits on (ix.indexrelid = inhrelid)
+ where relname like 'idxpart%' order by relname;
+ relname | relkind | inhparent
+------------------+---------+-----------------
+ idxpart | p |
+ idxpart1 | r |
+ idxpart1_a_b_idx | i | idxpart_a_b_idx
+ idxpart_a_b_idx | I |
+(4 rows)
+
+drop table idxpart;
+-- DROP behavior for partitioned indexes
+create table idxpart (a int) partition by range (a);
+create index on idxpart (a);
+create table idxpart1 partition of idxpart for values from (0) to (10);
+drop index idxpart1_a_idx; -- no way
+ERROR: cannot drop index idxpart1_a_idx because index idxpart_a_idx requires it
+HINT: You can drop index idxpart_a_idx instead.
+drop index idxpart_a_idx; -- both indexes go away
+select relname, relkind from pg_class
+ where relname like 'idxpart%' order by relname;
+ relname | relkind
+----------+---------
+ idxpart | p
+ idxpart1 | r
+(2 rows)
+
+create index on idxpart (a);
+drop table idxpart1; -- the index on partition goes away too
+select relname, relkind from pg_class
+ where relname like 'idxpart%' order by relname;
+ relname | relkind
+---------------+---------
+ idxpart | p
+ idxpart_a_idx | I
+(2 rows)
+
+drop table idxpart;
+-- ALTER INDEX .. ATTACH, error cases
+create table idxpart (a int, b int) partition by range (a, b);
+create table idxpart1 partition of idxpart for values from (0, 0) to (10, 10);
+create index idxpart_a_b_idx on only idxpart (a, b);
+create index idxpart1_a_b_idx on idxpart1 (a, b);
+create index idxpart1_tst1 on idxpart1 (b, a);
+create index idxpart1_tst2 on idxpart1 using hash (a);
+create index idxpart1_tst3 on idxpart1 (a, b) where a > 10;
+alter index idxpart attach partition idxpart1;
+ERROR: "idxpart" is not an index
+alter index idxpart_a_b_idx attach partition idxpart1;
+ERROR: "idxpart1" is not an index
+alter index idxpart_a_b_idx attach partition idxpart_a_b_idx;
+ERROR: cannot attach index "idxpart_a_b_idx" as a partition of index "idxpart_a_b_idx"
+DETAIL: Index "idxpart_a_b_idx" is not on a partition of table "idxpart".
+alter index idxpart_a_b_idx attach partition idxpart1_b_idx;
+ERROR: relation "idxpart1_b_idx" does not exist
+alter index idxpart_a_b_idx attach partition idxpart1_tst1;
+ERROR: cannot attach index "idxpart1_tst1" as a partition of index "idxpart_a_b_idx"
+DETAIL: The index definitions do not match.
+alter index idxpart_a_b_idx attach partition idxpart1_tst2;
+ERROR: cannot attach index "idxpart1_tst2" as a partition of index "idxpart_a_b_idx"
+DETAIL: The index definitions do not match.
+alter index idxpart_a_b_idx attach partition idxpart1_tst3;
+ERROR: cannot attach index "idxpart1_tst3" as a partition of index "idxpart_a_b_idx"
+DETAIL: The index definitions do not match.
+-- OK
+alter index idxpart_a_b_idx attach partition idxpart1_a_b_idx;
+alter index idxpart_a_b_idx attach partition idxpart1_a_b_idx; -- quiet
+-- reject dupe
+create index idxpart1_2_a_b on idxpart1 (a, b);
+alter index idxpart_a_b_idx attach partition idxpart1_2_a_b;
+ERROR: cannot attach index "idxpart1_2_a_b" as a partition of index "idxpart_a_b_idx"
+DETAIL: Another index is already attached for partition "idxpart1".
+drop table idxpart;
+-- make sure everything's gone
+select indexrelid::regclass, indrelid::regclass
+ from pg_index where indexrelid::regclass::text like 'idxpart%';
+ indexrelid | indrelid
+------------+----------
+(0 rows)
+
+-- Don't auto-attach incompatible indexes
+create table idxpart (a int, b int) partition by range (a);
+create table idxpart1 (a int, b int);
+create index on idxpart1 using hash (a);
+create index on idxpart1 (a) where b > 1;
+create index on idxpart1 ((a + 0));
+create index on idxpart1 (a, a);
+create index on idxpart (a);
+alter table idxpart attach partition idxpart1 for values from (0) to (1000);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | integer | | |
+Partition of: idxpart FOR VALUES FROM (0) TO (1000)
+Indexes:
+ "idxpart1_a_a1_idx" btree (a, a)
+ "idxpart1_a_idx" hash (a)
+ "idxpart1_a_idx1" btree (a) WHERE b > 1
+ "idxpart1_a_idx2" btree (a)
+ "idxpart1_expr_idx" btree ((a + 0))
+
+drop table idxpart;
+-- If CREATE INDEX ONLY, don't create indexes on partitions; and existing
+-- indexes on partitions don't change parent. ALTER INDEX ATTACH can change
+-- the parent after the fact.
+create table idxpart (a int) partition by range (a);
+create table idxpart1 partition of idxpart for values from (0) to (100);
+create table idxpart2 partition of idxpart for values from (100) to (1000)
+ partition by range (a);
+create table idxpart21 partition of idxpart2 for values from (100) to (200);
+create table idxpart22 partition of idxpart2 for values from (200) to (300);
+create index on idxpart22 (a);
+create index on only idxpart2 (a);
+create index on idxpart (a);
+-- Here we expect that idxpart1 and idxpart2 have a new index, but idxpart21
+-- does not; also, idxpart22 is not attached.
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition of: idxpart FOR VALUES FROM (0) TO (100)
+Indexes:
+ "idxpart1_a_idx" btree (a)
+
+\d idxpart2
+ Table "public.idxpart2"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition of: idxpart FOR VALUES FROM (100) TO (1000)
+Partition key: RANGE (a)
+Indexes:
+ "idxpart2_a_idx" btree (a) INVALID
+Number of partitions: 2 (Use \d+ to list them.)
+
+\d idxpart21
+ Table "public.idxpart21"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition of: idxpart2 FOR VALUES FROM (100) TO (200)
+
+select indexrelid::regclass, indrelid::regclass, inhparent::regclass
+ from pg_index idx left join pg_inherits inh on (idx.indexrelid = inh.inhrelid)
+where indexrelid::regclass::text like 'idxpart%'
+ order by indrelid::regclass;
+ indexrelid | indrelid | inhparent
+-----------------+-----------+---------------
+ idxpart_a_idx | idxpart |
+ idxpart1_a_idx | idxpart1 | idxpart_a_idx
+ idxpart2_a_idx | idxpart2 | idxpart_a_idx
+ idxpart22_a_idx | idxpart22 |
+(4 rows)
+
+alter index idxpart2_a_idx attach partition idxpart22_a_idx;
+select indexrelid::regclass, indrelid::regclass, inhparent::regclass
+ from pg_index idx left join pg_inherits inh on (idx.indexrelid = inh.inhrelid)
+where indexrelid::regclass::text like 'idxpart%'
+ order by indrelid::regclass;
+ indexrelid | indrelid | inhparent
+-----------------+-----------+----------------
+ idxpart_a_idx | idxpart |
+ idxpart1_a_idx | idxpart1 | idxpart_a_idx
+ idxpart2_a_idx | idxpart2 | idxpart_a_idx
+ idxpart22_a_idx | idxpart22 | idxpart2_a_idx
+(4 rows)
+
+-- attaching idxpart22 is not enough to set idxpart22_a_idx valid ...
+alter index idxpart2_a_idx attach partition idxpart22_a_idx;
+\d idxpart2
+ Table "public.idxpart2"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition of: idxpart FOR VALUES FROM (100) TO (1000)
+Partition key: RANGE (a)
+Indexes:
+ "idxpart2_a_idx" btree (a) INVALID
+Number of partitions: 2 (Use \d+ to list them.)
+
+-- ... but this one is.
+create index on idxpart21 (a);
+alter index idxpart2_a_idx attach partition idxpart21_a_idx;
+\d idxpart2
+ Table "public.idxpart2"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition of: idxpart FOR VALUES FROM (100) TO (1000)
+Partition key: RANGE (a)
+Indexes:
+ "idxpart2_a_idx" btree (a)
+Number of partitions: 2 (Use \d+ to list them.)
+
+drop table idxpart;
+-- When a table is attached a partition and it already has an index, a
+-- duplicate index should not get created, but rather the index becomes
+-- attached to the parent's index.
+create table idxpart (a int, b int, c text) partition by range (a);
+create index idxparti on idxpart (a);
+create index idxparti2 on idxpart (b, c);
+create table idxpart1 (like idxpart including indexes);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | integer | | |
+ c | text | | |
+Indexes:
+ "idxpart1_a_idx" btree (a)
+ "idxpart1_b_c_idx" btree (b, c)
+
+select relname, relkind, inhparent::regclass
+ from pg_class left join pg_index ix on (indexrelid = oid)
+ left join pg_inherits on (ix.indexrelid = inhrelid)
+ where relname like 'idxpart%' order by relname;
+ relname | relkind | inhparent
+------------------+---------+-----------
+ idxpart | p |
+ idxpart1 | r |
+ idxpart1_a_idx | i |
+ idxpart1_b_c_idx | i |
+ idxparti | I |
+ idxparti2 | I |
+(6 rows)
+
+alter table idxpart attach partition idxpart1 for values from (0) to (10);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | integer | | |
+ c | text | | |
+Partition of: idxpart FOR VALUES FROM (0) TO (10)
+Indexes:
+ "idxpart1_a_idx" btree (a)
+ "idxpart1_b_c_idx" btree (b, c)
+
+select relname, relkind, inhparent::regclass
+ from pg_class left join pg_index ix on (indexrelid = oid)
+ left join pg_inherits on (ix.indexrelid = inhrelid)
+ where relname like 'idxpart%' order by relname;
+ relname | relkind | inhparent
+------------------+---------+-----------
+ idxpart | p |
+ idxpart1 | r |
+ idxpart1_a_idx | i | idxparti
+ idxpart1_b_c_idx | i | idxparti2
+ idxparti | I |
+ idxparti2 | I |
+(6 rows)
+
+drop table idxpart;
+-- Verify that attaching an invalid index does not mark the parent index valid.
+-- On the other hand, attaching a valid index marks not only its direct
+-- ancestor valid, but also any indirect ancestor that was only missing the one
+-- that was just made valid
+create table idxpart (a int, b int) partition by range (a);
+create table idxpart1 partition of idxpart for values from (1) to (1000) partition by range (a);
+create table idxpart11 partition of idxpart1 for values from (1) to (100);
+create index on only idxpart1 (a);
+create index on only idxpart (a);
+-- this results in two invalid indexes:
+select relname, indisvalid from pg_class join pg_index on indexrelid = oid
+ where relname like 'idxpart%' order by relname;
+ relname | indisvalid
+----------------+------------
+ idxpart1_a_idx | f
+ idxpart_a_idx | f
+(2 rows)
+
+-- idxpart1_a_idx is not valid, so idxpart_a_idx should not become valid:
+alter index idxpart_a_idx attach partition idxpart1_a_idx;
+select relname, indisvalid from pg_class join pg_index on indexrelid = oid
+ where relname like 'idxpart%' order by relname;
+ relname | indisvalid
+----------------+------------
+ idxpart1_a_idx | f
+ idxpart_a_idx | f
+(2 rows)
+
+-- after creating and attaching this, both idxpart1_a_idx and idxpart_a_idx
+-- should become valid
+create index on idxpart11 (a);
+alter index idxpart1_a_idx attach partition idxpart11_a_idx;
+select relname, indisvalid from pg_class join pg_index on indexrelid = oid
+ where relname like 'idxpart%' order by relname;
+ relname | indisvalid
+-----------------+------------
+ idxpart11_a_idx | t
+ idxpart1_a_idx | t
+ idxpart_a_idx | t
+(3 rows)
+
+drop table idxpart;
+-- verify dependency handling during ALTER TABLE DETACH PARTITION
+create table idxpart (a int) partition by range (a);
+create table idxpart1 (like idxpart);
+create index on idxpart1 (a);
+create index on idxpart (a);
+create table idxpart2 (like idxpart);
+alter table idxpart attach partition idxpart1 for values from (0000) to (1000);
+alter table idxpart attach partition idxpart2 for values from (1000) to (2000);
+create table idxpart3 partition of idxpart for values from (2000) to (3000);
+select relname, relkind from pg_class where relname like 'idxpart%' order by relname;
+ relname | relkind
+----------------+---------
+ idxpart | p
+ idxpart1 | r
+ idxpart1_a_idx | i
+ idxpart2 | r
+ idxpart2_a_idx | i
+ idxpart3 | r
+ idxpart3_a_idx | i
+ idxpart_a_idx | I
+(8 rows)
+
+-- a) after detaching partitions, the indexes can be dropped independently
+alter table idxpart detach partition idxpart1;
+alter table idxpart detach partition idxpart2;
+alter table idxpart detach partition idxpart3;
+drop index idxpart1_a_idx;
+drop index idxpart2_a_idx;
+drop index idxpart3_a_idx;
+select relname, relkind from pg_class where relname like 'idxpart%' order by relname;
+ relname | relkind
+---------------+---------
+ idxpart | p
+ idxpart1 | r
+ idxpart2 | r
+ idxpart3 | r
+ idxpart_a_idx | I
+(5 rows)
+
+drop table idxpart, idxpart1, idxpart2, idxpart3;
+select relname, relkind from pg_class where relname like 'idxpart%' order by relname;
+ relname | relkind
+---------+---------
+(0 rows)
+
+create table idxpart (a int) partition by range (a);
+create table idxpart1 (like idxpart);
+create index on idxpart1 (a);
+create index on idxpart (a);
+create table idxpart2 (like idxpart);
+alter table idxpart attach partition idxpart1 for values from (0000) to (1000);
+alter table idxpart attach partition idxpart2 for values from (1000) to (2000);
+create table idxpart3 partition of idxpart for values from (2000) to (3000);
+-- b) after detaching, dropping the index on parent does not remove the others
+select relname, relkind from pg_class where relname like 'idxpart%' order by relname;
+ relname | relkind
+----------------+---------
+ idxpart | p
+ idxpart1 | r
+ idxpart1_a_idx | i
+ idxpart2 | r
+ idxpart2_a_idx | i
+ idxpart3 | r
+ idxpart3_a_idx | i
+ idxpart_a_idx | I
+(8 rows)
+
+alter table idxpart detach partition idxpart1;
+alter table idxpart detach partition idxpart2;
+alter table idxpart detach partition idxpart3;
+drop index idxpart_a_idx;
+select relname, relkind from pg_class where relname like 'idxpart%' order by relname;
+ relname | relkind
+----------------+---------
+ idxpart | p
+ idxpart1 | r
+ idxpart1_a_idx | i
+ idxpart2 | r
+ idxpart2_a_idx | i
+ idxpart3 | r
+ idxpart3_a_idx | i
+(7 rows)
+
+drop table idxpart, idxpart1, idxpart2, idxpart3;
+select relname, relkind from pg_class where relname like 'idxpart%' order by relname;
+ relname | relkind
+---------+---------
+(0 rows)
+
+-- Verify that expression indexes inherit correctly
+create table idxpart (a int, b int) partition by range (a);
+create table idxpart1 (like idxpart);
+create index on idxpart1 ((a + b));
+create index on idxpart ((a + b));
+create table idxpart2 (like idxpart);
+alter table idxpart attach partition idxpart1 for values from (0000) to (1000);
+alter table idxpart attach partition idxpart2 for values from (1000) to (2000);
+create table idxpart3 partition of idxpart for values from (2000) to (3000);
+select relname as child, inhparent::regclass as parent, pg_get_indexdef as childdef
+ from pg_class join pg_inherits on inhrelid = oid,
+ lateral pg_get_indexdef(pg_class.oid)
+ where relkind in ('i', 'I') and relname like 'idxpart%' order by relname;
+ child | parent | childdef
+-------------------+------------------+--------------------------------------------------------------------
+ idxpart1_expr_idx | idxpart_expr_idx | CREATE INDEX idxpart1_expr_idx ON idxpart1 USING btree (((a + b)))
+ idxpart2_expr_idx | idxpart_expr_idx | CREATE INDEX idxpart2_expr_idx ON idxpart2 USING btree (((a + b)))
+ idxpart3_expr_idx | idxpart_expr_idx | CREATE INDEX idxpart3_expr_idx ON idxpart3 USING btree (((a + b)))
+(3 rows)
+
+drop table idxpart;
+-- Make sure the partition columns are mapped correctly
+create table idxpart (a int, b int, c text) partition by range (a);
+create index idxparti on idxpart (a);
+create index idxparti2 on idxpart (c, b);
+create table idxpart1 (c text, a int, b int);
+alter table idxpart attach partition idxpart1 for values from (0) to (10);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ c | text | | |
+ a | integer | | |
+ b | integer | | |
+Partition of: idxpart FOR VALUES FROM (0) TO (10)
+Indexes:
+ "idxpart1_a_idx" btree (a)
+ "idxpart1_c_b_idx" btree (c, b)
+
+drop table idxpart;
+-- Column number mapping: dropped columns in the partition
+create table idxpart1 (drop_1 int, drop_2 int, col_keep int, drop_3 int);
+alter table idxpart1 drop column drop_1;
+alter table idxpart1 drop column drop_2;
+alter table idxpart1 drop column drop_3;
+create index on idxpart1 (col_keep);
+create table idxpart (col_keep int) partition by range (col_keep);
+create index on idxpart (col_keep);
+alter table idxpart attach partition idxpart1 for values from (0) to (1000);
+\d idxpart
+ Table "public.idxpart"
+ Column | Type | Collation | Nullable | Default
+----------+---------+-----------+----------+---------
+ col_keep | integer | | |
+Partition key: RANGE (col_keep)
+Indexes:
+ "idxpart_col_keep_idx" btree (col_keep)
+Number of partitions: 1 (Use \d+ to list them.)
