Re: 16.14 regression: startup process self-deadlocks during multixact WAL replay in RecordNewMultiXact

> On 27 May 2026, at 17:33, Olegs Germanovs  wrote:
> 
> After upgrading from 16.13 to 16.14, archive recovery of a basebackup
>   hangs indefinitely during multixact WAL replay.

Hi Olegs!

Thanks for the detailed report! Your analysis of the self-deadlock is spot on.

The fix for this problem has already been committed to REL_16_STABLE as 42a3194e5483 [0].
It was discussed on the pgsql-bugs thread "BUG #19490: Streaming standby on 16.14 stops
applying WAL on MultiXactOffsetSLRU when primary is 16.8" [1].

Please let us know if you still observe the problem or any other unusual behavior.


Best regards, Andrey Borodin.



[0] https://git.postgresql.org/cgit/postgresql.git/commit/?h=REL_16_STABLE&id=42a3194e548349b658a808347df3d3d5e6b968af
[1] https://www.postgresql.org/message-id/flat/46FE61C9-F273-45FD-BED7-0F8CDA6EB992%40yandex-team.ru#69de260aa4dc6a1882732d466783afc9