Horton

Re: New WAL record to detect the checkpoint redo location

Michael Paquier <michael@paquier.xyz>

View thread View on postgr.es
From: Michael Paquier <michael@paquier.xyz>
To: Robert Haas <robertmhaas@gmail.com>
Cc: Andres Freund <andres@anarazel.de>, Dilip Kumar <dilipbalaut@gmail.com>, Heikki Linnakangas <hlinnaka@iki.fi>, pgsql-hackers <pgsql-hackers@postgresql.org>
Date: 2023-10-19T05:53:12Z
Lists: pgsql-hackers 
On Wed, Oct 18, 2023 at 10:24:50AM -0400, Robert Haas wrote:
> I added a variant of this test case. Here's v10.

+-- Verify that an XLOG_CHECKPOINT_REDO record begins at precisely the redo LSN
+-- of the checkpoint we just performed.
+SELECT redo_lsn FROM pg_control_checkpoint() \gset
+SELECT start_lsn = :'redo_lsn'::pg_lsn AS same_lsn, resource_manager,
+    record_type FROM pg_get_wal_record_info(:'redo_lsn');
+ same_lsn | resource_manager |   record_type   
+----------+------------------+-----------------
+ t        | XLOG             | CHECKPOINT_REDO
+(1 row)

Seems fine to me.  Thanks for considering the idea.
--
Michael

Commits

  1. During online checkpoints, insert XLOG_CHECKPOINT_REDO at redo point.

  2. Unify two isLogSwitch tests in XLogInsertRecord.