Re: Performance degradation on concurrent COPY into a single relation in PG16. - Mailing list pgsql-hackers

From Tom Lane
Subject Re: Performance degradation on concurrent COPY into a single relation in PG16.
Date
Msg-id 1578568.1694037713@sss.pgh.pa.us
Whole thread Raw
In response to Re: Performance degradation on concurrent COPY into a single relation in PG16.  (Andres Freund <andres@anarazel.de>)
Responses Re: Performance degradation on concurrent COPY into a single relation in PG16.
List pgsql-hackers
Andres Freund <andres@anarazel.de> writes:
> On 2023-08-16 13:15:46 +0200, Alvaro Herrera wrote:
>> Since the wins from this patch were replicated and it has been pushed, I
>> understand that this open item can be marked as closed, so I've done
>> that.

> Thanks!

It turns out that this patch is what's making buildfarm member
chipmunk fail in contrib/pg_visibility [1].  That's easily reproduced
by running the test with shared_buffers = 10MB.  I didn't dig further
than the "git bisect" result:

$ git bisect bad
82a4edabd272f70d044faec8cf7fd1eab92d9991 is the first bad commit
commit 82a4edabd272f70d044faec8cf7fd1eab92d9991
Author: Andres Freund <andres@anarazel.de>
Date:   Mon Aug 14 09:54:03 2023 -0700

    hio: Take number of prior relation extensions into account

but I imagine the problem is that the patch's more aggressive
relation-extension heuristic is causing the table to have more
pages than the test case expects.  Is it really a good idea
for such a heuristic to depend on shared_buffers?  If you don't
want to change the heuristic then we'll have to find a way to
tweak this test to avoid it.

            regards, tom lane

[1] https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=chipmunk&dt=2023-09-06%2014%3A14%3A51



pgsql-hackers by date:

Previous
From: Melanie Plageman
Date:
Subject: Re: Eliminate redundant tuple visibility check in vacuum
Next
From: Jacob Champion
Date:
Subject: Re: [PoC] Federated Authn/z with OAUTHBEARER