+
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+----------+---------+-----------+----------+---------
+ col_keep | integer | | |
+Partition of: idxpart FOR VALUES FROM (0) TO (1000)
+Indexes:
+ "idxpart1_col_keep_idx" btree (col_keep)
+
+select attrelid::regclass, attname, attnum from pg_attribute
+ where attrelid::regclass::text like 'idxpart%' and attnum > 0
+ order by attrelid::regclass, attnum;
+ attrelid | attname | attnum
+-----------------------+------------------------------+--------
+ idxpart1 | ........pg.dropped.1........ | 1
+ idxpart1 | ........pg.dropped.2........ | 2
+ idxpart1 | col_keep | 3
+ idxpart1 | ........pg.dropped.4........ | 4
+ idxpart1_col_keep_idx | col_keep | 1
+ idxpart | col_keep | 1
+ idxpart_col_keep_idx | col_keep | 1
+(7 rows)
+
+drop table idxpart;
+-- Column number mapping: dropped columns in the parent table
+create table idxpart(drop_1 int, drop_2 int, col_keep int, drop_3 int) partition by range (col_keep);
+alter table idxpart drop column drop_1;
+alter table idxpart drop column drop_2;
+alter table idxpart drop column drop_3;
+create table idxpart1 (col_keep int);
+create index on idxpart1 (col_keep);
+create index on idxpart (col_keep);
+alter table idxpart attach partition idxpart1 for values from (0) to (1000);
+\d idxpart
+ Table "public.idxpart"
+ Column | Type | Collation | Nullable | Default
+----------+---------+-----------+----------+---------
+ col_keep | integer | | |
+Partition key: RANGE (col_keep)
+Indexes:
+ "idxpart_col_keep_idx" btree (col_keep)
+Number of partitions: 1 (Use \d+ to list them.)
+
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+----------+---------+-----------+----------+---------
+ col_keep | integer | | |
+Partition of: idxpart FOR VALUES FROM (0) TO (1000)
+Indexes:
+ "idxpart1_col_keep_idx" btree (col_keep)
+
+select attrelid::regclass, attname, attnum from pg_attribute
+ where attrelid::regclass::text like 'idxpart%' and attnum > 0
+ order by attrelid::regclass, attnum;
+ attrelid | attname | attnum
+-----------------------+------------------------------+--------
+ idxpart | ........pg.dropped.1........ | 1
+ idxpart | ........pg.dropped.2........ | 2
+ idxpart | col_keep | 3
+ idxpart | ........pg.dropped.4........ | 4
+ idxpart1 | col_keep | 1
+ idxpart1_col_keep_idx | col_keep | 1
+ idxpart_col_keep_idx | col_keep | 1
+(7 rows)
+
+drop table idxpart;
+-- intentionally leave some objects around
+create table idxpart (a int) partition by range (a);
+create table idxpart1 partition of idxpart for values from (0) to (100);
+create table idxpart2 partition of idxpart for values from (100) to (1000)
+ partition by range (a);
+create table idxpart21 partition of idxpart2 for values from (100) to (200);
+create table idxpart22 partition of idxpart2 for values from (200) to (300);
+create index on idxpart22 (a);
+create index on only idxpart2 (a);
+alter index idxpart2_a_idx attach partition idxpart22_a_idx;
+create index on idxpart (a);
diff --git a/src/test/regress/parallel_schedule b/src/test/regress/parallel_schedule
index e224977791..ad9434fb87 100644
--- a/src/test/regress/parallel_schedule
+++ b/src/test/regress/parallel_schedule
@@ -116,7 +116,7 @@ test: plancache limit plpgsql copy2 temp domain rangefuncs prepare without_oid c
# ----------
# Another group of parallel tests
# ----------
-test: identity partition_join partition_prune reloptions hash_part
+test: identity partition_join partition_prune reloptions hash_part indexing
# event triggers cannot run concurrently with any test that runs DDL
test: event_trigger
diff --git a/src/test/regress/serial_schedule b/src/test/regress/serial_schedule
index 9fc5f1a268..27cd49845e 100644
--- a/src/test/regress/serial_schedule
+++ b/src/test/regress/serial_schedule
@@ -184,5 +184,6 @@ test: partition_join
test: partition_prune
test: reloptions
test: hash_part
+test: indexing
test: event_trigger
test: stats
diff --git a/src/test/regress/sql/alter_table.sql b/src/test/regress/sql/alter_table.sql
index 02a33ca7c4..af25ee9e77 100644
--- a/src/test/regress/sql/alter_table.sql
+++ b/src/test/regress/sql/alter_table.sql
@@ -1330,6 +1330,22 @@ create table tab1 (a int, b text);
create table tab2 (x int, y tab1);
alter table tab1 alter column b type varchar; -- fails
+-- Alter column type that's part of a partitioned index
+create table at_partitioned (a int, b text) partition by range (a);
+create table at_part_1 partition of at_partitioned for values from (0) to (1000);
+insert into at_partitioned values (512, '0.123');
+create table at_part_2 (b text, a int);
+insert into at_part_2 values ('1.234', 1024);
+create index on at_partitioned (b);
+create index on at_partitioned (a);
+\d at_part_1
+\d at_part_2
+alter table at_partitioned attach partition at_part_2 for values from (1000) to (2000);
+\d at_part_2
+alter table at_partitioned alter column b type numeric using b::numeric;
+\d at_part_1
+\d at_part_2
+
-- disallow recursive containment of row types
create temp table recur1 (f1 int);
alter table recur1 add column f2 recur1; -- fails
diff --git a/src/test/regress/sql/indexing.sql b/src/test/regress/sql/indexing.sql
new file mode 100644
index 0000000000..2380c7cdab
--- /dev/null
+++ b/src/test/regress/sql/indexing.sql
@@ -0,0 +1,283 @@
+-- Creating an index on a partitioned table makes the partitions
+-- automatically get the index
+create table idxpart (a int, b int, c text) partition by range (a);
+create table idxpart1 partition of idxpart for values from (0) to (10);
+create table idxpart2 partition of idxpart for values from (10) to (100)
+ partition by range (b);
+create table idxpart21 partition of idxpart2 for values from (0) to (100);
+create index on idxpart (a);
+select relname, relkind, inhparent::regclass
+ from pg_class left join pg_index ix on (indexrelid = oid)
+ left join pg_inherits on (ix.indexrelid = inhrelid)
+ where relname like 'idxpart%' order by relname;
+drop table idxpart;
+
+-- Some unsupported features
+create table idxpart (a int, b int, c text) partition by range (a);
+create table idxpart1 partition of idxpart for values from (0) to (10);
+create unique index on idxpart (a);
+create index concurrently on idxpart (a);
+drop table idxpart;
+
+-- If a table without index is attached as partition to a table with
+-- an index, the index is automatically created
+create table idxpart (a int, b int, c text) partition by range (a);
+create index idxparti on idxpart (a);
+create index idxparti2 on idxpart (b, c);
+create table idxpart1 (like idxpart);
+\d idxpart1
+alter table idxpart attach partition idxpart1 for values from (0) to (10);
+\d idxpart1
+drop table idxpart;
+
+-- If a partition already has an index, don't create a duplicative one
+create table idxpart (a int, b int) partition by range (a, b);
+create table idxpart1 partition of idxpart for values from (0, 0) to (10, 10);
+create index on idxpart1 (a, b);
+create index on idxpart (a, b);
+\d idxpart1
+select relname, relkind, inhparent::regclass
+ from pg_class left join pg_index ix on (indexrelid = oid)
+ left join pg_inherits on (ix.indexrelid = inhrelid)
+ where relname like 'idxpart%' order by relname;
+drop table idxpart;
+
+-- DROP behavior for partitioned indexes
+create table idxpart (a int) partition by range (a);
+create index on idxpart (a);
+create table idxpart1 partition of idxpart for values from (0) to (10);
+drop index idxpart1_a_idx; -- no way
+drop index idxpart_a_idx; -- both indexes go away
+select relname, relkind from pg_class
+ where relname like 'idxpart%' order by relname;
+create index on idxpart (a);
+drop table idxpart1; -- the index on partition goes away too
+select relname, relkind from pg_class
+ where relname like 'idxpart%' order by relname;
+drop table idxpart;
+
+-- ALTER INDEX .. ATTACH, error cases
+create table idxpart (a int, b int) partition by range (a, b);
+create table idxpart1 partition of idxpart for values from (0, 0) to (10, 10);
+create index idxpart_a_b_idx on only idxpart (a, b);
+create index idxpart1_a_b_idx on idxpart1 (a, b);
+create index idxpart1_tst1 on idxpart1 (b, a);
+create index idxpart1_tst2 on idxpart1 using hash (a);
+create index idxpart1_tst3 on idxpart1 (a, b) where a > 10;
+
+alter index idxpart attach partition idxpart1;
+alter index idxpart_a_b_idx attach partition idxpart1;
+alter index idxpart_a_b_idx attach partition idxpart_a_b_idx;
+alter index idxpart_a_b_idx attach partition idxpart1_b_idx;
+alter index idxpart_a_b_idx attach partition idxpart1_tst1;
+alter index idxpart_a_b_idx attach partition idxpart1_tst2;
+alter index idxpart_a_b_idx attach partition idxpart1_tst3;
+-- OK
+alter index idxpart_a_b_idx attach partition idxpart1_a_b_idx;
+alter index idxpart_a_b_idx attach partition idxpart1_a_b_idx; -- quiet
+
+-- reject dupe
+create index idxpart1_2_a_b on idxpart1 (a, b);
+alter index idxpart_a_b_idx attach partition idxpart1_2_a_b;
+drop table idxpart;
+-- make sure everything's gone
+select indexrelid::regclass, indrelid::regclass
+ from pg_index where indexrelid::regclass::text like 'idxpart%';
+
+-- Don't auto-attach incompatible indexes
+create table idxpart (a int, b int) partition by range (a);
+create table idxpart1 (a int, b int);
+create index on idxpart1 using hash (a);
+create index on idxpart1 (a) where b > 1;
+create index on idxpart1 ((a + 0));
+create index on idxpart1 (a, a);
+create index on idxpart (a);
+alter table idxpart attach partition idxpart1 for values from (0) to (1000);
+\d idxpart1
+drop table idxpart;
+
+-- If CREATE INDEX ONLY, don't create indexes on partitions; and existing
+-- indexes on partitions don't change parent. ALTER INDEX ATTACH can change
+-- the parent after the fact.
+create table idxpart (a int) partition by range (a);
+create table idxpart1 partition of idxpart for values from (0) to (100);
+create table idxpart2 partition of idxpart for values from (100) to (1000)
+ partition by range (a);
+create table idxpart21 partition of idxpart2 for values from (100) to (200);
+create table idxpart22 partition of idxpart2 for values from (200) to (300);
+create index on idxpart22 (a);
+create index on only idxpart2 (a);
+create index on idxpart (a);
+-- Here we expect that idxpart1 and idxpart2 have a new index, but idxpart21
+-- does not; also, idxpart22 is not attached.
+\d idxpart1
+\d idxpart2
+\d idxpart21
+select indexrelid::regclass, indrelid::regclass, inhparent::regclass
+ from pg_index idx left join pg_inherits inh on (idx.indexrelid = inh.inhrelid)
+where indexrelid::regclass::text like 'idxpart%'
+ order by indrelid::regclass;
+alter index idxpart2_a_idx attach partition idxpart22_a_idx;
+select indexrelid::regclass, indrelid::regclass, inhparent::regclass
+ from pg_index idx left join pg_inherits inh on (idx.indexrelid = inh.inhrelid)
+where indexrelid::regclass::text like 'idxpart%'
+ order by indrelid::regclass;
+-- attaching idxpart22 is not enough to set idxpart22_a_idx valid ...
+alter index idxpart2_a_idx attach partition idxpart22_a_idx;
+\d idxpart2
+-- ... but this one is.
+create index on idxpart21 (a);
+alter index idxpart2_a_idx attach partition idxpart21_a_idx;
+\d idxpart2
+drop table idxpart;
+
+-- When a table is attached a partition and it already has an index, a
+-- duplicate index should not get created, but rather the index becomes
+-- attached to the parent's index.
+create table idxpart (a int, b int, c text) partition by range (a);
+create index idxparti on idxpart (a);
+create index idxparti2 on idxpart (b, c);
+create table idxpart1 (like idxpart including indexes);
+\d idxpart1
+select relname, relkind, inhparent::regclass
+ from pg_class left join pg_index ix on (indexrelid = oid)
+ left join pg_inherits on (ix.indexrelid = inhrelid)
+ where relname like 'idxpart%' order by relname;
+alter table idxpart attach partition idxpart1 for values from (0) to (10);
+\d idxpart1
+select relname, relkind, inhparent::regclass
+ from pg_class left join pg_index ix on (indexrelid = oid)
+ left join pg_inherits on (ix.indexrelid = inhrelid)
+ where relname like 'idxpart%' order by relname;
+drop table idxpart;
+
+-- Verify that attaching an invalid index does not mark the parent index valid.
+-- On the other hand, attaching a valid index marks not only its direct
+-- ancestor valid, but also any indirect ancestor that was only missing the one
+-- that was just made valid
+create table idxpart (a int, b int) partition by range (a);
+create table idxpart1 partition of idxpart for values from (1) to (1000) partition by range (a);
+create table idxpart11 partition of idxpart1 for values from (1) to (100);
+create index on only idxpart1 (a);
+create index on only idxpart (a);
+-- this results in two invalid indexes:
+select relname, indisvalid from pg_class join pg_index on indexrelid = oid
+ where relname like 'idxpart%' order by relname;
+-- idxpart1_a_idx is not valid, so idxpart_a_idx should not become valid:
+alter index idxpart_a_idx attach partition idxpart1_a_idx;
+select relname, indisvalid from pg_class join pg_index on indexrelid = oid
+ where relname like 'idxpart%' order by relname;
+-- after creating and attaching this, both idxpart1_a_idx and idxpart_a_idx
+-- should become valid
+create index on idxpart11 (a);
+alter index idxpart1_a_idx attach partition idxpart11_a_idx;
+select relname, indisvalid from pg_class join pg_index on indexrelid = oid
+ where relname like 'idxpart%' order by relname;
+drop table idxpart;
+
+-- verify dependency handling during ALTER TABLE DETACH PARTITION
+create table idxpart (a int) partition by range (a);
+create table idxpart1 (like idxpart);
+create index on idxpart1 (a);
+create index on idxpart (a);
+create table idxpart2 (like idxpart);
+alter table idxpart attach partition idxpart1 for values from (0000) to (1000);
+alter table idxpart attach partition idxpart2 for values from (1000) to (2000);
+create table idxpart3 partition of idxpart for values from (2000) to (3000);
+select relname, relkind from pg_class where relname like 'idxpart%' order by relname;
+-- a) after detaching partitions, the indexes can be dropped independently
+alter table idxpart detach partition idxpart1;
+alter table idxpart detach partition idxpart2;
+alter table idxpart detach partition idxpart3;
+drop index idxpart1_a_idx;
+drop index idxpart2_a_idx;
+drop index idxpart3_a_idx;
+select relname, relkind from pg_class where relname like 'idxpart%' order by relname;
+drop table idxpart, idxpart1, idxpart2, idxpart3;
+select relname, relkind from pg_class where relname like 'idxpart%' order by relname;
+
+create table idxpart (a int) partition by range (a);
+create table idxpart1 (like idxpart);
+create index on idxpart1 (a);
+create index on idxpart (a);
+create table idxpart2 (like idxpart);
+alter table idxpart attach partition idxpart1 for values from (0000) to (1000);
+alter table idxpart attach partition idxpart2 for values from (1000) to (2000);
+create table idxpart3 partition of idxpart for values from (2000) to (3000);
+-- b) after detaching, dropping the index on parent does not remove the others
+select relname, relkind from pg_class where relname like 'idxpart%' order by relname;
+alter table idxpart detach partition idxpart1;
+alter table idxpart detach partition idxpart2;
+alter table idxpart detach partition idxpart3;
+drop index idxpart_a_idx;
+select relname, relkind from pg_class where relname like 'idxpart%' order by relname;
+drop table idxpart, idxpart1, idxpart2, idxpart3;
+select relname, relkind from pg_class where relname like 'idxpart%' order by relname;
+
+-- Verify that expression indexes inherit correctly
+create table idxpart (a int, b int) partition by range (a);
+create table idxpart1 (like idxpart);
+create index on idxpart1 ((a + b));
+create index on idxpart ((a + b));
+create table idxpart2 (like idxpart);
+alter table idxpart attach partition idxpart1 for values from (0000) to (1000);
+alter table idxpart attach partition idxpart2 for values from (1000) to (2000);
+create table idxpart3 partition of idxpart for values from (2000) to (3000);
+select relname as child, inhparent::regclass as parent, pg_get_indexdef as childdef
+ from pg_class join pg_inherits on inhrelid = oid,
+ lateral pg_get_indexdef(pg_class.oid)
+ where relkind in ('i', 'I') and relname like 'idxpart%' order by relname;
+drop table idxpart;
+
+-- Make sure the partition columns are mapped correctly
+create table idxpart (a int, b int, c text) partition by range (a);
+create index idxparti on idxpart (a);
+create index idxparti2 on idxpart (c, b);
+create table idxpart1 (c text, a int, b int);
+alter table idxpart attach partition idxpart1 for values from (0) to (10);
+\d idxpart1
+drop table idxpart;
+
+-- Column number mapping: dropped columns in the partition
+create table idxpart1 (drop_1 int, drop_2 int, col_keep int, drop_3 int);
+alter table idxpart1 drop column drop_1;
+alter table idxpart1 drop column drop_2;
+alter table idxpart1 drop column drop_3;
+create index on idxpart1 (col_keep);
+create table idxpart (col_keep int) partition by range (col_keep);
+create index on idxpart (col_keep);
+alter table idxpart attach partition idxpart1 for values from (0) to (1000);
+\d idxpart
+\d idxpart1
+select attrelid::regclass, attname, attnum from pg_attribute
+ where attrelid::regclass::text like 'idxpart%' and attnum > 0
+ order by attrelid::regclass, attnum;
+drop table idxpart;
+
+-- Column number mapping: dropped columns in the parent table
+create table idxpart(drop_1 int, drop_2 int, col_keep int, drop_3 int) partition by range (col_keep);
+alter table idxpart drop column drop_1;
+alter table idxpart drop column drop_2;
+alter table idxpart drop column drop_3;
+create table idxpart1 (col_keep int);
+create index on idxpart1 (col_keep);
+create index on idxpart (col_keep);
+alter table idxpart attach partition idxpart1 for values from (0) to (1000);
+\d idxpart
+\d idxpart1
+select attrelid::regclass, attname, attnum from pg_attribute
+ where attrelid::regclass::text like 'idxpart%' and attnum > 0
+ order by attrelid::regclass, attnum;
+drop table idxpart;
+
+-- intentionally leave some objects around
+create table idxpart (a int) partition by range (a);
+create table idxpart1 partition of idxpart for values from (0) to (100);
+create table idxpart2 partition of idxpart for values from (100) to (1000)
+ partition by range (a);
+create table idxpart21 partition of idxpart2 for values from (100) to (200);
+create table idxpart22 partition of idxpart2 for values from (200) to (300);
+create index on idxpart22 (a);
+create index on only idxpart2 (a);
+alter index idxpart2_a_idx attach partition idxpart22_a_idx;
+create index on idxpart (a);
--
2.11.0
v3-0002-allow-indexes-on-partitioned-tables-to-be-unique.patchtext/plain; charset=us-asciiDownload
From 1aedc8ceb90b59f9964317093f15c9d3a0108a1c Mon Sep 17 00:00:00 2001
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
Date: Mon, 6 Nov 2017 17:04:55 +0100
Subject: [PATCH v3 2/2] allow indexes on partitioned tables to be unique
---
doc/src/sgml/ref/alter_table.sgml | 9 +-
doc/src/sgml/ref/create_table.sgml | 16 ++-
src/backend/bootstrap/bootparse.y | 2 +
src/backend/catalog/index.c | 28 ++++-
src/backend/catalog/toasting.c | 4 +-
src/backend/commands/indexcmds.c | 80 +++++++++++++--
src/backend/commands/tablecmds.c | 12 ++-
src/backend/parser/analyze.c | 7 ++
src/backend/parser/parse_utilcmd.c | 31 +-----
src/backend/tcop/utility.c | 1 +
src/include/catalog/index.h | 5 +-
src/include/commands/defrem.h | 1 +
src/include/parser/parse_utilcmd.h | 3 +-
src/test/regress/expected/alter_table.out | 8 --
src/test/regress/expected/create_table.out | 12 ---
src/test/regress/expected/indexing.out | 142 +++++++++++++++++++++++++-
src/test/regress/expected/insert_conflict.out | 2 +-
src/test/regress/sql/alter_table.sql | 2 -
src/test/regress/sql/create_table.sql | 8 --
src/test/regress/sql/indexing.sql | 73 ++++++++++++-
20 files changed, 363 insertions(+), 83 deletions(-)
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index 2c41f2107e..ee4c802c03 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -782,8 +782,9 @@ ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
This form attaches an existing table (which might itself be partitioned)
as a partition of the target table. The table can be attached
as a partition for specific values using <literal>FOR VALUES
- </literal> or as a default partition by using <literal>DEFAULT
- </literal>. For each index in the target table, a corresponding
+ </literal> or as a default partition by using
+ <literal>DEFAULT</literal>.
