Re: pgsql: Replace existing durable_rename_excl() calls with durable_rename - Mailing list pgsql-committers

On Wed, Apr 27, 2022 at 11:28:18PM -0400, Tom Lane wrote:
> Assert'ing that an external (filesystem) condition holds seems like a bad
> idea.  See buildfarm.

Yes, I saw the failures:
https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=kestrel&dt=2022-04-28%2002%3A13%3A27
https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=rorqual&dt=2022-04-28%2002%3A14%3A08
https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=calliphoridae&dt=2022-04-28%2002%3A59%3A26

All of them are on 025_stuck_on_old_timeline.pl when the WAL receiver
writes a TLI history file when fetching it from a primary, meaning
that we need to think harder about this case as we would overwrite an
existing TLI file once we use durable_rename() over
durable_rename_excl().

I don't want to keep the buildfarm unstable, so I'll revert for now.
--
Michael