From: | Kyotaro Horiguchi <horikyota(dot)ntt(at)gmail(dot)com> |
---|---|
To: | tgl(at)sss(dot)pgh(dot)pa(dot)us |
Cc: | masahiko(dot)sawada(at)2ndquadrant(dot)com, masao(dot)fujii(at)oss(dot)nttdata(dot)com, pgsql-hackers(at)lists(dot)postgresql(dot)org |
Subject: | Re: Race condition in SyncRepGetSyncStandbysPriority |
Date: | 2020-04-13 06:34:24 |
Message-ID: | 20200413.153424.688075393737383141.horikyota.ntt@gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
At Mon, 13 Apr 2020 15:31:01 +0900 (JST), Kyotaro Horiguchi <horikyota(dot)ntt(at)gmail(dot)com> wrote in
> At Sat, 11 Apr 2020 18:30:30 -0400, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote in
> > The point that I was trying to make originally is that it seems quite
> > insane to imagine that a walsender's sync_standby_priority value is
> > somehow more stable than the very existence of the process. Yet we
> > only require a walsender to lock its own mutex while claiming or
> > disowning its WalSnd entry (by setting or clearing the pid field).
> > So I think it's nuts to define those fields as being protected by
> > the global SyncRepLock.
>
> Right. FWIW, furthermore, even SyncRepConfig->syncrep_method can be
> inconsistent among walsenders. I haven't thought that it can be
> relied on as always consistent and it is enough that it makes a
> consistent result only while the setting and the set of walsenders is
> stable.
Yes, the sentene "and (I haven't thought that) it is enough .." is a
mistake of "and I have thought that it is enough that..".
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center
From | Date | Subject | |
---|---|---|---|
Next Message | Kyotaro Horiguchi | 2020-04-13 06:45:56 | Re: pg_validatebackup -> pg_verifybackup? |
Previous Message | Kyotaro Horiguchi | 2020-04-13 06:31:01 | Re: Race condition in SyncRepGetSyncStandbysPriority |