+ For each index in the target table, a corresponding
one will be created in the attached table; or, if an equivalent
index already exists, will be attached to the target table's index,
as if <command>ALTER INDEX ATTACH PARTITION</command> had been executed.
@@ -798,8 +799,10 @@ ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
as the target table and no more; moreover, the column types must also
match. Also, it must have all the <literal>NOT NULL</literal> and
<literal>CHECK</literal> constraints of the target table. Currently
- <literal>UNIQUE</literal>, <literal>PRIMARY KEY</literal>, and
<literal>FOREIGN KEY</literal> constraints are not considered.
+ <literal>UNIQUE</literal> and <literal>PRIMARY KEY</literal> constraints
+ from the parent table will be created in the partition, if they don't
+ already exist.
If any of the <literal>CHECK</literal> constraints of the table being
attached is marked <literal>NO INHERIT</literal>, the command will fail;
such a constraint must be recreated without the <literal>NO INHERIT</literal>
diff --git a/doc/src/sgml/ref/create_table.sgml b/doc/src/sgml/ref/create_table.sgml
index a0c9a6d257..4c56df8960 100644
--- a/doc/src/sgml/ref/create_table.sgml
+++ b/doc/src/sgml/ref/create_table.sgml
@@ -546,8 +546,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</para>
<para>
- Partitioned tables do not support <literal>UNIQUE</literal>,
- <literal>PRIMARY KEY</literal>, <literal>EXCLUDE</literal>, or
+ Partitioned tables do not support
+ <literal>EXCLUDE</literal>, or
<literal>FOREIGN KEY</literal> constraints; however, you can define
these constraints on individual partitions.
</para>
@@ -786,6 +786,11 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
primary key constraint defined for the table. (Otherwise it
would just be the same constraint listed twice.)
</para>
+
+ <para>
+ When used on partitioned tables, <literal>UNIQUE</literal> constraints
+ must include all the columns of the partition key.
+ </para>
</listitem>
</varlistentry>
@@ -814,6 +819,13 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
about the design of the schema, since a primary key implies that other
tables can rely on this set of columns as a unique identifier for rows.
</para>
+
+ <para>
+ <literal>PRIMARY KEY</literal> constraints share the restrictions that
+ <literal>UNIQUE</literal> constraints have when placed on partitioned
+ tables.
+ </para>
+
</listitem>
</varlistentry>
diff --git a/src/backend/bootstrap/bootparse.y b/src/backend/bootstrap/bootparse.y
index dfd53fa054..9e81f9514d 100644
--- a/src/backend/bootstrap/bootparse.y
+++ b/src/backend/bootstrap/bootparse.y
@@ -322,6 +322,7 @@ Boot_DeclareIndexStmt:
stmt,
$4,
InvalidOid,
+ InvalidOid,
false,
false,
false,
@@ -367,6 +368,7 @@ Boot_DeclareUniqueIndexStmt:
stmt,
$5,
InvalidOid,
+ InvalidOid,
false,
false,
false,
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 6a7fa91e25..bf69ab53e7 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -678,6 +678,8 @@ UpdateIndexRelation(Oid indexoid,
* nonzero to specify a preselected OID.
* parentIndexRelid: if creating an index partition, the OID of the
* parent index; otherwise InvalidOid.
+ * parentConstraintId: if creating a constraint on a partition, the OID
+ * of the constraint in the parent; otherwise InvalidOid.
* relFileNode: normally, pass InvalidOid to get new storage. May be
* nonzero to attach an existing valid build.
* indexInfo: same info executor uses to insert into the index
@@ -709,6 +711,7 @@ UpdateIndexRelation(Oid indexoid,
* (only if INDEX_CREATE_ADD_CONSTRAINT is set)
* allow_system_table_mods: allow table to be a system catalog
* is_internal: if true, post creation hook for new index
+ * constraintId: if not NULL, receives OID of created constraint
*
* Returns the OID of the created index.
*/
@@ -717,6 +720,7 @@ index_create(Relation heapRelation,
const char *indexRelationName,
Oid indexRelationId,
Oid parentIndexRelid,
+ Oid parentConstraintId,
Oid relFileNode,
IndexInfo *indexInfo,
List *indexColNames,
@@ -729,7 +733,8 @@ index_create(Relation heapRelation,
bits16 flags,
bits16 constr_flags,
bool allow_system_table_mods,
- bool is_internal)
+ bool is_internal,
+ Oid *constraintId)
{
Oid heapRelationId = RelationGetRelid(heapRelation);
Relation pg_class;
@@ -976,6 +981,7 @@ index_create(Relation heapRelation,
if ((flags & INDEX_CREATE_ADD_CONSTRAINT) != 0)
{
char constraintType;
+ ObjectAddress localaddr;
if (isprimary)
constraintType = CONSTRAINT_PRIMARY;
@@ -989,14 +995,17 @@ index_create(Relation heapRelation,
constraintType = 0; /* keep compiler quiet */
}
- index_constraint_create(heapRelation,
+ localaddr = index_constraint_create(heapRelation,
indexRelationId,
+ parentConstraintId,
indexInfo,
indexRelationName,
constraintType,
constr_flags,
allow_system_table_mods,
is_internal);
+ if (constraintId)
+ *constraintId = localaddr.objectId;
}
else
{
@@ -1167,6 +1176,8 @@ index_create(Relation heapRelation,
*
* heapRelation: table owning the index (must be suitably locked by caller)
* indexRelationId: OID of the index
+ * parentConstraintId: if constraint is on a partition, the OID of the
+ * constraint in the parent.
* indexInfo: same info executor uses to insert into the index
* constraintName: what it say (generally, should match name of index)
* constraintType: one of CONSTRAINT_PRIMARY, CONSTRAINT_UNIQUE, or
@@ -1184,6 +1195,7 @@ index_create(Relation heapRelation,
ObjectAddress
index_constraint_create(Relation heapRelation,
Oid indexRelationId,
+ Oid parentConstraintId,
IndexInfo *indexInfo,
const char *constraintName,
char constraintType,
@@ -1281,6 +1293,18 @@ index_constraint_create(Relation heapRelation,
recordDependencyOn(&myself, &referenced, DEPENDENCY_INTERNAL);
/*
+ * Also, if this is a constraint on a partition, mark it as depending
+ * on the constraint in the parent.
+ */
+ if (OidIsValid(parentConstraintId))
+ {
+ ObjectAddress third;
+
+ ObjectAddressSet(third, ConstraintRelationId, parentConstraintId);
+ recordDependencyOn(&referenced, &third, DEPENDENCY_INTERNAL);
+ }
+
+ /*
* If the constraint is deferrable, create the deferred uniqueness
* checking trigger. (The trigger will be given an internal dependency on
* the constraint by CreateTrigger.)
diff --git a/src/backend/catalog/toasting.c b/src/backend/catalog/toasting.c
index cf37011b73..f4e7b83fee 100644
--- a/src/backend/catalog/toasting.c
+++ b/src/backend/catalog/toasting.c
@@ -329,13 +329,13 @@ create_toast_table(Relation rel, Oid toastOid, Oid toastIndexOid,
coloptions[1] = 0;
index_create(toast_rel, toast_idxname, toastIndexOid, InvalidOid,
- InvalidOid,
+ InvalidOid, InvalidOid,
indexInfo,
list_make2("chunk_id", "chunk_seq"),
BTREE_AM_OID,
rel->rd_rel->reltablespace,
collationObjectId, classObjectId, coloptions, (Datum) 0,
- INDEX_CREATE_IS_PRIMARY, 0, true, true);
+ INDEX_CREATE_IS_PRIMARY, 0, true, true, NULL);
heap_close(toast_rel, NoLock);
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 566a27cc1d..208327dd86 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -300,6 +300,8 @@ CheckIndexCompatible(Oid oldId,
* nonzero to specify a preselected OID for the index.
* 'parentIndexId': the OID of the parent index; InvalidOid if not the child
* of a partitioned index.
+ * 'parentConstraintId': the OID of the parent constraint; InvalidOid if not
+ * the child of a constraint (only used when recursing)
* 'is_alter_table': this is due to an ALTER rather than a CREATE operation.
* 'check_rights': check for CREATE rights in namespace and tablespace. (This
* should be true except when ALTER is deleting/recreating an index.)
@@ -316,6 +318,7 @@ DefineIndex(Oid relationId,
IndexStmt *stmt,
Oid indexRelationId,
Oid parentIndexId,
+ Oid parentConstraintId,
bool is_alter_table,
bool check_rights,
bool check_not_in_use,
@@ -330,6 +333,7 @@ DefineIndex(Oid relationId,
Oid accessMethodId;
Oid namespaceId;
Oid tablespaceId;
+ Oid createdConstraintId;
List *indexColNames;
Relation rel;
Relation indexRelation;
@@ -427,20 +431,11 @@ DefineIndex(Oid relationId,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot create index on partitioned table \"%s\" concurrently",
RelationGetRelationName(rel))));
- if (stmt->unique)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot create unique index on partitioned table \"%s\"",
- RelationGetRelationName(rel))));
if (stmt->excludeOpNames)
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot create exclusion constraints on partitioned table \"%s\"",
RelationGetRelationName(rel))));
- if (stmt->primary || stmt->isconstraint)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot create constraints on partitioned tables")));
}
/*
@@ -638,6 +633,68 @@ DefineIndex(Oid relationId,
index_check_primary_key(rel, indexInfo, is_alter_table);
/*
+ * If this table is partitioned and we're creating a unique index or a
+ * primary key, make sure that the indexed columns are part of the
+ * partition key. Otherwise it would be possible to violate uniqueness by
+ * putting values that ought to be unique in different partitions.
+ *
+ * We could lift this limitation if we had global indexes, but those have
+ * their own problems, so this is a useful feature combination.
+ */
+ if (partitioned && (stmt->unique || stmt->primary))
+ {
+ PartitionKey key = rel->rd_partkey;
+ int i;
+
+ /*
+ * A partitioned table can have unique indexes, as long as all the
+ * columns in the partition key appear in the unique key. A
+ * partition-local index can enforce global uniqueness iff the PK
+ * value completely determines the partition that a row is in.
+ *
+ * Thus, verify that all the columns in the partition key appear
+ * in the unique key definition.
+ */
+ for (i = 0; i < key->partnatts; i++)
+ {
+ bool found = false;
+ int j;
+
+ /*
+ * It may be possible to support UNIQUE constraints when partition
+ * keys are expressions, but is it worth it? Give up for now.
+ */
+ if (key->partattrs[i] == 0)
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ /* XXX reformulate error message? */
+ errmsg("UNIQUE constraints are not supported on partitioned tables using expressions as partition keys")));
+
+ for (j = 0; j < indexInfo->ii_NumIndexAttrs; j++)
+ {
+ if (key->partattrs[i] == indexInfo->ii_KeyAttrNumbers[j])
+ {
+ found = true;
+ break;
+ }
+ }
+ if (!found)
+ {
+ Form_pg_attribute att;
+
+ att = TupleDescAttr(RelationGetDescr(rel), key->partattrs[i] - 1);
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("insufficient columns in UNIQUE constraint definition"),
+ errdetail("UNIQUE constraint on table \"%s\" does not include column \"%s\" which is part of the partition key.",
+ RelationGetRelationName(rel),
+ NameStr(att->attname))));
+ }
+ }
+ }
+
+
+ /*
* We disallow indexes on system columns other than OID. They would not
* necessarily get updated correctly, and they don't seem useful anyway.
*/
@@ -734,12 +791,14 @@ DefineIndex(Oid relationId,
indexRelationId =
index_create(rel, indexRelationName, indexRelationId, parentIndexId,
+ parentConstraintId,
stmt->oldNode, indexInfo, indexColNames,
accessMethodId, tablespaceId,
collationObjectId, classObjectId,
coloptions, reloptions,
flags, constr_flags,
- allowSystemTableMods, !check_rights);
+ allowSystemTableMods, !check_rights,
+ &createdConstraintId);
ObjectAddressSet(address, RelationRelationId, indexRelationId);
@@ -853,6 +912,7 @@ DefineIndex(Oid relationId,
DefineIndex(childRelid, childStmt,
InvalidOid, /* no predefined OID */
indexRelationId, /* this is our child */
+ createdConstraintId,
false, check_rights, check_not_in_use,
false, quiet);
}
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index c8575ea17b..939e57140d 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -938,17 +938,20 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
Relation idxRel = index_open(lfirst_oid(cell), AccessShareLock);
AttrNumber *attmap;
IndexStmt *idxstmt;
+ Oid constraintOid;
attmap = convert_tuples_by_name_map(RelationGetDescr(rel),
RelationGetDescr(parent),
gettext_noop("could not convert row type"));
idxstmt =
generateClonedIndexStmt(NULL, RelationGetRelid(rel), idxRel,
- attmap, RelationGetDescr(rel)->natts);
+ attmap, RelationGetDescr(rel)->natts,
+ &constraintOid);
DefineIndex(RelationGetRelid(rel),
idxstmt,
InvalidOid,
RelationGetRelid(idxRel),
+ constraintOid,
false, false, false, false, false);
index_close(idxRel, AccessShareLock);
@@ -6805,6 +6808,7 @@ ATExecAddIndex(AlteredTableInfo *tab, Relation rel,
stmt,
InvalidOid, /* no predefined OID */
InvalidOid, /* no parent index */
+ InvalidOid, /* no parent constraint */
true, /* is_alter_table */
check_rights,
false, /* check_not_in_use - we did it already */
@@ -6897,6 +6901,7 @@ ATExecAddIndexConstraint(AlteredTableInfo *tab, Relation rel,
address = index_constraint_create(rel,
index_oid,
+ InvalidOid,
indexInfo,
constraintName,
constraintType,
@@ -14174,12 +14179,15 @@ AttachPartitionEnsureIndexes(Relation rel, Relation attachrel)
if (!found)
{
IndexStmt *stmt;
+ Oid constraintOid;
stmt = generateClonedIndexStmt(NULL, RelationGetRelid(attachrel),
idxRel, attmap,
- RelationGetDescr(rel)->natts);
+ RelationGetDescr(rel)->natts,
+ &constraintOid);
DefineIndex(RelationGetRelid(attachrel), stmt, InvalidOid,
RelationGetRelid(idxRel),
+ constraintOid,
false, false, false, false, false);
}
diff --git a/src/backend/parser/analyze.c b/src/backend/parser/analyze.c
index e7b2bc7e73..5b3a610cf9 100644
--- a/src/backend/parser/analyze.c
+++ b/src/backend/parser/analyze.c
@@ -1017,6 +1017,13 @@ transformOnConflictClause(ParseState *pstate,
TargetEntry *te;
int attno;
+ if (targetrel->rd_partdesc)
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("%s cannot be applied to partitioned table \"%s\"",
+ "ON CONFLICT DO UPDATE",
+ RelationGetRelationName(targetrel))));
+
/*
* All INSERT expressions have been parsed, get ready for potentially
* existing SET statements that need to be processed like an UPDATE.
