Modify the document of Logical Replication configuration settings

On Wed, Jan 18, 2023 at 07:00:43AM +0000, Takamichi Osumi (Fujitsu) wrote:

The attached patch includes below two changes for the description of
Logical Replication "Configuration Settings".

1. Add one brief description about wal_sender_timeout.
I made it similar to one other sentence for subscriber.
2. Fix a wrong GUC name "wal_receiver_retry_interval".
I think this doesn't seem to exist and would mean "wal_retrieve_retry_interval".

Kindly have a look at it.

Looks right to me, thanks!
--
Michael

On Wednesday, January 18, 2023 4:11 PM Michael Paquier <michael@paquier.xyz> wrote:

On Wed, Jan 18, 2023 at 07:00:43AM +0000, Takamichi Osumi (Fujitsu) wrote:

The attached patch includes below two changes for the description of
Logical Replication "Configuration Settings".

1. Add one brief description about wal_sender_timeout.
I made it similar to one other sentence for subscriber.
2. Fix a wrong GUC name "wal_receiver_retry_interval".
I think this doesn't seem to exist and would mean

"wal_retrieve_retry_interval".

Kindly have a look at it.

Looks right to me, thanks!

Thank you for checking, too !

Best Regards,
Takamichi Osumi

On Wed, Jan 18, 2023 at 12:31 PM Takamichi Osumi (Fujitsu)
<osumi.takamichi@fujitsu.com> wrote:

Hi, hackers

The attached patch includes below two changes for the description of
Logical Replication "Configuration Settings".

1. Add one brief description about wal_sender_timeout.
I made it similar to one other sentence for subscriber.

+   <para>
+    Logical replication walsender is also affected by
+    <link linkend="guc-wal-sender-timeout"><varname>wal_sender_timeout</varname></link>.
+   </para>

Looks fine. Adding something like [1]diff --git a/doc/src/sgml/config.sgml b/doc/src/sgml/config.sgml index 89d53f2a64..6f9509267c 100644 --- a/doc/src/sgml/config.sgml +++ b/doc/src/sgml/config.sgml @@ -4326,7 +4326,8 @@ restore_command = 'copy "C:\\server\\archivedir\\%f" "%p"' # Windows <para> Terminate replication connections that are inactive for longer than this amount of time. This is useful for - the sending server to detect a standby crash or network outage. + the sending server to detect a standby crash or logical replication + subscriber crash or network outage. If this value is specified without units, it is taken as milliseconds. The default value is 60 seconds. A value of zero disables the timeout mechanism. in wal_sender_timeout GUC's
description might be a good idea just to give specific information
that the logical replication subscribers too get affected. Perhaps,
it's not required since the postgres glossary wraps logical
replication subscriber under standby anyway -
https://www.postgresql.org/docs/devel/glossary.html. To me personally,
the typical notion of standby is the one connected to primary via
streaming replication.

2. Fix a wrong GUC name "wal_receiver_retry_interval".
I think this doesn't seem to exist and would mean "wal_retrieve_retry_interval".

Good catch. +1.

[1]
diff --git a/doc/src/sgml/config.sgml b/doc/src/sgml/config.sgml
index 89d53f2a64..6f9509267c 100644
--- a/doc/src/sgml/config.sgml
+++ b/doc/src/sgml/config.sgml
@@ -4326,7 +4326,8 @@ restore_command = 'copy
"C:\\server\\archivedir\\%f" "%p"'  # Windows
        <para>
         Terminate replication connections that are inactive for longer
         than this amount of time. This is useful for
-        the sending server to detect a standby crash or network outage.
+        the sending server to detect a standby crash or logical replication
+        subscriber crash or network outage.
         If this value is specified without units, it is taken as milliseconds.
         The default value is 60 seconds.
         A value of zero disables the timeout mechanism.

--
Bharath Rupireddy
PostgreSQL Contributors Team
RDS Open Source Databases
Amazon Web Services: https://aws.amazon.com

On Wed, Jan 18, 2023 at 02:04:16PM +0530, Bharath Rupireddy wrote:

[1]
diff --git a/doc/src/sgml/config.sgml b/doc/src/sgml/config.sgml
index 89d53f2a64..6f9509267c 100644
--- a/doc/src/sgml/config.sgml
+++ b/doc/src/sgml/config.sgml
@@ -4326,7 +4326,8 @@ restore_command = 'copy
"C:\\server\\archivedir\\%f" "%p"'  # Windows
<para>
Terminate replication connections that are inactive for longer
than this amount of time. This is useful for
-        the sending server to detect a standby crash or network outage.
+        the sending server to detect a standby crash or logical replication
+        subscriber crash or network outage.
If this value is specified without units, it is taken as milliseconds.
The default value is 60 seconds.
A value of zero disables the timeout mechanism.

Perhaps we could do that, I am not sure whether this brings much in
this section, though.
--
Michael

Hi,

On Thursday, January 19, 2023 3:14 PM Michael Paquier <michael@paquier.xyz> wrote:

On Wed, Jan 18, 2023 at 02:04:16PM +0530, Bharath Rupireddy wrote:

[1]
diff --git a/doc/src/sgml/config.sgml b/doc/src/sgml/config.sgml index
89d53f2a64..6f9509267c 100644
--- a/doc/src/sgml/config.sgml
+++ b/doc/src/sgml/config.sgml
@@ -4326,7 +4326,8 @@ restore_command = 'copy
"C:\\server\\archivedir\\%f" "%p"'  # Windows
<para>
Terminate replication connections that are inactive for longer
than this amount of time. This is useful for
-        the sending server to detect a standby crash or network outage.
+        the sending server to detect a standby crash or logical replication
+        subscriber crash or network outage.
If this value is specified without units, it is taken as milliseconds.
The default value is 60 seconds.
A value of zero disables the timeout mechanism.

Perhaps we could do that, I am not sure whether this brings much in this
section, though.

This might increase comprehensiveness of the doc slightly.

If we want to do this, it might be better to
add this kind of additions to other parameters such as
wal_receiver_timeout, wal_retrieve_retry_interval
and wal_receiver_status_interval, too.

BTH, thank you for having taken care of my patch, Michael-san!

Best Regards,
Takamichi Osumi