Re: BUG #16799: postgresql log issue(last completed transaction was at log time) - Mailing list pgsql-bugs

From Tom Lane
Subject Re: BUG #16799: postgresql log issue(last completed transaction was at log time)
Date
Msg-id 2457719.1609432047@sss.pgh.pa.us
Whole thread Raw
In response to Re: BUG #16799: postgresql log issue(last completed transaction was at log time)  (oleg9301 xz <oleg9301@gmail.com>)
List pgsql-bugs
oleg9301 xz <oleg9301@gmail.com> writes:
> чт, 31 дек. 2020 г. в 18:21, Tom Lane <tgl@sss.pgh.pa.us>:
>> PG Bug reporting form <noreply@postgresql.org> writes:
>>> After recovering postgresql from barman, in version 12.5 I don't have
>>> next string
>>> "last completed transaction was at log time %s"
>>> In postgresql 11.9-1.pgdg100+1 everything fine.

>> I don't see any reason to believe there's a bug here.  That message
>> would only appear if at least one transaction commit or abort record was
>> replayed from WAL, and it's quite possible there wouldn't have been.

> There is always transations to commit, I(barman) do pg_basebackup once a
> day, and recieve wal with pg_receivewal(standby slot). And everything is
> fine, i have all data on OLAP DB(restore 6 times in day)
> (all WAL files playing correctly).

Hmm.  I poked a bit into the code, and I see what is different in v12:
commit 7acf8a876 ("Make crash recovery ignore recovery target settings")
caused all of recoveryStopsAfter() to be skipped when
!ArchiveRecoveryRequested ... including the part that updates
recoveryLastXTime.

Fujii-san, Peter, was that actually intended?  The commit message
doesn't mention it.  It's certainly arguable that this is cosmetic
and we don't care enough to contort the code to keep the old behavior
here.

            regards, tom lane



pgsql-bugs by date:

Previous
From: oleg9301 xz
Date:
Subject: Re: BUG #16799: postgresql log issue(last completed transaction was at log time)
Next
From: Dana Burd
Date:
Subject: Re: BUG #16797: EXTRACT(EPOCH FROM timestamp) is not using local timezone