diff --git a/src/backend/parser/parse_utilcmd.c b/src/backend/parser/parse_utilcmd.c
index 90bb356df8..e1ded02ecc 100644
--- a/src/backend/parser/parse_utilcmd.c
+++ b/src/backend/parser/parse_utilcmd.c
@@ -704,12 +704,6 @@ transformColumnDefinition(CreateStmtContext *cxt, ColumnDef *column)
errmsg("primary key constraints are not supported on foreign tables"),
parser_errposition(cxt->pstate,
constraint->location)));
- if (cxt->ispartitioned)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("primary key constraints are not supported on partitioned tables"),
- parser_errposition(cxt->pstate,
- constraint->location)));
/* FALL THRU */
case CONSTR_UNIQUE:
@@ -719,12 +713,6 @@ transformColumnDefinition(CreateStmtContext *cxt, ColumnDef *column)
errmsg("unique constraints are not supported on foreign tables"),
parser_errposition(cxt->pstate,
constraint->location)));
- if (cxt->ispartitioned)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("unique constraints are not supported on partitioned tables"),
- parser_errposition(cxt->pstate,
- constraint->location)));
if (constraint->keys == NIL)
constraint->keys = list_make1(makeString(column->colname));
cxt->ixconstraints = lappend(cxt->ixconstraints, constraint);
@@ -821,12 +809,6 @@ transformTableConstraint(CreateStmtContext *cxt, Constraint *constraint)
errmsg("primary key constraints are not supported on foreign tables"),
parser_errposition(cxt->pstate,
constraint->location)));
- if (cxt->ispartitioned)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("primary key constraints are not supported on partitioned tables"),
- parser_errposition(cxt->pstate,
- constraint->location)));
cxt->ixconstraints = lappend(cxt->ixconstraints, constraint);
break;
@@ -837,12 +819,6 @@ transformTableConstraint(CreateStmtContext *cxt, Constraint *constraint)
errmsg("unique constraints are not supported on foreign tables"),
parser_errposition(cxt->pstate,
constraint->location)));
- if (cxt->ispartitioned)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("unique constraints are not supported on partitioned tables"),
- parser_errposition(cxt->pstate,
- constraint->location)));
cxt->ixconstraints = lappend(cxt->ixconstraints, constraint);
break;
@@ -1184,7 +1160,7 @@ transformTableLikeClause(CreateStmtContext *cxt, TableLikeClause *table_like_cla
/* Build CREATE INDEX statement to recreate the parent_index */
index_stmt = generateClonedIndexStmt(cxt->relation, InvalidOid,
parent_index,
- attmap, tupleDesc->natts);
+ attmap, tupleDesc->natts, NULL);
/* Copy comment on index, if requested */
if (table_like_clause->options & CREATE_TABLE_LIKE_COMMENTS)
@@ -1267,7 +1243,7 @@ transformOfType(CreateStmtContext *cxt, TypeName *ofTypename)
*/
IndexStmt *
generateClonedIndexStmt(RangeVar *heapRel, Oid heapRelid, Relation source_idx,
- const AttrNumber *attmap, int attmap_length)
+ const AttrNumber *attmap, int attmap_length, Oid *constraintOid)
{
Oid source_relid = RelationGetRelid(source_idx);
HeapTuple ht_idxrel;
@@ -1365,6 +1341,9 @@ generateClonedIndexStmt(RangeVar *heapRel, Oid heapRelid, Relation source_idx,
HeapTuple ht_constr;
Form_pg_constraint conrec;
+ if (constraintOid)
+ *constraintOid = constraintId;
+
ht_constr = SearchSysCache1(CONSTROID,
ObjectIdGetDatum(constraintId));
if (!HeapTupleIsValid(ht_constr))
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 1a29dda783..fa09724277 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1337,6 +1337,7 @@ ProcessUtilitySlow(ParseState *pstate,
stmt,
InvalidOid, /* no predefined OID */
InvalidOid, /* no parent index */
+ InvalidOid, /* no parent constraint */
false, /* is_alter_table */
true, /* check_rights */
true, /* check_not_in_use */
diff --git a/src/include/catalog/index.h b/src/include/catalog/index.h
index 8ff3a0732e..9d501fe944 100644
--- a/src/include/catalog/index.h
+++ b/src/include/catalog/index.h
@@ -54,6 +54,7 @@ extern Oid index_create(Relation heapRelation,
const char *indexRelationName,
Oid indexRelationId,
Oid parentIndexRelid,
+ Oid parentConstraintId,
Oid relFileNode,
IndexInfo *indexInfo,
List *indexColNames,
@@ -66,7 +67,8 @@ extern Oid index_create(Relation heapRelation,
bits16 flags,
bits16 constr_flags,
bool allow_system_table_mods,
- bool is_internal);
+ bool is_internal,
+ Oid *constraintId);
#define INDEX_CONSTR_CREATE_MARK_AS_PRIMARY (1 << 0)
#define INDEX_CONSTR_CREATE_DEFERRABLE (1 << 1)
@@ -76,6 +78,7 @@ extern Oid index_create(Relation heapRelation,
extern ObjectAddress index_constraint_create(Relation heapRelation,
Oid indexRelationId,
+ Oid parentConstraintId,
IndexInfo *indexInfo,
const char *constraintName,
char constraintType,
diff --git a/src/include/commands/defrem.h b/src/include/commands/defrem.h
index 41007162aa..d1876d0b08 100644
--- a/src/include/commands/defrem.h
+++ b/src/include/commands/defrem.h
@@ -26,6 +26,7 @@ extern ObjectAddress DefineIndex(Oid relationId,
IndexStmt *stmt,
Oid indexRelationId,
Oid parentIndexId,
+ Oid parentConstraintId,
bool is_alter_table,
bool check_rights,
bool check_not_in_use,
diff --git a/src/include/parser/parse_utilcmd.h b/src/include/parser/parse_utilcmd.h
index 64aa8234e5..35ac97940a 100644
--- a/src/include/parser/parse_utilcmd.h
+++ b/src/include/parser/parse_utilcmd.h
@@ -29,6 +29,7 @@ extern PartitionBoundSpec *transformPartitionBound(ParseState *pstate, Relation
PartitionBoundSpec *spec);
extern IndexStmt *generateClonedIndexStmt(RangeVar *heapRel, Oid heapOid,
Relation source_idx,
- const AttrNumber *attmap, int attmap_length);
+ const AttrNumber *attmap, int attmap_length,
+ Oid *constraintOid);
#endif /* PARSE_UTILCMD_H */
diff --git a/src/test/regress/expected/alter_table.out b/src/test/regress/expected/alter_table.out
index 517fb080bd..2caf930242 100644
--- a/src/test/regress/expected/alter_table.out
+++ b/src/test/regress/expected/alter_table.out
@@ -3290,14 +3290,6 @@ CREATE TABLE partitioned (
a int,
b int
) PARTITION BY RANGE (a, (a+b+1));
-ALTER TABLE partitioned ADD UNIQUE (a);
-ERROR: unique constraints are not supported on partitioned tables
-LINE 1: ALTER TABLE partitioned ADD UNIQUE (a);
- ^
-ALTER TABLE partitioned ADD PRIMARY KEY (a);
-ERROR: primary key constraints are not supported on partitioned tables
-LINE 1: ALTER TABLE partitioned ADD PRIMARY KEY (a);
- ^
ALTER TABLE partitioned ADD FOREIGN KEY (a) REFERENCES blah;
ERROR: foreign key constraints are not supported on partitioned tables
LINE 1: ALTER TABLE partitioned ADD FOREIGN KEY (a) REFERENCES blah;
diff --git a/src/test/regress/expected/create_table.out b/src/test/regress/expected/create_table.out
index 8e745402ae..866cc99b9f 100644
--- a/src/test/regress/expected/create_table.out
+++ b/src/test/regress/expected/create_table.out
@@ -276,12 +276,6 @@ CREATE TABLE partitioned (
) PARTITION BY LIST (a1, a2); -- fail
ERROR: cannot use "list" partition strategy with more than one column
-- unsupported constraint type for partitioned tables
-CREATE TABLE partitioned (
- a int PRIMARY KEY
-) PARTITION BY RANGE (a);
-ERROR: primary key constraints are not supported on partitioned tables
-LINE 2: a int PRIMARY KEY
- ^
CREATE TABLE pkrel (
a int PRIMARY KEY
);
@@ -293,12 +287,6 @@ LINE 2: a int REFERENCES pkrel(a)
^
DROP TABLE pkrel;
CREATE TABLE partitioned (
- a int UNIQUE
-) PARTITION BY RANGE (a);
-ERROR: unique constraints are not supported on partitioned tables
-LINE 2: a int UNIQUE
- ^
-CREATE TABLE partitioned (
a int,
EXCLUDE USING gist (a WITH &&)
) PARTITION BY RANGE (a);
diff --git a/src/test/regress/expected/indexing.out b/src/test/regress/expected/indexing.out
index da5f557b72..791ebfc07e 100644
--- a/src/test/regress/expected/indexing.out
+++ b/src/test/regress/expected/indexing.out
@@ -26,8 +26,6 @@ drop table idxpart;
-- Some unsupported features
create table idxpart (a int, b int, c text) partition by range (a);
create table idxpart1 partition of idxpart for values from (0) to (10);
-create unique index on idxpart (a);
-ERROR: cannot create unique index on partitioned table "idxpart"
create index concurrently on idxpart (a);
ERROR: cannot create index on partitioned table "idxpart" concurrently
drop table idxpart;
@@ -591,6 +589,146 @@ select attrelid::regclass, attname, attnum from pg_attribute
(7 rows)
drop table idxpart;
+--
+-- Constraint-related indexes
+--
+-- Verify that it works to add primary key / unique to partitioned tables
+create table idxpart (a int primary key, b int) partition by range (a);
+\d idxpart
+ Table "public.idxpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | not null |
+ b | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "idxpart_pkey" PRIMARY KEY, btree (a)
+Number of partitions: 0
+
+drop table idxpart;
+-- but not if you fail to use the full partition key
+create table idxpart (a int unique, b int) partition by range (a, b);
+ERROR: insufficient columns in UNIQUE constraint definition
+DETAIL: UNIQUE constraint on table "idxpart" does not include column "b" which is part of the partition key.
+create table idxpart (a int, b int unique) partition by range (a, b);
+ERROR: insufficient columns in UNIQUE constraint definition
+DETAIL: UNIQUE constraint on table "idxpart" does not include column "a" which is part of the partition key.
+create table idxpart (a int primary key, b int) partition by range (b, a);
+ERROR: insufficient columns in UNIQUE constraint definition
+DETAIL: UNIQUE constraint on table "idxpart" does not include column "b" which is part of the partition key.
+create table idxpart (a int, b int primary key) partition by range (b, a);
+ERROR: insufficient columns in UNIQUE constraint definition
+DETAIL: UNIQUE constraint on table "idxpart" does not include column "a" which is part of the partition key.
+-- OK if you use them in some other order
+create table idxpart (a int, b int, c text, primary key (a, b, c)) partition by range (b, c, a);
+drop table idxpart;
+create table idxpart (a int primary key, b int) partition by range ((b + a));
+ERROR: UNIQUE constraints are not supported on partitioned tables using expressions as partition keys
+-- not other types of index-based constraints
+create table idxpart (a int, exclude (a with = )) partition by range (a);
+ERROR: exclusion constraints are not supported on partitioned tables
+LINE 1: create table idxpart (a int, exclude (a with = )) partition ...
+ ^
+-- It works to add primary keys after the partitioned table is created
+create table idxpart (a int, b int, c text) partition by range (a, b);
+alter table idxpart add primary key (a); -- not an incomplete one tho
+ERROR: insufficient columns in UNIQUE constraint definition
+DETAIL: UNIQUE constraint on table "idxpart" does not include column "b" which is part of the partition key.
+alter table idxpart add primary key (a, b);
+\d idxpart
+ Table "public.idxpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | not null |
+ b | integer | | not null |
+ c | text | | |
+Partition key: RANGE (a, b)
+Indexes:
+ "idxpart_pkey" PRIMARY KEY, btree (a, b)
+Number of partitions: 0
+
+create table idxpart1 partition of idxpart for values from (0, 0) to (1000, 1000);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | not null |
+ b | integer | | not null |
+ c | text | | |
+Partition of: idxpart FOR VALUES FROM (0, 0) TO (1000, 1000)
+Indexes:
+ "idxpart1_pkey" PRIMARY KEY, btree (a, b)
+
+drop table idxpart;
+-- It works to add unique constraints after the partitioned table is created
+create table idxpart (a int, b int) partition by range (a, b);
+alter table idxpart add unique (a); -- ... nope
+ERROR: insufficient columns in UNIQUE constraint definition
+DETAIL: UNIQUE constraint on table "idxpart" does not include column "b" which is part of the partition key.
+alter table idxpart add unique (b, a);
+\d idxpart
+ Table "public.idxpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | integer | | |
+Partition key: RANGE (a, b)
+Indexes:
+ "idxpart_b_a_key" UNIQUE CONSTRAINT, btree (b, a)
+Number of partitions: 0
+
+drop table idxpart;
+-- Exclusion constraints cannot be added
+create table idxpart (a int, b int) partition by range (a);
+alter table idxpart add exclude (a with =);
+ERROR: exclusion constraints are not supported on partitioned tables
+LINE 1: alter table idxpart add exclude (a with =);
+ ^
+drop table idxpart;
+-- When (sub)partitions are created, they also contain the constraint
+create table idxpart (a int, b int, primary key (a, b)) partition by range (a, b);
+create table idxpart1 partition of idxpart for values from (1, 1) to (10, 10);
+create table idxpart2 partition of idxpart for values from (10, 10) to (20, 20)
+ partition by range (b);
+create table idxpart21 partition of idxpart2 for values from (10) to (15);
+create table idxpart22 partition of idxpart2 for values from (15) to (20);
+create table idxpart3 (b int not null, a int not null);
+alter table idxpart attach partition idxpart3 for values from (20, 20) to (30, 30);
+select conname, contype, conrelid::regclass, conindid::regclass, conkey
+ from pg_constraint where conrelid::regclass::text like 'idxpart%'
+ order by conname;
+ conname | contype | conrelid | conindid | conkey
+----------------+---------+-----------+----------------+--------
+ idxpart1_pkey | p | idxpart1 | idxpart1_pkey | {1,2}
+ idxpart21_pkey | p | idxpart21 | idxpart21_pkey | {1,2}
+ idxpart22_pkey | p | idxpart22 | idxpart22_pkey | {1,2}
+ idxpart2_pkey | p | idxpart2 | idxpart2_pkey | {1,2}
+ idxpart3_pkey | p | idxpart3 | idxpart3_pkey | {2,1}
+ idxpart_pkey | p | idxpart | idxpart_pkey | {1,2}
+(6 rows)
+
+drop table idxpart;
+-- multi-layer partitioning honors the prohibition. So this fails:
+create table idxpart (a int, b int, primary key (a)) partition by range (a);
+create table idxpart2 partition of idxpart for values from (0) to (1000) partition by range (b);
+ERROR: insufficient columns in UNIQUE constraint definition
+DETAIL: UNIQUE constraint on table "idxpart2" does not include column "b" which is part of the partition key.
+drop table idxpart;
+-- but this works:
+create table idxpart (a int, b int, primary key (a, b)) partition by range (a);
+create table idxpart2 partition of idxpart for values from (0) to (1000) partition by range (b);
+create table idxpart21 partition of idxpart2 for values from (0) to (1000);
+select conname, contype, conrelid::regclass, conindid::regclass, conkey
+ from pg_constraint where conrelid::regclass::text like 'idxpart%'
+ order by conname;
+ conname | contype | conrelid | conindid | conkey
+----------------+---------+-----------+----------------+--------
+ idxpart21_pkey | p | idxpart21 | idxpart21_pkey | {1,2}
+ idxpart2_pkey | p | idxpart2 | idxpart2_pkey | {1,2}
+ idxpart_pkey | p | idxpart | idxpart_pkey | {1,2}
+(3 rows)
+
+drop table idxpart;
-- intentionally leave some objects around
create table idxpart (a int) partition by range (a);
create table idxpart1 partition of idxpart for values from (0) to (100);
diff --git a/src/test/regress/expected/insert_conflict.out b/src/test/regress/expected/insert_conflict.out
index 8fd2027d6a..2650faedee 100644
--- a/src/test/regress/expected/insert_conflict.out
+++ b/src/test/regress/expected/insert_conflict.out
@@ -794,7 +794,7 @@ insert into parted_conflict_test values (1, 'a') on conflict do nothing;
insert into parted_conflict_test values (1, 'a') on conflict do nothing;
-- however, on conflict do update is not supported yet
insert into parted_conflict_test values (1) on conflict (b) do update set a = excluded.a;
-ERROR: there is no unique or exclusion constraint matching the ON CONFLICT specification
+ERROR: ON CONFLICT DO UPDATE cannot be applied to partitioned table "parted_conflict_test"
-- but it works OK if we target the partition directly
insert into parted_conflict_test_1 values (1) on conflict (b) do
update set a = excluded.a;
diff --git a/src/test/regress/sql/alter_table.sql b/src/test/regress/sql/alter_table.sql
index af25ee9e77..ed0bb7845b 100644
--- a/src/test/regress/sql/alter_table.sql
+++ b/src/test/regress/sql/alter_table.sql
@@ -2016,8 +2016,6 @@ CREATE TABLE partitioned (
a int,
b int
) PARTITION BY RANGE (a, (a+b+1));
-ALTER TABLE partitioned ADD UNIQUE (a);
-ALTER TABLE partitioned ADD PRIMARY KEY (a);
ALTER TABLE partitioned ADD FOREIGN KEY (a) REFERENCES blah;
ALTER TABLE partitioned ADD EXCLUDE USING gist (a WITH &&);
diff --git a/src/test/regress/sql/create_table.sql b/src/test/regress/sql/create_table.sql
index 8f9991ef18..fefccf21a2 100644
--- a/src/test/regress/sql/create_table.sql
+++ b/src/test/regress/sql/create_table.sql
@@ -294,10 +294,6 @@ CREATE TABLE partitioned (
) PARTITION BY LIST (a1, a2); -- fail
-- unsupported constraint type for partitioned tables
-CREATE TABLE partitioned (
- a int PRIMARY KEY
-) PARTITION BY RANGE (a);
-
CREATE TABLE pkrel (
a int PRIMARY KEY
);
@@ -307,10 +303,6 @@ CREATE TABLE partitioned (
DROP TABLE pkrel;
CREATE TABLE partitioned (
- a int UNIQUE
-) PARTITION BY RANGE (a);
-
-CREATE TABLE partitioned (
a int,
EXCLUDE USING gist (a WITH &&)
) PARTITION BY RANGE (a);
diff --git a/src/test/regress/sql/indexing.sql b/src/test/regress/sql/indexing.sql
index 2380c7cdab..25ed3cbb07 100644
--- a/src/test/regress/sql/indexing.sql
+++ b/src/test/regress/sql/indexing.sql
@@ -15,7 +15,6 @@ drop table idxpart;
-- Some unsupported features
create table idxpart (a int, b int, c text) partition by range (a);
create table idxpart1 partition of idxpart for values from (0) to (10);
-create unique index on idxpart (a);
create index concurrently on idxpart (a);
drop table idxpart;
@@ -270,6 +269,78 @@ select attrelid::regclass, attname, attnum from pg_attribute
order by attrelid::regclass, attnum;
drop table idxpart;
+--
+-- Constraint-related indexes
+--
+
+-- Verify that it works to add primary key / unique to partitioned tables
+create table idxpart (a int primary key, b int) partition by range (a);
+\d idxpart
+drop table idxpart;
+
+-- but not if you fail to use the full partition key
+create table idxpart (a int unique, b int) partition by range (a, b);
+create table idxpart (a int, b int unique) partition by range (a, b);
+create table idxpart (a int primary key, b int) partition by range (b, a);
+create table idxpart (a int, b int primary key) partition by range (b, a);
+
+-- OK if you use them in some other order
+create table idxpart (a int, b int, c text, primary key (a, b, c)) partition by range (b, c, a);
+drop table idxpart;
+
+create table idxpart (a int primary key, b int) partition by range ((b + a));
+-- not other types of index-based constraints
+create table idxpart (a int, exclude (a with = )) partition by range (a);
+
+-- It works to add primary keys after the partitioned table is created
+create table idxpart (a int, b int, c text) partition by range (a, b);
+alter table idxpart add primary key (a); -- not an incomplete one tho
+alter table idxpart add primary key (a, b);
+\d idxpart
+create table idxpart1 partition of idxpart for values from (0, 0) to (1000, 1000);
+\d idxpart1
+drop table idxpart;
+
+-- It works to add unique constraints after the partitioned table is created
+create table idxpart (a int, b int) partition by range (a, b);
+alter table idxpart add unique (a); -- ... nope
+alter table idxpart add unique (b, a);
+\d idxpart
+drop table idxpart;
+
+-- Exclusion constraints cannot be added
+create table idxpart (a int, b int) partition by range (a);
+alter table idxpart add exclude (a with =);
+drop table idxpart;
+
+-- When (sub)partitions are created, they also contain the constraint
+create table idxpart (a int, b int, primary key (a, b)) partition by range (a, b);
+create table idxpart1 partition of idxpart for values from (1, 1) to (10, 10);
+create table idxpart2 partition of idxpart for values from (10, 10) to (20, 20)
+ partition by range (b);
+create table idxpart21 partition of idxpart2 for values from (10) to (15);
+create table idxpart22 partition of idxpart2 for values from (15) to (20);
+create table idxpart3 (b int not null, a int not null);
+alter table idxpart attach partition idxpart3 for values from (20, 20) to (30, 30);
+select conname, contype, conrelid::regclass, conindid::regclass, conkey
+ from pg_constraint where conrelid::regclass::text like 'idxpart%'
+ order by conname;
+drop table idxpart;
+
+-- multi-layer partitioning honors the prohibition. So this fails:
+create table idxpart (a int, b int, primary key (a)) partition by range (a);
+create table idxpart2 partition of idxpart for values from (0) to (1000) partition by range (b);
+drop table idxpart;
+
+-- but this works:
+create table idxpart (a int, b int, primary key (a, b)) partition by range (a);
+create table idxpart2 partition of idxpart for values from (0) to (1000) partition by range (b);
+create table idxpart21 partition of idxpart2 for values from (0) to (1000);
+select conname, contype, conrelid::regclass, conindid::regclass, conkey
+ from pg_constraint where conrelid::regclass::text like 'idxpart%'
+ order by conname;
+drop table idxpart;
+
-- intentionally leave some objects around
create table idxpart (a int) partition by range (a);
create table idxpart1 partition of idxpart for values from (0) to (100);
--
2.11.0
Re: unique indexes on partitioned tables
Version 4 of this patch, rebased on today's master.
