Horton

Re: Checkpointer sync queue fills up / loops around pg_usleep() are bad

Michael Paquier <michael@paquier.xyz>

View thread View on postgr.es
From: Michael Paquier <michael@paquier.xyz>
To: Andres Freund <andres@anarazel.de>
Cc: pgsql-hackers@postgresql.org, Thomas Munro <thomas.munro@gmail.com>
Date: 2022-02-27T09:10:45Z
Lists: pgsql-hackers 
On Sat, Feb 26, 2022 at 01:39:42PM -0800, Andres Freund wrote:
> I suspect the easiest is to just convert that usleep to a WaitLatch(). That'd
> require adding a new enum value to WaitEventTimeout in 14. Which probably is
> fine?

We've added wait events in back-branches in the past, so this does not
strike me as a problem as long as you add the new entry at the end of
the enum, while keeping things ordered on HEAD.  In recent memory, I
think that only some of the extensions published by PostgresPro rely
on the enums in this area. 
--
Michael

Commits

  1. Fix waiting in RegisterSyncRequest().

  2. Wake up for latches in CheckpointWriteDelay().