Horton

Re: [PATCH] Clarify the behavior of the system when approaching XID wraparound

Peter Geoghegan <pg@bowt.ie>

View thread View on postgr.es
From: Peter Geoghegan <pg@bowt.ie>
To: John Naylor <john.naylor@enterprisedb.com>
Cc: Aleksander Alekseev <aleksander@timescale.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>, Pavel Borisov <pashkin.elfe@gmail.com>
Date: 2023-05-02T03:03:47Z
Lists: pgsql-hackers 
On Mon, May 1, 2023 at 7:55 PM Peter Geoghegan <pg@bowt.ie> wrote:
> Obviously there are certain things that can hold back OldestMXact by a
> wildly excessive amount. But I don't think that there is anything that
> can hold back OldestMXact by a wildly excessive amount that won't more
> or less do the same thing to OldestXmin.

Actually, it's probably possible for a transaction that only has a
virtual transaction ID to call MultiXactIdSetOldestVisible(), which
will then have the effect of holding back OldestMXact without also
holding back OldestXmin (in READ COMMITTED mode).

Will have to check to make sure, but that won't happen today.

--
Peter Geoghegan

Commits

  1. Reword messages about impending (M)XID exhaustion.

  2. Talk about assigning, rather than generating, new MultiXactIds.

  3. Update the documentation on recovering from (M)XID exhaustion.