The main change is in dependency handling for the constraints: you now
can't drop a constraint from a partition, if it's attached to a
constraint in the parent (you can't drop indexes from under the
constraints either, but that was true in previous versions too). Also
some error message rewording. I added a bunch of additional tests.
I implemented the dependencies using pg_depend entries. However,
pg_constraint has the notion of "coninhcount" and "conislocal", so I
update those values for the partition's pg_constraint row, because the
error messages are nicer that way. We could remove those lines from the
patch and the mechanics should be pretty much identical.
I'll review the doc additions, per Simon upthread.
--
�lvaro Herrera https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
Re: unique indexes on partitioned tables
Alvaro Herrera wrote:
Version 4 of this patch, rebased on today's master.
--
�lvaro Herrera https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
Attachments:
v4-0001-allow-indexes-on-partitioned-tables-to-be-unique.patchtext/plain; charset=us-asciiDownload
From 1a02e7f359c94e5db0bb069666b950775cd3e2af Mon Sep 17 00:00:00 2001
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
Date: Mon, 6 Nov 2017 17:04:55 +0100
Subject: [PATCH v4] allow indexes on partitioned tables to be unique
---
doc/src/sgml/ref/alter_table.sgml | 9 +-
doc/src/sgml/ref/create_table.sgml | 16 +-
src/backend/bootstrap/bootparse.y | 2 +
src/backend/catalog/index.c | 45 ++++-
src/backend/catalog/pg_constraint.c | 76 ++++++++
src/backend/catalog/toasting.c | 4 +-
src/backend/commands/indexcmds.c | 125 +++++++++++--
src/backend/commands/tablecmds.c | 62 ++++++-
src/backend/parser/analyze.c | 7 +
src/backend/parser/parse_utilcmd.c | 31 +---
src/backend/tcop/utility.c | 1 +
src/include/catalog/index.h | 5 +-
src/include/catalog/pg_constraint_fn.h | 4 +-
src/include/commands/defrem.h | 1 +
src/include/parser/parse_utilcmd.h | 3 +-
src/test/regress/expected/alter_table.out | 8 -
src/test/regress/expected/create_table.out | 12 --
src/test/regress/expected/indexing.out | 254 +++++++++++++++++++++++++-
src/test/regress/expected/insert_conflict.out | 2 +-
src/test/regress/sql/alter_table.sql | 2 -
src/test/regress/sql/create_table.sql | 8 -
src/test/regress/sql/indexing.sql | 151 ++++++++++++++-
22 files changed, 740 insertions(+), 88 deletions(-)
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index 286c7a8589..c00fd09fe1 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -804,8 +804,9 @@ ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
This form attaches an existing table (which might itself be partitioned)
as a partition of the target table. The table can be attached
as a partition for specific values using <literal>FOR VALUES
- </literal> or as a default partition by using <literal>DEFAULT
- </literal>. For each index in the target table, a corresponding
+ </literal> or as a default partition by using
+ <literal>DEFAULT</literal>.
+ For each index in the target table, a corresponding
one will be created in the attached table; or, if an equivalent
index already exists, will be attached to the target table's index,
as if <command>ALTER INDEX ATTACH PARTITION</command> had been executed.
@@ -820,8 +821,10 @@ ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
as the target table and no more; moreover, the column types must also
match. Also, it must have all the <literal>NOT NULL</literal> and
<literal>CHECK</literal> constraints of the target table. Currently
- <literal>UNIQUE</literal>, <literal>PRIMARY KEY</literal>, and
<literal>FOREIGN KEY</literal> constraints are not considered.
+ <literal>UNIQUE</literal> and <literal>PRIMARY KEY</literal> constraints
+ from the parent table will be created in the partition, if they don't
+ already exist.
If any of the <literal>CHECK</literal> constraints of the table being
attached is marked <literal>NO INHERIT</literal>, the command will fail;
such a constraint must be recreated without the <literal>NO INHERIT</literal>
diff --git a/doc/src/sgml/ref/create_table.sgml b/doc/src/sgml/ref/create_table.sgml
index a0c9a6d257..4c56df8960 100644
--- a/doc/src/sgml/ref/create_table.sgml
+++ b/doc/src/sgml/ref/create_table.sgml
@@ -546,8 +546,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</para>
<para>
- Partitioned tables do not support <literal>UNIQUE</literal>,
- <literal>PRIMARY KEY</literal>, <literal>EXCLUDE</literal>, or
+ Partitioned tables do not support
+ <literal>EXCLUDE</literal>, or
<literal>FOREIGN KEY</literal> constraints; however, you can define
these constraints on individual partitions.
</para>
@@ -786,6 +786,11 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
primary key constraint defined for the table. (Otherwise it
would just be the same constraint listed twice.)
</para>
+
+ <para>
+ When used on partitioned tables, <literal>UNIQUE</literal> constraints
+ must include all the columns of the partition key.
+ </para>
</listitem>
</varlistentry>
@@ -814,6 +819,13 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
about the design of the schema, since a primary key implies that other
tables can rely on this set of columns as a unique identifier for rows.
</para>
+
+ <para>
+ <literal>PRIMARY KEY</literal> constraints share the restrictions that
+ <literal>UNIQUE</literal> constraints have when placed on partitioned
+ tables.
+ </para>
+
</listitem>
</varlistentry>
diff --git a/src/backend/bootstrap/bootparse.y b/src/backend/bootstrap/bootparse.y
index dfd53fa054..9e81f9514d 100644
--- a/src/backend/bootstrap/bootparse.y
+++ b/src/backend/bootstrap/bootparse.y
@@ -322,6 +322,7 @@ Boot_DeclareIndexStmt:
stmt,
$4,
InvalidOid,
+ InvalidOid,
false,
false,
false,
@@ -367,6 +368,7 @@ Boot_DeclareUniqueIndexStmt:
stmt,
$5,
InvalidOid,
+ InvalidOid,
false,
false,
false,
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 849a469127..1660711fb0 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -690,6 +690,8 @@ UpdateIndexRelation(Oid indexoid,
* nonzero to specify a preselected OID.
* parentIndexRelid: if creating an index partition, the OID of the
* parent index; otherwise InvalidOid.
+ * parentConstraintId: if creating a constraint on a partition, the OID
+ * of the constraint in the parent; otherwise InvalidOid.
* relFileNode: normally, pass InvalidOid to get new storage. May be
* nonzero to attach an existing valid build.
* indexInfo: same info executor uses to insert into the index
@@ -721,6 +723,7 @@ UpdateIndexRelation(Oid indexoid,
* (only if INDEX_CREATE_ADD_CONSTRAINT is set)
* allow_system_table_mods: allow table to be a system catalog
* is_internal: if true, post creation hook for new index
+ * constraintId: if not NULL, receives OID of created constraint
*
* Returns the OID of the created index.
*/
@@ -729,6 +732,7 @@ index_create(Relation heapRelation,
const char *indexRelationName,
Oid indexRelationId,
Oid parentIndexRelid,
+ Oid parentConstraintId,
Oid relFileNode,
IndexInfo *indexInfo,
List *indexColNames,
@@ -741,7 +745,8 @@ index_create(Relation heapRelation,
bits16 flags,
bits16 constr_flags,
bool allow_system_table_mods,
- bool is_internal)
+ bool is_internal,
+ Oid *constraintId)
{
Oid heapRelationId = RelationGetRelid(heapRelation);
Relation pg_class;
@@ -988,6 +993,7 @@ index_create(Relation heapRelation,
if ((flags & INDEX_CREATE_ADD_CONSTRAINT) != 0)
{
char constraintType;
+ ObjectAddress localaddr;
if (isprimary)
constraintType = CONSTRAINT_PRIMARY;
@@ -1001,14 +1007,17 @@ index_create(Relation heapRelation,
constraintType = 0; /* keep compiler quiet */
}
- index_constraint_create(heapRelation,
+ localaddr = index_constraint_create(heapRelation,
indexRelationId,
+ parentConstraintId,
indexInfo,
indexRelationName,
constraintType,
constr_flags,
allow_system_table_mods,
is_internal);
+ if (constraintId)
+ *constraintId = localaddr.objectId;
}
else
{
@@ -1179,6 +1188,8 @@ index_create(Relation heapRelation,
*
* heapRelation: table owning the index (must be suitably locked by caller)
* indexRelationId: OID of the index
+ * parentConstraintId: if constraint is on a partition, the OID of the
+ * constraint in the parent.
* indexInfo: same info executor uses to insert into the index
* constraintName: what it say (generally, should match name of index)
* constraintType: one of CONSTRAINT_PRIMARY, CONSTRAINT_UNIQUE, or
@@ -1196,6 +1207,7 @@ index_create(Relation heapRelation,
ObjectAddress
index_constraint_create(Relation heapRelation,
Oid indexRelationId,
+ Oid parentConstraintId,
IndexInfo *indexInfo,
const char *constraintName,
char constraintType,
@@ -1210,6 +1222,8 @@ index_constraint_create(Relation heapRelation,
bool deferrable;
bool initdeferred;
bool mark_as_primary;
+ bool islocal;
+ int inhcount;
deferrable = (constr_flags & INDEX_CONSTR_CREATE_DEFERRABLE) != 0;
initdeferred = (constr_flags & INDEX_CONSTR_CREATE_INIT_DEFERRED) != 0;
@@ -1244,6 +1258,17 @@ index_constraint_create(Relation heapRelation,
deleteDependencyRecordsForClass(RelationRelationId, indexRelationId,
RelationRelationId, DEPENDENCY_AUTO);
+ if (OidIsValid(parentConstraintId))
+ {
+ islocal = false;
+ inhcount = 1;
+ }
+ else
+ {
+ islocal = true;
+ inhcount = 0;
+ }
+
/*
* Construct a pg_constraint entry.
*/
@@ -1271,8 +1296,8 @@ index_constraint_create(Relation heapRelation,
NULL, /* no check constraint */
NULL,
NULL,
- true, /* islocal */
- 0, /* inhcount */
+ islocal,
+ inhcount,
true, /* noinherit */
is_internal);
@@ -1293,6 +1318,18 @@ index_constraint_create(Relation heapRelation,
recordDependencyOn(&myself, &referenced, DEPENDENCY_INTERNAL);
/*
+ * Also, if this is a constraint on a partition, mark it as depending
+ * on the constraint in the parent.
+ */
+ if (OidIsValid(parentConstraintId))
+ {
+ ObjectAddress third;
+
+ ObjectAddressSet(third, ConstraintRelationId, parentConstraintId);
+ recordDependencyOn(&referenced, &third, DEPENDENCY_INTERNAL_AUTO);
+ }
+
+ /*
* If the constraint is deferrable, create the deferred uniqueness
* checking trigger. (The trigger will be given an internal dependency on
* the constraint by CreateTrigger.)
diff --git a/src/backend/catalog/pg_constraint.c b/src/backend/catalog/pg_constraint.c
index 442ae7e23d..731c5e4317 100644
--- a/src/backend/catalog/pg_constraint.c
+++ b/src/backend/catalog/pg_constraint.c
@@ -748,6 +748,43 @@ AlterConstraintNamespaces(Oid ownerId, Oid oldNspId,
}
/*
+ * ConstraintSetParentConstraint
+ * Set a partition's constraint as child of its parent table's
+ *
+ * This updates the constraint's pg_constraint row to show it as inherited, and
+ * add a dependency to the parent so that it cannot be removed on its own.
+ */
+void
+ConstraintSetParentConstraint(Oid childConstrId, Oid parentConstrId)
+{
+ Relation constrRel;
+ Form_pg_constraint constrForm;
+ HeapTuple tuple,
+ newtup;
+ ObjectAddress depender;
+ ObjectAddress referenced;
+
+ constrRel = heap_open(ConstraintRelationId, RowExclusiveLock);
+ tuple = SearchSysCache1(CONSTROID, ObjectIdGetDatum(childConstrId));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for constraint %u", childConstrId);
+ newtup = heap_copytuple(tuple);
+ constrForm = (Form_pg_constraint) GETSTRUCT(newtup);
+ constrForm->conislocal = false;
+ constrForm->coninhcount++;
+ CatalogTupleUpdate(constrRel, &tuple->t_self, newtup);
+ ReleaseSysCache(tuple);
+
+ ObjectAddressSet(referenced, ConstraintRelationId, parentConstrId);
+ ObjectAddressSet(depender, ConstraintRelationId, childConstrId);
+
+ recordDependencyOn(&depender, &referenced, DEPENDENCY_INTERNAL_AUTO);
+
+ heap_close(constrRel, RowExclusiveLock);
+}
+
+
+/*
* get_relation_constraint_oid
* Find a constraint on the specified relation with the specified name.
* Returns constraint's OID.
@@ -904,6 +941,45 @@ get_relation_constraint_attnos(Oid relid, const char *conname,
}
/*
+ * Return the OID of the constraint associated with the given index in the
+ * given relation; or InvalidOid if no such index is catalogued.
+ */
+Oid
+get_relation_idx_constraint_oid(Oid relationId, Oid indexId)
+{
+ Relation pg_constraint;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple tuple;
+ Oid constraintId = InvalidOid;
+
+ pg_constraint = heap_open(ConstraintRelationId, AccessShareLock);
+
+ ScanKeyInit(&key,
+ Anum_pg_constraint_conrelid,
+ BTEqualStrategyNumber,
+ F_OIDEQ,
+ ObjectIdGetDatum(relationId));
+ scan = systable_beginscan(pg_constraint, ConstraintRelidIndexId,
+ true, NULL, 1, &key);
+ while ((tuple = systable_getnext(scan)) != NULL)
+ {
+ Form_pg_constraint constrForm;
+
+ constrForm = (Form_pg_constraint) GETSTRUCT(tuple);
+ if (constrForm->conindid == indexId)
+ {
+ constraintId = HeapTupleGetOid(tuple);
+ break;
+ }
+ }
+ systable_endscan(scan);
+
+ heap_close(pg_constraint, AccessShareLock);
+ return constraintId;
+}
+
+/*
* get_domain_constraint_oid
* Find a constraint on the specified domain with the specified name.
* Returns constraint's OID.
diff --git a/src/backend/catalog/toasting.c b/src/backend/catalog/toasting.c
index cf37011b73..f4e7b83fee 100644
--- a/src/backend/catalog/toasting.c
+++ b/src/backend/catalog/toasting.c
@@ -329,13 +329,13 @@ create_toast_table(Relation rel, Oid toastOid, Oid toastIndexOid,
coloptions[1] = 0;
index_create(toast_rel, toast_idxname, toastIndexOid, InvalidOid,
- InvalidOid,
+ InvalidOid, InvalidOid,
indexInfo,
list_make2("chunk_id", "chunk_seq"),
BTREE_AM_OID,
rel->rd_rel->reltablespace,
collationObjectId, classObjectId, coloptions, (Datum) 0,
- INDEX_CREATE_IS_PRIMARY, 0, true, true);
+ INDEX_CREATE_IS_PRIMARY, 0, true, true, NULL);
heap_close(toast_rel, NoLock);
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index a9461a4b06..b0e5ede488 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -25,6 +25,7 @@
#include "catalog/indexing.h"
#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_constraint_fn.h"
#include "catalog/pg_inherits.h"
#include "catalog/pg_inherits_fn.h"
#include "catalog/pg_opclass.h"
@@ -301,6 +302,8 @@ CheckIndexCompatible(Oid oldId,
* nonzero to specify a preselected OID for the index.
* 'parentIndexId': the OID of the parent index; InvalidOid if not the child
* of a partitioned index.
+ * 'parentConstraintId': the OID of the parent constraint; InvalidOid if not
+ * the child of a constraint (only used when recursing)
* 'is_alter_table': this is due to an ALTER rather than a CREATE operation.
* 'check_rights': check for CREATE rights in namespace and tablespace. (This
* should be true except when ALTER is deleting/recreating an index.)
@@ -317,6 +320,7 @@ DefineIndex(Oid relationId,
IndexStmt *stmt,
Oid indexRelationId,
Oid parentIndexId,
+ Oid parentConstraintId,
bool is_alter_table,
bool check_rights,
bool check_not_in_use,
@@ -331,6 +335,7 @@ DefineIndex(Oid relationId,
Oid accessMethodId;
Oid namespaceId;
Oid tablespaceId;
+ Oid createdConstraintId = InvalidOid;
List *indexColNames;
Relation rel;
Relation indexRelation;
@@ -428,20 +433,11 @@ DefineIndex(Oid relationId,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot create index on partitioned table \"%s\" concurrently",
RelationGetRelationName(rel))));
- if (stmt->unique)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot create unique index on partitioned table \"%s\"",
- RelationGetRelationName(rel))));
if (stmt->excludeOpNames)
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot create exclusion constraints on partitioned table \"%s\"",
RelationGetRelationName(rel))));
- if (stmt->primary || stmt->isconstraint)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot create constraints on partitioned tables")));
}
/*
@@ -639,6 +635,84 @@ DefineIndex(Oid relationId,
index_check_primary_key(rel, indexInfo, is_alter_table);
/*
+ * If this table is partitioned and we're creating a unique index or a
+ * primary key, make sure that the indexed columns are part of the
+ * partition key. Otherwise it would be possible to violate uniqueness by
+ * putting values that ought to be unique in different partitions.
