Re: WAL_LOG CREATE DATABASE strategy broken for non-standard page layouts

From: Melanie Plageman <melanieplageman(at)gmail(dot)com>
To: Michael Paquier <michael(at)paquier(dot)xyz>
Cc: Zhijie Hou <houzhijie22(at)gmail(dot)com>, Robert Haas <robertmhaas(at)gmail(dot)com>, Rogers Wang <rogers(dot)ww(at)qq(dot)com>, pgsql-hackers(at)lists(dot)postgresql(dot)org, Matthias van de Meent <boekewurm+postgres(at)gmail(dot)com>, Álvaro Herrera <alvherre(at)kurilemu(dot)de>, Andres Freund <andres(at)anarazel(dot)de>
Subject: Re: WAL_LOG CREATE DATABASE strategy broken for non-standard page layouts
Date: 2026-11-20 18:08:24
Message-ID: CAAKRu_YdgS113uEtf2DO7LszB5oFF41bR_Y4mShHTPwy0Z4xfw@mail.gmail.com
Views: Whole Thread | Raw Message | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Thu, Nov 19, 2026 at 7:01 PM Michael Paquier <michael(at)paquier(dot)xyz> wrote:
>
> On Tue, Nov 17, 2026 at 11:20:54PM +0800, Zhijie Hou wrote:
> > I think we also need to increase shared_buffers (e.g., to 16MB), because
> > $node->init(allows_streaming => 1) sets shared_buffers to 1MB internally,
> > causing the pg_class pages to be evicted from shared_buffers before CREATE
> > TABLE runs, so the test could not reach the old bug path.
>
> If we care about this level of tuning, I'd suspect that just bumping
> shared_buffers if the wal_consistency_checking switch is enabled
> would be fine to do instead of having a buildfarm member that uses a
> custom configuration file. wal_consistency_checking consumes a lot of
> resources, so this would not hurt more. Your ideas around lifting
> some of the masking of the VM pages would make that more valuable in
> the long run, as least it seems so to me.

We actually needed the increased shared_buffers to preserve the
regression coverage of the original test unrelated to
wal_consistency_checking.

- Melanie

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Greg Burd 2026-11-20 18:26:13 Re: Tepid: selective index updates for heap relations
Previous Message Ayush Tiwari 2026-11-20 17:15:07 Re: Add a pg_wal_preallocate() SQL function to eagerly create future WAL segments