Horton

Re: Test slots invalidations in 035_standby_logical_decoding.pl only if dead rows are removed

Michael Paquier <michael@paquier.xyz>

View thread View on postgr.es
From: Michael Paquier <michael@paquier.xyz>
To: Alexander Lakhin <exclusion@gmail.com>
Cc: "Drouvot, Bertrand" <bertranddrouvot.pg@gmail.com>, "Yu Shi (Fujitsu)" <shiy.fnst@fujitsu.com>, Tom Lane <tgl@sss.pgh.pa.us>, Robert Haas <robertmhaas@gmail.com>, Andres Freund <andres@anarazel.de>, Thomas Munro <thomas.munro@gmail.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2024-01-10T04:26:14Z
Lists: pgsql-hackers 
On Tue, Jan 09, 2024 at 08:00:00PM +0300, Alexander Lakhin wrote:
> Thus just adding FREEZE is not enough, seemingly. It makes me wonder if
> 0174c2d21 should be superseded by a patch like discussed (or just have
> autovacuum = off added)...

Adding an extra FREEZE offers an extra insurance, so I don't see why
it would be a problem to add it to stabilize the horizon conflicts on
the standbys.

> 09.01.2024 07:59, Michael Paquier wrote:
> Bertrand, thank you for updating the patch!
> 
> Michael, it definitely increases stability of the test (tens of iterations
> with 20 tests in parallel performed successfully), although I've managed to
> see another interesting failure (twice):
> 13      #   Failed test 'activeslot slot invalidation is logged with vacuum on pg_class'
> 13      #   at t/035_standby_logical_decoding.pl line 227.

Something I'd like to confirm here: you still see this failure with
the patch, but without an extra FREEZE, right?  If we do both, the
test would get more stable, wouldn't it?
--
Michael

Commits

  1. Improve stability of recovery test 035_standby_logical_decoding