Re: Drop replslot after pgstat_shutdown cause assert coredump

At Fri, 22 Oct 2021 02:10:21 +0900, Fujii Masao <masao(dot)fujii(at)oss(dot)nttdata(dot)com> wrote in
> Even in prior to the commit, pgstat_shutdown_hook() can be called
> before ProcKill() at the backend exit, so ISTM that the problem can
> be reproduced.
>
> Probably we need to make sure that pgstat_shutdown_hook() is called
> after ProcKill(), e.g., by registering pgstat_shutdown_hook() into

Considering the coming shared-memory based stats collector, pgstat
must be shutdown before shared memory shutdown. Every operation that
requires stats collector also must be shut down before the pgstat
shutdown. A naive solution would be having before-pgstat-shutdown hook
but I'm not sure it's the right direction.

> on_proc_exit_list (I'm not sure if this change is safe, though).
> Or maybe pgstat logic for replication slot drop needs to be
> overhauled.

I think we don't want to lose the stats numbers of the to-be-dropped
slot. So the slot-drop must happen before pgstat shutdown.

regards.

--
Kyotaro Horiguchi
NTT Open Source Software Center