Horton

Re: BUG #19490: Streaming standby on 16.14 stops applying WAL on MultiXactOffsetSLRU when primary is 16.8

Andrey Borodin <x4mmm@yandex-team.ru>

View thread
From: Andrey Borodin <x4mmm@yandex-team.ru>
To: Ayush Tiwari <ayushtiwari.slg01@gmail.com>
Cc: Michael Paquier <michael@paquier.xyz>, Radim Marek <radim@boringsql.com>, Heikki Linnakangas <hlinnaka@iki.fi>, Marko Tiikkaja <marko@joh.to>, PostgreSQL mailing lists <pgsql-bugs@lists.postgresql.org>
Date: 2026-05-26T08:41:03Z
Lists: pgsql-hackers 

> On 26 May 2026, at 13:30, Ayush Tiwari <ayushtiwari.slg01@gmail.com> wrote:
> 
> Releasing MultiXactOffsetSLRULock around SimpleLruWriteAll() preserves
> the flush-before-physical-check rule while avoiding the self-deadlock.

I think we don't need to release lock, we just need to acquire it later, as it is done
in 17+ branches.

FWIW I'm working on buildfarm module that will recovery regress WAL from
REL_x_0 through replay by REL_x_STABLE.


Best regards, Andrey Borodin.