Horton

Re: Recent pg_rewind test failures in buildfarm

Michael Paquier <michael@paquier.xyz>

View thread View on postgr.es
From: Michael Paquier <michael@paquier.xyz>
To: Bertrand Drouvot <bertranddrouvot.pg@gmail.com>
Cc: Alexander Lakhin <exclusion@gmail.com>, Tom Lane <tgl@sss.pgh.pa.us>, pgsql-hackers@lists.postgresql.org
Date: 2025-04-23T04:55:31Z
Lists: pgsql-hackers 
On Tue, Apr 22, 2025 at 06:22:51AM +0000, Bertrand Drouvot wrote:
> Yeah, unless that might come from fc415edf8ca but I don't think that's the case.

I've been eyeing this whole area of the code for a few hours to
convince myself on HEAD, and I cannot really find a defect directly
related to it.

I am wondering if we should do a PGSTAT_BACKEND_FLUSH_WAL in the WAL
sender, but the end of any transaction happening in a logical WAL
sender would make sure that this happens on a periodic basis.

Anyway, sorry for make everybody waiting here.  I've now removed the
assertion down to v15.
--
Michael

Commits

  1. Remove assertion based on pending_since in pgstat_report_stat()