Re: Many queries stuck for waiting MultiXactOffsetControlLock on PostgreSQL 9.6.1 replica - Mailing list pgsql-bugs

From David G. Johnston
Subject Re: Many queries stuck for waiting MultiXactOffsetControlLock on PostgreSQL 9.6.1 replica
Date
Msg-id CAKFQuwYJn8ZkLwiEfx27w4X4ONTg4-KF=YvQum-ZYaQwAj3OcA@mail.gmail.com
Whole thread Raw
In response to Many queries stuck for waiting MultiXactOffsetControlLock on PostgreSQL 9.6.1 replica  (Rui An <rueian.huang@gmail.com>)
Responses Re: Many queries stuck for waiting MultiXactOffsetControlLock on PostgreSQL 9.6.1 replica  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-bugs
On Saturday, July 31, 2021, Rui An <rueian.huang@gmail.com> wrote:

I have tried use pg_terminate_backend to kill the stuck prcoesses, but no help. They still stuck on waiting for MultiXactOffsetControlLock even if the pg_terminate_backend(pid) returns true.

All the true is telling you is your process sent the requested signal.  It, and related functions, provide no insight as to if/how receivers of said signal may have responded.

As for the question at hand - the axiom is “if it ain’t broke don’t fix it”…but as your system is basically broke you need to fix it by running a supported version of the software.

And you should also be prepared for the possibility that 9.6 is discontinued before there is a fix in which case you’d have to upgrade to v10 (at least) to get said fix.

David J.

pgsql-bugs by date:

Previous
From: Andrey Borodin
Date:
Subject: Re: Many queries stuck for waiting MultiXactOffsetControlLock on PostgreSQL 9.6.1 replica
Next
From: Kyotaro Horiguchi
Date:
Subject: Re: BUG #17103: WAL segments are not removed after exceeding max_slot_wal_keep_size