Allow logical decoding on standbys

Unsurprisingly, this requires wal_level = logical to be set on the primary and
standby. The infrastructure added in 26669757b6a ensures that slots are
invalidated if the primary's wal_level is lowered.

Creating a slot on a standby waits for a xl_running_xact record to be
processed. If the primary is idle (and thus not emitting xl_running_xact
records), that can take a while.  To make that faster, this commit also
introduces the pg_log_standby_snapshot() function. By executing it on the
primary, completion of slot creation on the standby can be accelerated.

Note that logical decoding on a standby does not itself enforce that required
catalog rows are not removed. The user has to use physical replication slots +
hot_standby_feedback or other measures to prevent that. If catalog rows
required for a slot are removed, the slot is invalidated.

See 6af1793954e for an overall design of logical decoding on a standby.

Bumps catversion, for the addition of the pg_log_standby_snapshot() function.

Author: "Drouvot, Bertrand" <bertranddrouvot.pg@gmail.com>
Author: Andres Freund <andres@anarazel.de> (in an older version)
Author: Amit Khandekar <amitdkhan.pg@gmail.com> (in an older version)
Reviewed-by: Andres Freund <andres@anarazel.de>
Reviewed-by: FabrÌzio de Royes Mello <fabriziomello@gmail.com>
Reviewed-by: Amit Kapila <amit.kapila16@gmail.com>
Reviewed-By: Robert Haas <robertmhaas@gmail.com>

2 files changed, 42 insertions, 0 deletions

diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 4211d31f30..bf4c61ccfb 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -27074,6 +27074,21 @@ postgres=# SELECT '0/0'::pg_lsn + pd.segment_number * ps.setting::int + :offset
         prepared with <xref linkend="sql-prepare-transaction"/>.
        </para></entry>
       </row>
+      <row>
+       <entry role="func_table_entry"><para role="func_signature">
+        <indexterm>
+         <primary>pg_log_standby_snapshot</primary>
+        </indexterm>
+        <function>pg_log_standby_snapshot</function> ()
+        <returnvalue>pg_lsn</returnvalue>
+       </para>
+       <para>
+        Take a snapshot of running transactions and write it to WAL, without
+        having to wait bgwriter or checkpointer to log one. This is useful for
+        logical decoding on standby, as logical slot creation has to wait
+        until such a record is replayed on the standby.
+       </para></entry>
+      </row>
      </tbody>
     </tgroup>
    </table>
diff --git a/doc/src/sgml/logicaldecoding.sgml b/doc/src/sgml/logicaldecoding.sgml
index 4e912b4bd4..ebe0376e3e 100644
--- a/doc/src/sgml/logicaldecoding.sgml
+++ b/doc/src/sgml/logicaldecoding.sgml
@@ -316,6 +316,33 @@ postgres=# select * from pg_logical_slot_get_changes('regression_slot', NULL, NU
      may consume changes from a slot at any given time.
     </para>
 
+    <para>
+     A logical replication slot can also be created on a hot standby. To prevent
+     <command>VACUUM</command> from removing required rows from the system
+     catalogs, <varname>hot_standby_feedback</varname> should be set on the
+     standby. In spite of that, if any required rows get removed, the slot gets
+     invalidated. It's highly recommended to use a physical slot between the primary
+     and the standby. Otherwise, hot_standby_feedback will work, but only while the
+     connection is alive (for example a node restart would break it). Then, the
+     primary may delete system catalog rows that could be needed by the logical
+     decoding on the standby (as it does not know about the catalog_xmin on the
+     standby). Existing logical slots on standby also get invalidated if wal_level
+     on primary is reduced to less than 'logical'. This is done as soon as the
+     standby detects such a change in the WAL stream. It means, that for walsenders
+     that are lagging (if any), some WAL records up to the wal_level parameter change
+     on the primary won't be decoded.
+    </para>
+
+    <para>
+     Creation of a logical slot requires information about all the currently
+     running transactions. On the primary, this information is available
+     directly, but on a standby, this information has to be obtained from
+     primary. Thus, slot creation may need to wait for some activity to happen
+     on the primary. If the primary is idle, creating a logical slot on
+     standby may take noticeable time. This can be sped up by calling the
+     <function>pg_log_standby_snapshot</function> on the primary.
+    </para>
+
     <caution>
      <para>
       Replication slots persist across crashes and know nothing about the state