+ *
+ * We could lift this limitation if we had global indexes, but those have
+ * their own problems, so this is a useful feature combination.
+ */
+ if (partitioned && (stmt->unique || stmt->primary))
+ {
+ PartitionKey key = rel->rd_partkey;
+ int i;
+
+ /*
+ * A partitioned table can have unique indexes, as long as all the
+ * columns in the partition key appear in the unique key. A
+ * partition-local index can enforce global uniqueness iff the PK
+ * value completely determines the partition that a row is in.
+ *
+ * Thus, verify that all the columns in the partition key appear
+ * in the unique key definition.
+ */
+ for (i = 0; i < key->partnatts; i++)
+ {
+ bool found = false;
+ int j;
+ const char *constraint_type;
+
+ if (stmt->primary)
+ constraint_type = "PRIMARY KEY";
+ else if (stmt->unique)
+ constraint_type = "UNIQUE";
+ else if (stmt->excludeOpNames != NIL)
+ constraint_type = "EXCLUDE";
+ else
+ {
+ elog(ERROR, "unknown constraint type");
+ constraint_type = NULL; /* keep compiler quiet */
+ }
+
+ /*
+ * It may be possible to support UNIQUE constraints when partition
+ * keys are expressions, but is it worth it? Give up for now.
+ */
+ if (key->partattrs[i] == 0)
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("unsupported %s constraint with partition key definition",
+ constraint_type),
+ errmsg("%s constraints cannot be used when partition keys include expressions.",
+ constraint_type)));
+
+ for (j = 0; j < indexInfo->ii_NumIndexAttrs; j++)
+ {
+ if (key->partattrs[i] == indexInfo->ii_KeyAttrNumbers[j])
+ {
+ found = true;
+ break;
+ }
+ }
+ if (!found)
+ {
+ Form_pg_attribute att;
+
+ att = TupleDescAttr(RelationGetDescr(rel), key->partattrs[i] - 1);
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("insufficient columns in %s constraint definition",
+ constraint_type),
+ errdetail("%s constraint on table \"%s\" lacks column \"%s\" which is part of the partition key.",
+ constraint_type, RelationGetRelationName(rel),
+ NameStr(att->attname))));
+ }
+ }
+ }
+
+
+ /*
* We disallow indexes on system columns other than OID. They would not
* necessarily get updated correctly, and they don't seem useful anyway.
*/
@@ -735,12 +809,14 @@ DefineIndex(Oid relationId,
indexRelationId =
index_create(rel, indexRelationName, indexRelationId, parentIndexId,
+ parentConstraintId,
stmt->oldNode, indexInfo, indexColNames,
accessMethodId, tablespaceId,
collationObjectId, classObjectId,
coloptions, reloptions,
flags, constr_flags,
- allowSystemTableMods, !check_rights);
+ allowSystemTableMods, !check_rights,
+ &createdConstraintId);
ObjectAddressSet(address, RelationRelationId, indexRelationId);
@@ -827,16 +903,40 @@ DefineIndex(Oid relationId,
opfamOids,
attmap, maplen))
{
+ Oid cldConstrOid = InvalidOid;
+
/*
- * Found a match. Attach index to parent and we're
- * done, but keep lock till commit.
+ * Found a match.
+ *
+ * If this index is being created in the parent
+ * because of a constraint, then the child needs to
+ * have a constraint also, so look for one. If there
+ * is no such constraint, this index is no good, so
+ * keep looking.
*/
+ if (createdConstraintId != InvalidOid)
+ {
+ cldConstrOid =
+ get_relation_idx_constraint_oid(childRelid,
+ cldidxid);
+ if (cldConstrOid == InvalidOid)
+ {
+ index_close(cldidx, lockmode);
+ continue;
+ }
+ }
+
+ /* Attach index to parent and we're done. */
IndexSetParentIndex(cldidx, indexRelationId);
+ if (createdConstraintId != InvalidOid)
+ ConstraintSetParentConstraint(cldConstrOid,
+ createdConstraintId);
if (!IndexIsValid(cldidx->rd_index))
invalidate_parent = true;
found = true;
+ /* keep lock till commit */
index_close(cldidx, NoLock);
break;
}
@@ -867,6 +967,7 @@ DefineIndex(Oid relationId,
DefineIndex(childRelid, childStmt,
InvalidOid, /* no predefined OID */
indexRelationId, /* this is our child */
+ createdConstraintId,
false, check_rights, check_not_in_use,
false, quiet);
}
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 2e768dd5e4..5ba7971c43 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -939,17 +939,20 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
Relation idxRel = index_open(lfirst_oid(cell), AccessShareLock);
AttrNumber *attmap;
IndexStmt *idxstmt;
+ Oid constraintOid;
attmap = convert_tuples_by_name_map(RelationGetDescr(rel),
RelationGetDescr(parent),
gettext_noop("could not convert row type"));
idxstmt =
generateClonedIndexStmt(NULL, RelationGetRelid(rel), idxRel,
- attmap, RelationGetDescr(rel)->natts);
+ attmap, RelationGetDescr(rel)->natts,
+ &constraintOid);
DefineIndex(RelationGetRelid(rel),
idxstmt,
InvalidOid,
RelationGetRelid(idxRel),
+ constraintOid,
false, false, false, false, false);
index_close(idxRel, AccessShareLock);
@@ -6809,6 +6812,7 @@ ATExecAddIndex(AlteredTableInfo *tab, Relation rel,
stmt,
InvalidOid, /* no predefined OID */
InvalidOid, /* no parent index */
+ InvalidOid, /* no parent constraint */
true, /* is_alter_table */
check_rights,
false, /* check_not_in_use - we did it already */
@@ -6901,6 +6905,7 @@ ATExecAddIndexConstraint(AlteredTableInfo *tab, Relation rel,
address = index_constraint_create(rel,
index_oid,
+ InvalidOid,
indexInfo,
constraintName,
constraintType,
@@ -14132,6 +14137,7 @@ AttachPartitionEnsureIndexes(Relation rel, Relation attachrel)
IndexInfo *info;
AttrNumber *attmap;
bool found = false;
+ Oid constraintOid;
/*
* Ignore indexes in the partitioned table other than partitioned
@@ -14148,6 +14154,7 @@ AttachPartitionEnsureIndexes(Relation rel, Relation attachrel)
attmap = convert_tuples_by_name_map(RelationGetDescr(attachrel),
RelationGetDescr(rel),
gettext_noop("could not convert row type"));
+ constraintOid = get_relation_idx_constraint_oid(RelationGetRelid(rel), idx);
/*
* Scan the list of existing indexes in the partition-to-be, and mark
@@ -14156,6 +14163,8 @@ AttachPartitionEnsureIndexes(Relation rel, Relation attachrel)
*/
for (i = 0; i < list_length(attachRelIdxs); i++)
{
+ Oid cldConstrOid;
+
/* does this index have a parent? if so, can't use it */
if (has_superclass(RelationGetRelid(attachrelIdxRels[i])))
continue;
@@ -14168,8 +14177,26 @@ AttachPartitionEnsureIndexes(Relation rel, Relation attachrel)
attmap,
RelationGetDescr(rel)->natts))
{
+ /*
+ * If this index is being created in the parent because of a
+ * constraint, then the child needs to have a constraint also,
+ * so look for one. If there is no such constraint, this
+ * index is no good, so keep looking.
+ */
+ if (OidIsValid(constraintOid))
+ {
+ cldConstrOid =
+ get_relation_idx_constraint_oid(RelationGetRelid(attachrel),
+ RelationGetRelid(attachrelIdxRels[i]));
+ /* no dice */
+ if (!OidIsValid(cldConstrOid))
+ continue;
+ }
+
/* bingo. */
IndexSetParentIndex(attachrelIdxRels[i], idx);
+ if (OidIsValid(constraintOid))
+ ConstraintSetParentConstraint(cldConstrOid, constraintOid);
found = true;
break;
}
@@ -14182,12 +14209,15 @@ AttachPartitionEnsureIndexes(Relation rel, Relation attachrel)
if (!found)
{
IndexStmt *stmt;
+ Oid constraintOid;
stmt = generateClonedIndexStmt(NULL, RelationGetRelid(attachrel),
idxRel, attmap,
- RelationGetDescr(rel)->natts);
+ RelationGetDescr(rel)->natts,
+ &constraintOid);
DefineIndex(RelationGetRelid(attachrel), stmt, InvalidOid,
RelationGetRelid(idxRel),
+ constraintOid,
false, false, false, false, false);
}
@@ -14430,6 +14460,8 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
bool found;
int i;
PartitionDesc partDesc;
+ Oid constraintOid,
+ cldConstrId;
/*
* If this partition already has an index attached, refuse the operation.
@@ -14485,8 +14517,34 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(parentIdx)),
errdetail("The index definitions do not match.")));
+ /*
+ * If there is a constraint in the parent, make sure there is one
+ * in the child too.
+ */
+ constraintOid = get_relation_idx_constraint_oid(RelationGetRelid(parentTbl),
+ RelationGetRelid(parentIdx));
+
+ if (OidIsValid(constraintOid))
+ {
+ cldConstrId = get_relation_idx_constraint_oid(RelationGetRelid(partTbl),
+ partIdxId);
+ if (!OidIsValid(cldConstrId))
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_OBJECT_DEFINITION),
+ errmsg("cannot attach index \"%s\" as a partition of index \"%s\"",
+ RelationGetRelationName(partIdx),
+ RelationGetRelationName(parentIdx)),
+ errdetail("The index \"%s\" belongs to a constraint in table \"%s\" but no constraint exists for index \"%s\".",
+ RelationGetRelationName(parentIdx),
+ RelationGetRelationName(parentTbl),
+ RelationGetRelationName(partIdx))));
+ }
+
/* All good -- do it */
IndexSetParentIndex(partIdx, RelationGetRelid(parentIdx));
+ if (OidIsValid(constraintOid))
+ ConstraintSetParentConstraint(cldConstrId, constraintOid);
+
pfree(attmap);
CommandCounterIncrement();
diff --git a/src/backend/parser/analyze.c b/src/backend/parser/analyze.c
index e7b2bc7e73..5b3a610cf9 100644
--- a/src/backend/parser/analyze.c
+++ b/src/backend/parser/analyze.c
@@ -1017,6 +1017,13 @@ transformOnConflictClause(ParseState *pstate,
TargetEntry *te;
int attno;
+ if (targetrel->rd_partdesc)
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("%s cannot be applied to partitioned table \"%s\"",
+ "ON CONFLICT DO UPDATE",
+ RelationGetRelationName(targetrel))));
+
/*
* All INSERT expressions have been parsed, get ready for potentially
* existing SET statements that need to be processed like an UPDATE.
diff --git a/src/backend/parser/parse_utilcmd.c b/src/backend/parser/parse_utilcmd.c
index 5afb363096..a93fe11828 100644
--- a/src/backend/parser/parse_utilcmd.c
+++ b/src/backend/parser/parse_utilcmd.c
@@ -704,12 +704,6 @@ transformColumnDefinition(CreateStmtContext *cxt, ColumnDef *column)
errmsg("primary key constraints are not supported on foreign tables"),
parser_errposition(cxt->pstate,
constraint->location)));
- if (cxt->ispartitioned)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("primary key constraints are not supported on partitioned tables"),
- parser_errposition(cxt->pstate,
- constraint->location)));
/* FALL THRU */
case CONSTR_UNIQUE:
@@ -719,12 +713,6 @@ transformColumnDefinition(CreateStmtContext *cxt, ColumnDef *column)
errmsg("unique constraints are not supported on foreign tables"),
parser_errposition(cxt->pstate,
constraint->location)));
- if (cxt->ispartitioned)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("unique constraints are not supported on partitioned tables"),
- parser_errposition(cxt->pstate,
- constraint->location)));
if (constraint->keys == NIL)
constraint->keys = list_make1(makeString(column->colname));
cxt->ixconstraints = lappend(cxt->ixconstraints, constraint);
@@ -821,12 +809,6 @@ transformTableConstraint(CreateStmtContext *cxt, Constraint *constraint)
errmsg("primary key constraints are not supported on foreign tables"),
parser_errposition(cxt->pstate,
constraint->location)));
- if (cxt->ispartitioned)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("primary key constraints are not supported on partitioned tables"),
- parser_errposition(cxt->pstate,
- constraint->location)));
cxt->ixconstraints = lappend(cxt->ixconstraints, constraint);
break;
@@ -837,12 +819,6 @@ transformTableConstraint(CreateStmtContext *cxt, Constraint *constraint)
errmsg("unique constraints are not supported on foreign tables"),
parser_errposition(cxt->pstate,
constraint->location)));
- if (cxt->ispartitioned)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("unique constraints are not supported on partitioned tables"),
- parser_errposition(cxt->pstate,
- constraint->location)));
cxt->ixconstraints = lappend(cxt->ixconstraints, constraint);
break;
@@ -1184,7 +1160,7 @@ transformTableLikeClause(CreateStmtContext *cxt, TableLikeClause *table_like_cla
/* Build CREATE INDEX statement to recreate the parent_index */
index_stmt = generateClonedIndexStmt(cxt->relation, InvalidOid,
parent_index,
- attmap, tupleDesc->natts);
+ attmap, tupleDesc->natts, NULL);
/* Copy comment on index, if requested */
if (table_like_clause->options & CREATE_TABLE_LIKE_COMMENTS)
@@ -1267,7 +1243,7 @@ transformOfType(CreateStmtContext *cxt, TypeName *ofTypename)
*/
IndexStmt *
generateClonedIndexStmt(RangeVar *heapRel, Oid heapRelid, Relation source_idx,
- const AttrNumber *attmap, int attmap_length)
+ const AttrNumber *attmap, int attmap_length, Oid *constraintOid)
{
Oid source_relid = RelationGetRelid(source_idx);
HeapTuple ht_idxrel;
@@ -1365,6 +1341,9 @@ generateClonedIndexStmt(RangeVar *heapRel, Oid heapRelid, Relation source_idx,
HeapTuple ht_constr;
Form_pg_constraint conrec;
+ if (constraintOid)
+ *constraintOid = constraintId;
+
ht_constr = SearchSysCache1(CONSTROID,
ObjectIdGetDatum(constraintId));
if (!HeapTupleIsValid(ht_constr))
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 3abe7d6155..8c23ee53e2 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1353,6 +1353,7 @@ ProcessUtilitySlow(ParseState *pstate,
stmt,
InvalidOid, /* no predefined OID */
InvalidOid, /* no parent index */
+ InvalidOid, /* no parent constraint */
false, /* is_alter_table */
true, /* check_rights */
true, /* check_not_in_use */
diff --git a/src/include/catalog/index.h b/src/include/catalog/index.h
index 235e180299..c8b6fe85c5 100644
--- a/src/include/catalog/index.h
+++ b/src/include/catalog/index.h
@@ -54,6 +54,7 @@ extern Oid index_create(Relation heapRelation,
const char *indexRelationName,
Oid indexRelationId,
Oid parentIndexRelid,
+ Oid parentConstraintId,
Oid relFileNode,
IndexInfo *indexInfo,
List *indexColNames,
@@ -66,7 +67,8 @@ extern Oid index_create(Relation heapRelation,
bits16 flags,
bits16 constr_flags,
bool allow_system_table_mods,
- bool is_internal);
+ bool is_internal,
+ Oid *constraintId);
#define INDEX_CONSTR_CREATE_MARK_AS_PRIMARY (1 << 0)
#define INDEX_CONSTR_CREATE_DEFERRABLE (1 << 1)
@@ -76,6 +78,7 @@ extern Oid index_create(Relation heapRelation,
extern ObjectAddress index_constraint_create(Relation heapRelation,
Oid indexRelationId,
+ Oid parentConstraintId,
IndexInfo *indexInfo,
const char *constraintName,
char constraintType,
diff --git a/src/include/catalog/pg_constraint_fn.h b/src/include/catalog/pg_constraint_fn.h
index 6bb1b09714..d3351f4a83 100644
--- a/src/include/catalog/pg_constraint_fn.h
+++ b/src/include/catalog/pg_constraint_fn.h
@@ -58,7 +58,6 @@ extern Oid CreateConstraintEntry(const char *constraintName,
extern void RemoveConstraintById(Oid conId);
extern void RenameConstraintById(Oid conId, const char *newname);
-extern void SetValidatedConstraintById(Oid conId);
extern bool ConstraintNameIsUsed(ConstraintCategory conCat, Oid objId,
Oid objNamespace, const char *conname);
@@ -68,10 +67,13 @@ extern char *ChooseConstraintName(const char *name1, const char *name2,
extern void AlterConstraintNamespaces(Oid ownerId, Oid oldNspId,
Oid newNspId, bool isType, ObjectAddresses *objsMoved);
+extern void ConstraintSetParentConstraint(Oid childConstrId,
+ Oid parentConstrId);
extern Oid get_relation_constraint_oid(Oid relid, const char *conname, bool missing_ok);
extern Bitmapset *get_relation_constraint_attnos(Oid relid, const char *conname,
bool missing_ok, Oid *constraintOid);
extern Oid get_domain_constraint_oid(Oid typid, const char *conname, bool missing_ok);
+extern Oid get_relation_idx_constraint_oid(Oid relationId, Oid indexId);
extern Bitmapset *get_primary_key_attnos(Oid relid, bool deferrableOk,
Oid *constraintOid);
diff --git a/src/include/commands/defrem.h b/src/include/commands/defrem.h
index 7b824c95af..f510f40945 100644
--- a/src/include/commands/defrem.h
+++ b/src/include/commands/defrem.h
@@ -26,6 +26,7 @@ extern ObjectAddress DefineIndex(Oid relationId,
IndexStmt *stmt,
Oid indexRelationId,
Oid parentIndexId,
+ Oid parentConstraintId,
bool is_alter_table,
bool check_rights,
bool check_not_in_use,
diff --git a/src/include/parser/parse_utilcmd.h b/src/include/parser/parse_utilcmd.h
index 64aa8234e5..35ac97940a 100644
--- a/src/include/parser/parse_utilcmd.h
+++ b/src/include/parser/parse_utilcmd.h
@@ -29,6 +29,7 @@ extern PartitionBoundSpec *transformPartitionBound(ParseState *pstate, Relation
PartitionBoundSpec *spec);
extern IndexStmt *generateClonedIndexStmt(RangeVar *heapRel, Oid heapOid,
Relation source_idx,
- const AttrNumber *attmap, int attmap_length);
+ const AttrNumber *attmap, int attmap_length,
+ Oid *constraintOid);
#endif /* PARSE_UTILCMD_H */
diff --git a/src/test/regress/expected/alter_table.out b/src/test/regress/expected/alter_table.out
index e9a1d37f6f..ccd2c38dbc 100644
--- a/src/test/regress/expected/alter_table.out
+++ b/src/test/regress/expected/alter_table.out
@@ -3305,14 +3305,6 @@ CREATE TABLE partitioned (
a int,
b int
) PARTITION BY RANGE (a, (a+b+1));
-ALTER TABLE partitioned ADD UNIQUE (a);
-ERROR: unique constraints are not supported on partitioned tables
-LINE 1: ALTER TABLE partitioned ADD UNIQUE (a);
- ^
-ALTER TABLE partitioned ADD PRIMARY KEY (a);
-ERROR: primary key constraints are not supported on partitioned tables
-LINE 1: ALTER TABLE partitioned ADD PRIMARY KEY (a);
- ^
ALTER TABLE partitioned ADD FOREIGN KEY (a) REFERENCES blah;
ERROR: foreign key constraints are not supported on partitioned tables
LINE 1: ALTER TABLE partitioned ADD FOREIGN KEY (a) REFERENCES blah;
diff --git a/src/test/regress/expected/create_table.out b/src/test/regress/expected/create_table.out
index 8e745402ae..866cc99b9f 100644
--- a/src/test/regress/expected/create_table.out
+++ b/src/test/regress/expected/create_table.out
@@ -276,12 +276,6 @@ CREATE TABLE partitioned (
) PARTITION BY LIST (a1, a2); -- fail
ERROR: cannot use "list" partition strategy with more than one column
-- unsupported constraint type for partitioned tables
-CREATE TABLE partitioned (
- a int PRIMARY KEY
-) PARTITION BY RANGE (a);
-ERROR: primary key constraints are not supported on partitioned tables
-LINE 2: a int PRIMARY KEY
- ^
CREATE TABLE pkrel (
a int PRIMARY KEY
);
@@ -293,12 +287,6 @@ LINE 2: a int REFERENCES pkrel(a)
^
DROP TABLE pkrel;
CREATE TABLE partitioned (
- a int UNIQUE
-) PARTITION BY RANGE (a);
-ERROR: unique constraints are not supported on partitioned tables
-LINE 2: a int UNIQUE
- ^
-CREATE TABLE partitioned (
a int,
EXCLUDE USING gist (a WITH &&)
) PARTITION BY RANGE (a);
diff --git a/src/test/regress/expected/indexing.out b/src/test/regress/expected/indexing.out
index ffd4b10c37..febd1b3162 100644
--- a/src/test/regress/expected/indexing.out
+++ b/src/test/regress/expected/indexing.out
@@ -26,8 +26,6 @@ drop table idxpart;
-- Some unsupported features
create table idxpart (a int, b int, c text) partition by range (a);
create table idxpart1 partition of idxpart for values from (0) to (10);
-create unique index on idxpart (a);
-ERROR: cannot create unique index on partitioned table "idxpart"
create index concurrently on idxpart (a);
ERROR: cannot create index on partitioned table "idxpart" concurrently
drop table idxpart;
@@ -744,6 +742,256 @@ select attrelid::regclass, attname, attnum from pg_attribute
(7 rows)
drop table idxpart;
+--
+-- Constraint-related indexes
+--
+-- Verify that it works to add primary key / unique to partitioned tables
+create table idxpart (a int primary key, b int) partition by range (a);
+\d idxpart
+ Table "public.idxpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | not null |
+ b | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "idxpart_pkey" PRIMARY KEY, btree (a)
+Number of partitions: 0
+
+drop table idxpart;
+-- but not if you fail to use the full partition key
+create table idxpart (a int unique, b int) partition by range (a, b);
+ERROR: insufficient columns in UNIQUE constraint definition
+DETAIL: UNIQUE constraint on table "idxpart" lacks column "b" which is part of the partition key.
