Horton

Re: New instability in stats regression test

Andres Freund <andres@anarazel.de>

View thread View on postgr.es
From: Andres Freund <andres@anarazel.de>
To: Bharath Rupireddy <bharath.rupireddyforpostgres@gmail.com>
Cc: Michael Paquier <michael@paquier.xyz>, Tom Lane <tgl@sss.pgh.pa.us>, pgsql-hackers@lists.postgresql.org
Date: 2023-11-27T17:52:59Z
Lists: pgsql-hackers 
Hi,

On 2023-11-27 15:49:01 +0530, Bharath Rupireddy wrote:
> On Mon, Nov 27, 2023 at 8:26 AM Michael Paquier <michael@paquier.xyz> wrote:
> > but two resets done on the same shared stats type would still
> > be prone to race conditions without all the previous activity done in
> > the tests (like pg_stat_wal).
> 
> Can running stats.sql in non-parallel mode help stabilize the tests as-is?

I think that'd be a cure *way* worse than the disease. Having concurrent stats
activity isn't exactly uncommon. And because of checkpoints, autovacuum etc,
you'd still have rare situations of concurrency.

Greetings,

Andres Freund

Commits

  1. Remove test for pg_stat_reset_shared() in stats.sql

  2. Add ability to reset all shared stats types in pg_stat_reset_shared()

  3. Add tests for pg_stat_io