On Wed, 5 Apr 2023 at 01:41, Greg Stark <stark@mit.edu> wrote:
>
> On Wed, 29 Mar 2023 at 17:48, Justin Pryzby <pryzby@telsasoft.com> wrote:
> >
> > The patch still occasionally fails its tests under freebsd.
> > https://cirrus-ci.com/github/postgresql-cfbot/postgresql/commitfest/42/3358
>
> I wonder if some other test is behaving differently on FreeBSD and
> leaving behind a prepared transaction or a zombie session in some idle
> state or something like that? Is there anything (aside from
> autovacuum) connecting or running in the background in the test
> environment that could be creating a transaction id and holding back
> snapshot xmin?
Ok, I've reproduced this here by running the tests under meson. It
doesn't look like it's platform dependent.
It seems under meson the different test suites are run in parallel or
at least isolation/deadlock-parallel are still running stuff when the
regression checks are running. If that's not expected then maybe
something's not behaving as expected? I've attached pg_stat_activity
from during the test run.
Regardless it shows these tests are obviously not robust enough to
include as they would break for anyone running make installcheck on a
non-idle cluster.
That's fine, as I said, the tests were just there to give a reviewer
more confidence and I think it's fine to just not include them in the
commit.
--
greg