+create table idxpart (a int, b int unique) partition by range (a, b);
+ERROR: insufficient columns in UNIQUE constraint definition
+DETAIL: UNIQUE constraint on table "idxpart" lacks column "a" which is part of the partition key.
+create table idxpart (a int primary key, b int) partition by range (b, a);
+ERROR: insufficient columns in PRIMARY KEY constraint definition
+DETAIL: PRIMARY KEY constraint on table "idxpart" lacks column "b" which is part of the partition key.
+create table idxpart (a int, b int primary key) partition by range (b, a);
+ERROR: insufficient columns in PRIMARY KEY constraint definition
+DETAIL: PRIMARY KEY constraint on table "idxpart" lacks column "a" which is part of the partition key.
+-- OK if you use them in some other order
+create table idxpart (a int, b int, c text, primary key (a, b, c)) partition by range (b, c, a);
+drop table idxpart;
+create table idxpart (a int primary key, b int) partition by range ((b + a));
+ERROR: unsupported PRIMARY KEY constraint with partition key definition
+-- not other types of index-based constraints
+create table idxpart (a int, exclude (a with = )) partition by range (a);
+ERROR: exclusion constraints are not supported on partitioned tables
+LINE 1: create table idxpart (a int, exclude (a with = )) partition ...
+ ^
+-- It works to add primary keys after the partitioned table is created
+create table idxpart (a int, b int, c text) partition by range (a, b);
+alter table idxpart add primary key (a); -- not an incomplete one tho
+ERROR: insufficient columns in PRIMARY KEY constraint definition
+DETAIL: PRIMARY KEY constraint on table "idxpart" lacks column "b" which is part of the partition key.
+alter table idxpart add primary key (a, b);
+\d idxpart
+ Table "public.idxpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | not null |
+ b | integer | | not null |
+ c | text | | |
+Partition key: RANGE (a, b)
+Indexes:
+ "idxpart_pkey" PRIMARY KEY, btree (a, b)
+Number of partitions: 0
+
+create table idxpart1 partition of idxpart for values from (0, 0) to (1000, 1000);
+\d idxpart1
+ Table "public.idxpart1"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | not null |
+ b | integer | | not null |
+ c | text | | |
+Partition of: idxpart FOR VALUES FROM (0, 0) TO (1000, 1000)
+Indexes:
+ "idxpart1_pkey" PRIMARY KEY, btree (a, b)
+
+drop table idxpart;
+-- It works to add unique constraints after the partitioned table is created
+create table idxpart (a int, b int) partition by range (a, b);
+alter table idxpart add unique (a); -- ... nope
+ERROR: insufficient columns in UNIQUE constraint definition
+DETAIL: UNIQUE constraint on table "idxpart" lacks column "b" which is part of the partition key.
+alter table idxpart add unique (b, a);
+\d idxpart
+ Table "public.idxpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+ b | integer | | |
+Partition key: RANGE (a, b)
+Indexes:
+ "idxpart_b_a_key" UNIQUE CONSTRAINT, btree (b, a)
+Number of partitions: 0
+
+drop table idxpart;
+-- Exclusion constraints cannot be added
+create table idxpart (a int, b int) partition by range (a);
+alter table idxpart add exclude (a with =);
+ERROR: exclusion constraints are not supported on partitioned tables
+LINE 1: alter table idxpart add exclude (a with =);
+ ^
+drop table idxpart;
+-- When (sub)partitions are created, they also contain the constraint
+create table idxpart (a int, b int, primary key (a, b)) partition by range (a, b);
+create table idxpart1 partition of idxpart for values from (1, 1) to (10, 10);
+create table idxpart2 partition of idxpart for values from (10, 10) to (20, 20)
+ partition by range (b);
+create table idxpart21 partition of idxpart2 for values from (10) to (15);
+create table idxpart22 partition of idxpart2 for values from (15) to (20);
+create table idxpart3 (b int not null, a int not null);
+alter table idxpart attach partition idxpart3 for values from (20, 20) to (30, 30);
+select conname, contype, conrelid::regclass, conindid::regclass, conkey
+ from pg_constraint where conrelid::regclass::text like 'idxpart%'
+ order by conname;
+ conname | contype | conrelid | conindid | conkey
+----------------+---------+-----------+----------------+--------
+ idxpart1_pkey | p | idxpart1 | idxpart1_pkey | {1,2}
+ idxpart21_pkey | p | idxpart21 | idxpart21_pkey | {1,2}
+ idxpart22_pkey | p | idxpart22 | idxpart22_pkey | {1,2}
+ idxpart2_pkey | p | idxpart2 | idxpart2_pkey | {1,2}
+ idxpart3_pkey | p | idxpart3 | idxpart3_pkey | {2,1}
+ idxpart_pkey | p | idxpart | idxpart_pkey | {1,2}
+(6 rows)
+
+drop table idxpart;
+-- Verify that multi-layer partitioning honors the requirement that all
+-- columns in the partition key must appear in primary key
+create table idxpart (a int, b int, primary key (a)) partition by range (a);
+create table idxpart2 partition of idxpart
+for values from (0) to (1000) partition by range (b); -- fail
+ERROR: insufficient columns in PRIMARY KEY constraint definition
+DETAIL: PRIMARY KEY constraint on table "idxpart2" lacks column "b" which is part of the partition key.
+drop table idxpart;
+-- Multi-layer partitioning works correctly in this case:
+create table idxpart (a int, b int, primary key (a, b)) partition by range (a);
+create table idxpart2 partition of idxpart for values from (0) to (1000) partition by range (b);
+create table idxpart21 partition of idxpart2 for values from (0) to (1000);
+select conname, contype, conrelid::regclass, conindid::regclass, conkey
+ from pg_constraint where conrelid::regclass::text like 'idxpart%'
+ order by conname;
+ conname | contype | conrelid | conindid | conkey
+----------------+---------+-----------+----------------+--------
+ idxpart21_pkey | p | idxpart21 | idxpart21_pkey | {1,2}
+ idxpart2_pkey | p | idxpart2 | idxpart2_pkey | {1,2}
+ idxpart_pkey | p | idxpart | idxpart_pkey | {1,2}
+(3 rows)
+
+drop table idxpart;
+-- If a partitioned table has a unique/PK constraint, then it's not possible
+-- to drop the corresponding constraint in the children; nor it's possible
+-- to drop the indexes individually. Dropping the constraint in the parent
+-- gets rid of the lot.
+create table idxpart (i int) partition by hash (i);
+create table idxpart0 partition of idxpart (i) for values with (modulus 2, remainder 0);
+create table idxpart1 partition of idxpart (i) for values with (modulus 2, remainder 1);
+alter table idxpart0 add primary key(i);
+alter table idxpart add primary key(i);
+select indrelid::regclass, indexrelid::regclass, inhparent::regclass, indisvalid,
+ conname, conislocal, coninhcount, connoinherit, convalidated
+ from pg_index idx left join pg_inherits inh on (idx.indexrelid = inh.inhrelid)
+ left join pg_constraint con on (idx.indexrelid = con.conindid)
+ where indrelid::regclass::text like 'idxpart%'
+ order by indexrelid::regclass::text collate "C";
+ indrelid | indexrelid | inhparent | indisvalid | conname | conislocal | coninhcount | connoinherit | convalidated
+----------+---------------+--------------+------------+---------------+------------+-------------+--------------+--------------
+ idxpart0 | idxpart0_pkey | idxpart_pkey | t | idxpart0_pkey | f | 1 | t | t
+ idxpart1 | idxpart1_pkey | idxpart_pkey | t | idxpart1_pkey | f | 1 | t | t
+ idxpart | idxpart_pkey | | t | idxpart_pkey | t | 0 | t | t
+(3 rows)
+
+drop index idxpart0_pkey; -- fail
+ERROR: cannot drop index idxpart0_pkey because index idxpart_pkey requires it
+HINT: You can drop index idxpart_pkey instead.
+drop index idxpart1_pkey; -- fail
+ERROR: cannot drop index idxpart1_pkey because index idxpart_pkey requires it
+HINT: You can drop index idxpart_pkey instead.
+alter table idxpart0 drop constraint idxpart0_pkey; -- fail
+ERROR: cannot drop inherited constraint "idxpart0_pkey" of relation "idxpart0"
+alter table idxpart1 drop constraint idxpart1_pkey; -- fail
+ERROR: cannot drop inherited constraint "idxpart1_pkey" of relation "idxpart1"
+alter table idxpart drop constraint idxpart_pkey; -- ok
+select indrelid::regclass, indexrelid::regclass, inhparent::regclass, indisvalid,
+ conname, conislocal, coninhcount, connoinherit, convalidated
+ from pg_index idx left join pg_inherits inh on (idx.indexrelid = inh.inhrelid)
+ left join pg_constraint con on (idx.indexrelid = con.conindid)
+ where indrelid::regclass::text like 'idxpart%'
+ order by indexrelid::regclass::text collate "C";
+ indrelid | indexrelid | inhparent | indisvalid | conname | conislocal | coninhcount | connoinherit | convalidated
+----------+------------+-----------+------------+---------+------------+-------------+--------------+--------------
+(0 rows)
+
+drop table idxpart;
+-- If a partitioned table has a constraint whose index is not valid,
+-- attaching a missing partition makes it valid.
+create table idxpart (a int) partition by range (a);
+create table idxpart0 (like idxpart);
+alter table idxpart0 add primary key (a);
+alter table idxpart attach partition idxpart0 for values from (0) to (1000);
+alter table only idxpart add primary key (a);
+select indrelid::regclass, indexrelid::regclass, inhparent::regclass, indisvalid,
+ conname, conislocal, coninhcount, connoinherit, convalidated
+ from pg_index idx left join pg_inherits inh on (idx.indexrelid = inh.inhrelid)
+ left join pg_constraint con on (idx.indexrelid = con.conindid)
+ where indrelid::regclass::text like 'idxpart%'
+ order by indexrelid::regclass::text collate "C";
+ indrelid | indexrelid | inhparent | indisvalid | conname | conislocal | coninhcount | connoinherit | convalidated
+----------+---------------+-----------+------------+---------------+------------+-------------+--------------+--------------
+ idxpart0 | idxpart0_pkey | | t | idxpart0_pkey | t | 0 | t | t
+ idxpart | idxpart_pkey | | f | idxpart_pkey | t | 0 | t | t
+(2 rows)
+
+alter index idxpart_pkey attach partition idxpart0_pkey;
+select indrelid::regclass, indexrelid::regclass, inhparent::regclass, indisvalid,
+ conname, conislocal, coninhcount, connoinherit, convalidated
+ from pg_index idx left join pg_inherits inh on (idx.indexrelid = inh.inhrelid)
+ left join pg_constraint con on (idx.indexrelid = con.conindid)
+ where indrelid::regclass::text like 'idxpart%'
+ order by indexrelid::regclass::text collate "C";
+ indrelid | indexrelid | inhparent | indisvalid | conname | conislocal | coninhcount | connoinherit | convalidated
+----------+---------------+--------------+------------+---------------+------------+-------------+--------------+--------------
+ idxpart0 | idxpart0_pkey | idxpart_pkey | t | idxpart0_pkey | f | 1 | t | t
+ idxpart | idxpart_pkey | | t | idxpart_pkey | t | 0 | t | t
+(2 rows)
+
+drop table idxpart;
+-- if a partition has a unique index without a constraint, does not attach
+-- automatically; creates a new index instead.
+create table idxpart (a int, b int) partition by range (a);
+create table idxpart1 (a int not null, b int);
+create unique index on idxpart1 (a);
+alter table idxpart add primary key (a);
+alter table idxpart attach partition idxpart1 for values from (1) to (1000);
+select indrelid::regclass, indexrelid::regclass, inhparent::regclass, indisvalid,
+ conname, conislocal, coninhcount, connoinherit, convalidated
+ from pg_index idx left join pg_inherits inh on (idx.indexrelid = inh.inhrelid)
+ left join pg_constraint con on (idx.indexrelid = con.conindid)
+ where indrelid::regclass::text like 'idxpart%'
+ order by indexrelid::regclass::text collate "C";
+ indrelid | indexrelid | inhparent | indisvalid | conname | conislocal | coninhcount | connoinherit | convalidated
+----------+----------------+--------------+------------+---------------+------------+-------------+--------------+--------------
+ idxpart1 | idxpart1_a_idx | | t | | | | |
+ idxpart1 | idxpart1_pkey | idxpart_pkey | t | idxpart1_pkey | f | 1 | t | t
+ idxpart | idxpart_pkey | | t | idxpart_pkey | t | 0 | t | t
+(3 rows)
+
+drop table idxpart;
+-- Can't attach an index without a corresponding constraint
+create table idxpart (a int, b int) partition by range (a);
+create table idxpart1 (a int not null, b int);
+create unique index on idxpart1 (a);
+alter table idxpart attach partition idxpart1 for values from (1) to (1000);
+alter table only idxpart add primary key (a);
+alter index idxpart_pkey attach partition idxpart1_a_idx; -- fail
+ERROR: cannot attach index "idxpart1_a_idx" as a partition of index "idxpart_pkey"
+DETAIL: The index "idxpart_pkey" belongs to a constraint in table "idxpart" but no constraint exists for index "idxpart1_a_idx".
