On Mon, Jan 29, 2024 at 1:21 PM Nathan Bossart <nathandbossart@gmail.com> wrote:
> > Ah, I think this query:
> >
> > SELECT tli, start_lsn, end_lsn from pg_available_wal_summaries()
> > WHERE tli = $summarized_tli AND end_lsn > '$summarized_lsn'
> >
> > is returning more than one row in some cases. I attached a quick sketch of
> > an easy way to reproduce the issue as well as one way to fix it.
>
> The buildfarm just caught a failure with the new logging in place:
>
> https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=calliphoridae&dt=2024-01-29%2018%3A09%3A10
>
> I'm not totally sure my "fix" is correct, but I think this does confirm the
> theory.
Ah. The possibilities of ending up with TWO new WAL summaries never
occurred to me. Things that never occurred to the developer are a
leading cause of bugs, and so here.
I'm wondering if what we need to do is run pg_walsummary on both
summary files in that case. If we just pick one or the other, how do
we know which one to pick?
--
Robert Haas
EDB: http://www.enterprisedb.com