On Tue, Jun 22, 2021 at 10:33 AM Thomas Munro <thomas.munro@gmail.com> wrote:
>
> While scanning for assertion failures on the build farm that don't
> appear to have been discussed, I found this[1] in
> 010_truncate_publisher.log on the 13 branch:
>
> TRAP: FailedAssertion("tupdesc->tdrefcount <= 0", File:
> "/home/bf/build/buildfarm-desmoxytes/REL_13_STABLE/pgsql.build/../pgsql/src/backend/access/common/tupdesc.c",
> Line: 321)
>
I guess this could be similar to what we see at:
https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=skink&dt=2021-06-15%2020%3A49%3A26
We have discussed this in another thread at:
https://www.postgresql.org/message-id/648020.1623854904%40sss.pgh.pa.us
The reason why I think it is the same is that assertion failure shown
in the report is from function FreeTupleDesc() which we can call from
pgoutput.c while processing the invalidation. Ideally, we shouldn't
call invalidation before initializing the tuple conversion map for
partitions but in some rare cases, that was happening which we have
fixed in commit
(https://git.postgresql.org/gitweb/?p=postgresql.git;a=commitdiff;h=357cb8f07f95665ea533ff534821c22c35b01288).
I see this report is from 16th June 2021 and the commit is on 18th
June 2021. So, I am hoping this should have been fixed but if we see
it again then probably we need to investigate it further.
--
With Regards,
Amit Kapila.