From: | Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> |
---|---|
To: | Masahiko Sawada <sawada(dot)mshk(at)gmail(dot)com> |
Cc: | depesz(at)depesz(dot)com, pgsql-bugs mailing list <pgsql-bugs(at)postgresql(dot)org> |
Subject: | Re: WAL segments removed from primary despite the fact that logical replication slot needs it. |
Date: | 2022-11-15 05:11:44 |
Message-ID: | CAA4eK1K=N7Ft1PJUHGqVMgtV_GGv_svoz4WzOvaZnmCEV43kyQ@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-bugs |
On Tue, Nov 15, 2022 at 7:50 AM Masahiko Sawada <sawada(dot)mshk(at)gmail(dot)com> wrote:
>
> On Fri, Nov 11, 2022 at 11:50 PM hubert depesz lubaczewski
> <depesz(at)depesz(dot)com> wrote:
> >
> > 2022-11-10 20:59:55 UTC 103A7/7B666388 focal14 pgoutput logical 16607 canvas f t 10414 \N 3241434855 1039D/83825958 1039D/8DECBA18
> > 2022-11-10 21:00:54 UTC 103A7/88225C28 focal14 pgoutput logical 16607 canvas f t 10414 \N 3241434855 1039D/83825958 1039D/911A8DB0
> > 2022-11-10 21:01:53 UTC 103A7/9E77EF90 focal14 pgoutput logical 16607 canvas f f \N \N 3241434855 1039D/83825958 1039D/911A8DB0
> > 2022-11-10 21:02:52 UTC 103A7/B45ECEA8 focal14 pgoutput logical 16607 canvas f f \N \N 3241434855 1039D/83825958 1039D/911A8DB0
> > 2022-11-10 21:03:52 UTC 103A7/C4355F60 focal14 pgoutput logical 16607 canvas f t 21748 \N 3241443528 1039D/83825958 1039D/955633D0
> > 2022-11-10 21:04:51 UTC 103A7/D5A6FDD8 focal14 pgoutput logical 16607 canvas f t 21748 \N 3241443528 1039D/83825958 1039D/96453F38
> > 2022-11-10 21:05:50 UTC 103A7/E2C7F590 focal14 pgoutput logical 16607 canvas f f \N \N 3241443528 1039D/83825958 1039D/96453F38
> > 2022-11-10 21:06:49 UTC 103A7/F28B2000 focal14 pgoutput logical 16607 canvas f f \N \N 3241443528 1039D/83825958 1039D/96453F38
> > 2022-11-10 21:07:48 UTC 103A8/3828D20 focal14 pgoutput logical 16607 canvas f f \N \N 3241443528 1039D/83825958 1039D/96453F38
> > 2022-11-10 21:08:47 UTC 103A8/15BBF088 focal14 pgoutput logical 16607 canvas f f \N \N 3241443528 1039D/83825958 1039D/96453F38
>
> It happened twice that the walsender process was active for at least
> one minute while not advancing restart_lsn and exited. Could you share
> the server logs from around 2022-11-10 20:59:55 to 21:08:47 so that we
> can find out what the wal sender processes were doing?
>
> Also, you may have logs for "failed to increase restart lsn: proposed
> %X/%X, after %X/%X, current candidate %X/%X, current after %X/%X,
> flushed up to %X/%X" and "got new restart lsn %X/%X at %X/%X". If so,
> please share it as well.
>
> I've tried to reproduce this issue with a similar setting in a loop
> but could not.
>
> Given that you have the log "attempting to remove WAL segments older
> than log file 000000000001039D0000008A", the last WAL file we need to
> retain calculated by KeepLogSeg() was wrong.
>
> KeepLogSeg() retreated the last segment that we need to retain while
> considering replication slots' minimum LSN (and wal_keep_segments) but
> it seemed not to work fine. The possibilities are (1) we got the wrong
> slots' minimum LSN that is newer than a slot's LSN (ie, retreated
> logSegNo but it's still newer than required WAL files) or (2) we got
> an invalid slots's minimum LSN (ie, didn't retreated at all). If we
> know what the RedoRecPtr was, we can calculate _logSegNo that has not
> been retreated by KeepLogSeg().
>
I was checking but didn't find if we have RedoRecPtr-related
information available. I think it could be valuable to have RedoRecPtr
and probably replicationSlotMinLSN, lastRemovedSegNo, etc. kind of
information available via some view (either in pg_stat_bgwriter or a
similar view) to debug this kind of issue.
--
With Regards,
Amit Kapila.
From | Date | Subject | |
---|---|---|---|
Next Message | hubert depesz lubaczewski | 2022-11-15 07:55:36 | Re: WAL segments removed from primary despite the fact that logical replication slot needs it. |
Previous Message | Amit Kapila | 2022-11-15 04:51:18 | Re: WAL segments removed from primary despite the fact that logical replication slot needs it. |