+drop table idxpart;
-- intentionally leave some objects around
create table idxpart (a int) partition by range (a);
create table idxpart1 partition of idxpart for values from (0) to (100);
@@ -755,3 +1003,5 @@ create index on idxpart22 (a);
create index on only idxpart2 (a);
alter index idxpart2_a_idx attach partition idxpart22_a_idx;
create index on idxpart (a);
+create table idxpart_another (a int, b int, primary key (a, b)) partition by range (a);
+create table idxpart_another_1 partition of idxpart_another for values from (0) to (100);
diff --git a/src/test/regress/expected/insert_conflict.out b/src/test/regress/expected/insert_conflict.out
index 8fd2027d6a..2650faedee 100644
--- a/src/test/regress/expected/insert_conflict.out
+++ b/src/test/regress/expected/insert_conflict.out
@@ -794,7 +794,7 @@ insert into parted_conflict_test values (1, 'a') on conflict do nothing;
insert into parted_conflict_test values (1, 'a') on conflict do nothing;
-- however, on conflict do update is not supported yet
insert into parted_conflict_test values (1) on conflict (b) do update set a = excluded.a;
-ERROR: there is no unique or exclusion constraint matching the ON CONFLICT specification
+ERROR: ON CONFLICT DO UPDATE cannot be applied to partitioned table "parted_conflict_test"
-- but it works OK if we target the partition directly
insert into parted_conflict_test_1 values (1) on conflict (b) do
update set a = excluded.a;
diff --git a/src/test/regress/sql/alter_table.sql b/src/test/regress/sql/alter_table.sql
index b27e8f6777..b73f523e8a 100644
--- a/src/test/regress/sql/alter_table.sql
+++ b/src/test/regress/sql/alter_table.sql
@@ -2035,8 +2035,6 @@ CREATE TABLE partitioned (
a int,
b int
) PARTITION BY RANGE (a, (a+b+1));
-ALTER TABLE partitioned ADD UNIQUE (a);
-ALTER TABLE partitioned ADD PRIMARY KEY (a);
ALTER TABLE partitioned ADD FOREIGN KEY (a) REFERENCES blah;
ALTER TABLE partitioned ADD EXCLUDE USING gist (a WITH &&);
diff --git a/src/test/regress/sql/create_table.sql b/src/test/regress/sql/create_table.sql
index 8f9991ef18..fefccf21a2 100644
--- a/src/test/regress/sql/create_table.sql
+++ b/src/test/regress/sql/create_table.sql
@@ -294,10 +294,6 @@ CREATE TABLE partitioned (
) PARTITION BY LIST (a1, a2); -- fail
-- unsupported constraint type for partitioned tables
-CREATE TABLE partitioned (
- a int PRIMARY KEY
-) PARTITION BY RANGE (a);
-
CREATE TABLE pkrel (
a int PRIMARY KEY
);
@@ -307,10 +303,6 @@ CREATE TABLE partitioned (
DROP TABLE pkrel;
CREATE TABLE partitioned (
- a int UNIQUE
-) PARTITION BY RANGE (a);
-
-CREATE TABLE partitioned (
a int,
EXCLUDE USING gist (a WITH &&)
) PARTITION BY RANGE (a);
diff --git a/src/test/regress/sql/indexing.sql b/src/test/regress/sql/indexing.sql
index 2f985ec866..439d19c621 100644
--- a/src/test/regress/sql/indexing.sql
+++ b/src/test/regress/sql/indexing.sql
@@ -15,7 +15,6 @@ drop table idxpart;
-- Some unsupported features
create table idxpart (a int, b int, c text) partition by range (a);
create table idxpart1 partition of idxpart for values from (0) to (10);
-create unique index on idxpart (a);
create index concurrently on idxpart (a);
drop table idxpart;
@@ -375,6 +374,154 @@ select attrelid::regclass, attname, attnum from pg_attribute
order by attrelid::regclass, attnum;
drop table idxpart;
+--
+-- Constraint-related indexes
+--
+
+-- Verify that it works to add primary key / unique to partitioned tables
+create table idxpart (a int primary key, b int) partition by range (a);
+\d idxpart
+drop table idxpart;
+
+-- but not if you fail to use the full partition key
+create table idxpart (a int unique, b int) partition by range (a, b);
+create table idxpart (a int, b int unique) partition by range (a, b);
+create table idxpart (a int primary key, b int) partition by range (b, a);
+create table idxpart (a int, b int primary key) partition by range (b, a);
+
+-- OK if you use them in some other order
+create table idxpart (a int, b int, c text, primary key (a, b, c)) partition by range (b, c, a);
+drop table idxpart;
+
+create table idxpart (a int primary key, b int) partition by range ((b + a));
+-- not other types of index-based constraints
+create table idxpart (a int, exclude (a with = )) partition by range (a);
+
+-- It works to add primary keys after the partitioned table is created
+create table idxpart (a int, b int, c text) partition by range (a, b);
+alter table idxpart add primary key (a); -- not an incomplete one tho
+alter table idxpart add primary key (a, b);
+\d idxpart
+create table idxpart1 partition of idxpart for values from (0, 0) to (1000, 1000);
+\d idxpart1
+drop table idxpart;
+
+-- It works to add unique constraints after the partitioned table is created
+create table idxpart (a int, b int) partition by range (a, b);
+alter table idxpart add unique (a); -- ... nope
+alter table idxpart add unique (b, a);
+\d idxpart
+drop table idxpart;
+
+-- Exclusion constraints cannot be added
+create table idxpart (a int, b int) partition by range (a);
+alter table idxpart add exclude (a with =);
+drop table idxpart;
+
+-- When (sub)partitions are created, they also contain the constraint
+create table idxpart (a int, b int, primary key (a, b)) partition by range (a, b);
+create table idxpart1 partition of idxpart for values from (1, 1) to (10, 10);
+create table idxpart2 partition of idxpart for values from (10, 10) to (20, 20)
+ partition by range (b);
+create table idxpart21 partition of idxpart2 for values from (10) to (15);
+create table idxpart22 partition of idxpart2 for values from (15) to (20);
+create table idxpart3 (b int not null, a int not null);
+alter table idxpart attach partition idxpart3 for values from (20, 20) to (30, 30);
+select conname, contype, conrelid::regclass, conindid::regclass, conkey
+ from pg_constraint where conrelid::regclass::text like 'idxpart%'
+ order by conname;
+drop table idxpart;
+
+-- Verify that multi-layer partitioning honors the requirement that all
+-- columns in the partition key must appear in primary key
+create table idxpart (a int, b int, primary key (a)) partition by range (a);
+create table idxpart2 partition of idxpart
+for values from (0) to (1000) partition by range (b); -- fail
+drop table idxpart;
+
+-- Multi-layer partitioning works correctly in this case:
+create table idxpart (a int, b int, primary key (a, b)) partition by range (a);
+create table idxpart2 partition of idxpart for values from (0) to (1000) partition by range (b);
+create table idxpart21 partition of idxpart2 for values from (0) to (1000);
+select conname, contype, conrelid::regclass, conindid::regclass, conkey
+ from pg_constraint where conrelid::regclass::text like 'idxpart%'
+ order by conname;
+drop table idxpart;
+
+-- If a partitioned table has a unique/PK constraint, then it's not possible
+-- to drop the corresponding constraint in the children; nor it's possible
+-- to drop the indexes individually. Dropping the constraint in the parent
+-- gets rid of the lot.
+create table idxpart (i int) partition by hash (i);
+create table idxpart0 partition of idxpart (i) for values with (modulus 2, remainder 0);
+create table idxpart1 partition of idxpart (i) for values with (modulus 2, remainder 1);
+alter table idxpart0 add primary key(i);
+alter table idxpart add primary key(i);
+select indrelid::regclass, indexrelid::regclass, inhparent::regclass, indisvalid,
+ conname, conislocal, coninhcount, connoinherit, convalidated
+ from pg_index idx left join pg_inherits inh on (idx.indexrelid = inh.inhrelid)
+ left join pg_constraint con on (idx.indexrelid = con.conindid)
+ where indrelid::regclass::text like 'idxpart%'
+ order by indexrelid::regclass::text collate "C";
+drop index idxpart0_pkey; -- fail
+drop index idxpart1_pkey; -- fail
+alter table idxpart0 drop constraint idxpart0_pkey; -- fail
+alter table idxpart1 drop constraint idxpart1_pkey; -- fail
+alter table idxpart drop constraint idxpart_pkey; -- ok
+select indrelid::regclass, indexrelid::regclass, inhparent::regclass, indisvalid,
+ conname, conislocal, coninhcount, connoinherit, convalidated
+ from pg_index idx left join pg_inherits inh on (idx.indexrelid = inh.inhrelid)
+ left join pg_constraint con on (idx.indexrelid = con.conindid)
+ where indrelid::regclass::text like 'idxpart%'
+ order by indexrelid::regclass::text collate "C";
+drop table idxpart;
+
+-- If a partitioned table has a constraint whose index is not valid,
+-- attaching a missing partition makes it valid.
+create table idxpart (a int) partition by range (a);
+create table idxpart0 (like idxpart);
+alter table idxpart0 add primary key (a);
+alter table idxpart attach partition idxpart0 for values from (0) to (1000);
+alter table only idxpart add primary key (a);
+select indrelid::regclass, indexrelid::regclass, inhparent::regclass, indisvalid,
+ conname, conislocal, coninhcount, connoinherit, convalidated
+ from pg_index idx left join pg_inherits inh on (idx.indexrelid = inh.inhrelid)
+ left join pg_constraint con on (idx.indexrelid = con.conindid)
+ where indrelid::regclass::text like 'idxpart%'
+ order by indexrelid::regclass::text collate "C";
+alter index idxpart_pkey attach partition idxpart0_pkey;
+select indrelid::regclass, indexrelid::regclass, inhparent::regclass, indisvalid,
+ conname, conislocal, coninhcount, connoinherit, convalidated
+ from pg_index idx left join pg_inherits inh on (idx.indexrelid = inh.inhrelid)
+ left join pg_constraint con on (idx.indexrelid = con.conindid)
+ where indrelid::regclass::text like 'idxpart%'
+ order by indexrelid::regclass::text collate "C";
+drop table idxpart;
+
+-- if a partition has a unique index without a constraint, does not attach
+-- automatically; creates a new index instead.
+create table idxpart (a int, b int) partition by range (a);
+create table idxpart1 (a int not null, b int);
+create unique index on idxpart1 (a);
+alter table idxpart add primary key (a);
+alter table idxpart attach partition idxpart1 for values from (1) to (1000);
+select indrelid::regclass, indexrelid::regclass, inhparent::regclass, indisvalid,
+ conname, conislocal, coninhcount, connoinherit, convalidated
+ from pg_index idx left join pg_inherits inh on (idx.indexrelid = inh.inhrelid)
+ left join pg_constraint con on (idx.indexrelid = con.conindid)
+ where indrelid::regclass::text like 'idxpart%'
+ order by indexrelid::regclass::text collate "C";
+drop table idxpart;
+
+-- Can't attach an index without a corresponding constraint
+create table idxpart (a int, b int) partition by range (a);
+create table idxpart1 (a int not null, b int);
+create unique index on idxpart1 (a);
+alter table idxpart attach partition idxpart1 for values from (1) to (1000);
+alter table only idxpart add primary key (a);
+alter index idxpart_pkey attach partition idxpart1_a_idx; -- fail
+drop table idxpart;
+
-- intentionally leave some objects around
create table idxpart (a int) partition by range (a);
create table idxpart1 partition of idxpart for values from (0) to (100);
@@ -386,3 +533,5 @@ create index on idxpart22 (a);
create index on only idxpart2 (a);
alter index idxpart2_a_idx attach partition idxpart22_a_idx;
create index on idxpart (a);
+create table idxpart_another (a int, b int, primary key (a, b)) partition by range (a);
+create table idxpart_another_1 partition of idxpart_another for values from (0) to (100);
--
2.11.0
Re: unique indexes on partitioned tables
Hi Alvaro,
On 01/22/2018 05:55 PM, Alvaro Herrera wrote:
Alvaro Herrera wrote:
Version 4 of this patch, rebased on today's master.
Passes make check-world.
Maybe add a test case to indexing.sql that highlights that hash indexes
doesn't support UNIQUE; although not unique to partitioned indexes.
Thanks for working on this !
Best regards,
Jesper
Re: unique indexes on partitioned tables
On 1/22/18 17:55, Alvaro Herrera wrote:
Alvaro Herrera wrote:
Version 4 of this patch, rebased on today's master.
+ if (key->partattrs[i] == 0)
+ ereport(ERROR,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("unsupported %s constraint with
partition key definition",
+ constraint_type),
+ errmsg("%s constraints cannot be used when
partition keys include expressions.",
+ constraint_type)));Double errmsg(). (Maybe an Assert somewhere should help catch this?)
+alter table idxpart add primary key (a); -- not an incomplete one tho
"though"?
I would like to see some tests that the unique constraints are actually
enforced. That is, insert some duplicate values and see it fail. Throw
some null values in, to check PK behavior as well. It should be
trivial, but seems kind of useful.
Other than that, this looks pretty good to me. A logical extension of
the previous partitioned index patch.
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
Re: unique indexes on partitioned tables
Hi Alvaro.
On 2018/01/23 7:55, Alvaro Herrera wrote:
Alvaro Herrera wrote:
Version 4 of this patch, rebased on today's master.
With the latest patch, I noticed what I think is an unintended behavior.
create table p (a int, b int) partition by list (a);
create table p1 partition of p for values in (1) partition by range (b);
create table p11 partition of p1 for values from (1) to (10);
create table p2 partition of p for values in (2);
create unique index on p (a);
ERROR: insufficient columns in UNIQUE constraint definition
DETAIL: UNIQUE constraint on table "p1" lacks column "b" which is part of
the partition key.
It seems that after recursing to p1 which is itself partitioned,
DefineIndex() mistakenly looks for column b (which is in the p1's
partition key) in the unique key. I think that's unnecessary.
DefineIndex() should check that only once, that is, before recursing.
Please find attached a fix, a delta patch which applies on top of your v4
patch. With it:
create unique index on p (a);
insert into p values (1, 1);
insert into p values (1, 1);
ERROR: duplicate key value violates unique constraint "p11_a_idx"
DETAIL: Key (a)=(1) already exists.
insert into p values (2, 1);
insert into p values (2, 1);
ERROR: duplicate key value violates unique constraint "p2_a_idx"
DETAIL: Key (a)=(2) already exists.
drop index p_a_idx;
create unique index on p (a, b);
insert into p values (1, 1);
insert into p values (1, 1);
ERROR: duplicate key value violates unique constraint "p11_a_b_idx"
DETAIL: Key (a, b)=(1, 1) already exists.
insert into p values (2, 1);
insert into p values (2, 1);
ERROR: duplicate key value violates unique constraint "p2_a_b_idx"
DETAIL: Key (a, b)=(2, 1) already exists.
Am I missing something?
Thanks,
Amit
Attachments:
v4-delta.patchtext/plain; charset=UTF-8; name=v4-delta.patchDownload
diff --git a/src/backend/bootstrap/bootparse.y b/src/backend/bootstrap/bootparse.y
index 9e81f9514d..d15e13d984 100644
--- a/src/backend/bootstrap/bootparse.y
+++ b/src/backend/bootstrap/bootparse.y
@@ -327,6 +327,7 @@ Boot_DeclareIndexStmt:
false,
false,
true, /* skip_build */
+ false,
false);
do_end();
}
@@ -373,6 +374,7 @@ Boot_DeclareUniqueIndexStmt:
false,
false,
true, /* skip_build */
+ false,
false);
do_end();
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index b0e5ede488..f994c478d9 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -325,7 +325,8 @@ DefineIndex(Oid relationId,
bool check_rights,
bool check_not_in_use,
bool skip_build,
- bool quiet)
+ bool quiet,
+ bool recursing)
{
char *indexRelationName;
char *accessMethodName;
@@ -642,8 +643,11 @@ DefineIndex(Oid relationId,
*
* We could lift this limitation if we had global indexes, but those have
* their own problems, so this is a useful feature combination.
+ *
+ * If recursing for an index being defined on some ancestor, this must
+ * have been checked already.
*/
- if (partitioned && (stmt->unique || stmt->primary))
+ if (partitioned && (stmt->unique || stmt->primary) && !recursing)
{
PartitionKey key = rel->rd_partkey;
int i;
@@ -969,7 +973,7 @@ DefineIndex(Oid relationId,
indexRelationId, /* this is our child */
createdConstraintId,
false, check_rights, check_not_in_use,
- false, quiet);
+ false, quiet, true);
}
pfree(attmap);
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 2b22946c3c..a32fd4d86c 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -953,7 +953,7 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
InvalidOid,
RelationGetRelid(idxRel),
constraintOid,
- false, false, false, false, false);
+ false, false, false, false, false, false);
index_close(idxRel, AccessShareLock);
}
@@ -6817,7 +6817,7 @@ ATExecAddIndex(AlteredTableInfo *tab, Relation rel,
check_rights,
false, /* check_not_in_use - we did it already */
skip_build,
- quiet);
+ quiet, false);
/*
* If TryReuseIndex() stashed a relfilenode for us, we used it for the new
@@ -14218,7 +14218,7 @@ AttachPartitionEnsureIndexes(Relation rel, Relation attachrel)
DefineIndex(RelationGetRelid(attachrel), stmt, InvalidOid,
RelationGetRelid(idxRel),
constraintOid,
- false, false, false, false, false);
+ false, false, false, false, false, false);
}
index_close(idxRel, AccessShareLock);
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 8c23ee53e2..ab96c1f2ca 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1358,7 +1358,8 @@ ProcessUtilitySlow(ParseState *pstate,
true, /* check_rights */
true, /* check_not_in_use */
false, /* skip_build */
- false); /* quiet */
+ false, /* quiet */
+ false); /* not recursing */
/*
* Add the CREATE INDEX node itself to stash right away;
diff --git a/src/include/commands/defrem.h b/src/include/commands/defrem.h
index f510f40945..e38d173c3a 100644
--- a/src/include/commands/defrem.h
+++ b/src/include/commands/defrem.h
@@ -31,7 +31,8 @@ extern ObjectAddress DefineIndex(Oid relationId,
bool check_rights,
bool check_not_in_use,
bool skip_build,
- bool quiet);
+ bool quiet,
+ bool recursing);
extern void ReindexIndex(RangeVar *indexRelation, int options);
extern Oid ReindexTable(RangeVar *relation, int options);
extern void ReindexMultipleTables(const char *objectName, ReindexObjectType objectKind,
Re: unique indexes on partitioned tables
On 2018/01/29 16:28, Amit Langote wrote:
create table p (a int, b int) partition by list (a);
create table p1 partition of p for values in (1) partition by range (b);
create table p11 partition of p1 for values from (1) to (10);
create table p2 partition of p for values in (2);create unique index on p (a);
ERROR: insufficient columns in UNIQUE constraint definition
DETAIL: UNIQUE constraint on table "p1" lacks column "b" which is part of
the partition key.It seems that after recursing to p1 which is itself partitioned,
DefineIndex() mistakenly looks for column b (which is in the p1's
partition key) in the unique key. I think that's unnecessary.
DefineIndex() should check that only once, that is, before recursing.
Hmm, scratch that...
Am I missing something?
Yes, I am.
create table p (a int, b int) partition by list (a);
create table p1 partition of p for values in (1) partition by range (b);
create table p11 partition of p1 for values from (1) to (10);
create table p12 partition of p1 for values from (10) to (20);
create table p2 partition of p for values in (2);
-- after applying my delta patch
create unique index on p (a);
insert into p values (1, 1); -- unique index p11 (a) says all fine
insert into p values (1, 10); -- unique index p12 (a) says all fine
That can't be right, because p (a) is no longer unique.
So, a unique key on a partitioned table must include the partition key
columns of *all* downstream partitioned tables, as your patch correctly
enforces. Sorry about the noise.
That said, I think that it might be a good idea to include the above
detail in the documentation of CREATE INDEX and ALTER TABLE ADD UNIQUE.
Thanks,
Amit
Re: unique indexes on partitioned tables
On 1/26/18 13:42, Peter Eisentraut wrote:
Other than that, this looks pretty good to me. A logical extension of
the previous partitioned index patch.
Moved to next CF.
Seems close to ready.
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services