On Mon, Oct 09, 2023 at 12:20:18PM -0400, Tom Lane wrote:
> There will be a window where the worker has logged "database
> "noconndb" is not currently accepting connections" but hasn't yet
> exited, so that conceivably this query could see a positive count.
I don't think that's possible here. The check on datallowconn is done
before a backend calls pgstat_bestart() which would make its backend
entry reported to pg_stat_activity. So there is no window where a
backend would be in pg_stat_activity if this check fails.
> We could just drop this test, reasoning that the appearance of
> the error message is sufficient evidence that the right thing
> happened. (If the failed worker is still around, it won't break
> the remaining tests AFAICS.) Or we could convert this to a
> poll_query_until() loop.
Saying that, I'm OK with just dropping this query, as it could also be
possible that one decides that calling pgstat_bestart() before the
datallowconn check is a good idea for a reason or another.
--
Michael