> On 13 Jan 2024, at 17:00, Alexander Lakhin <exclusion@gmail.com> wrote:
> I suspected that this failure was caused by autovacuum, and I've managed to
> reproduce it without Valgrind or slowing down a machine.
This might be due to the fact that the cleanup codepath to remove the flag when
there is no login event trigger doesn't block on locking pg_database to avoid
stalling connections. There are no guarantees when the flag is cleared, so a
test like this will always have potential to be flaky it seems.
--
Daniel Gustafsson