Re: [PATCH] Clarify the behavior of the system when approaching XID wraparound (stop telling users to "vacuum that database in single-user mode")

Justin Pryzby <pryzby@telsasoft.com>

From: Justin Pryzby <pryzby@telsasoft.com>
To: Aleksander Alekseev <aleksander@timescale.com>
Cc: pgsql-hackers@lists.postgresql.org
Date: 2023-01-25T23:28:43Z
Lists: pgsql-hackers
On Mon, Jan 16, 2023 at 03:50:57PM +0300, Aleksander Alekseev wrote:
> Hi hackers,
> 
> > The proposed patchset changes the documentation and the error messages
> > accordingly, making them less misleading. 0001 corrects the
> > documentation but doesn't touch the code. 0002 and 0003 correct the
> > messages shown when approaching xidWrapLimit and xidWarnLimit
> > accordingly.
> 
> A colleague of mine, Oleksii Kliukin, pointed out that the
> recommendation about running VACUUM in a single-user mode is also
> outdated, as it was previously reported in [1]. I didn't believe it at
> first and decided to double-check:

and again at:
https://www.postgresql.org/message-id/flat/CA%2BTgmoYPfofQmRtUan%3DA3aWE9wFsJaOFr%2BW_ys2pPkNPr-2FZw%40mail.gmail.com#e7dd25fdcd171c5775f3f9e3f86b2082

> Unfortunately the [1] discussion went nowhere. 

likewise...

> So I figured it would be appropriate to add corresponding changes to
> the proposed patchset since it's relevant and is registered in the CF
> app already. PFA patchset v2 which now also includes 0004.
> 
> [1]:
> https://www.postgresql.org/message-id/flat/CAMT0RQTmRj_Egtmre6fbiMA9E2hM3BsLULiV8W00stwa3URvzA%40mail.gmail.com

I suggest to resend this with a title like the 2021 thread [1] (I was
unable to find this just now when I looked)
| doc: stop telling users to "vacuum that database in single-user mode"

And copy the participants of the previous two iterations of this thread.

-- 
Justin



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.