Horton

Re: recovering from "found xmin ... from before relfrozenxid ..."

Stephen Frost <sfrost@snowman.net>

View thread View on postgr.es
From: Stephen Frost <sfrost@snowman.net>
To: Robert Haas <robertmhaas@gmail.com>
Cc: "pgsql-hackers@postgresql.org" <pgsql-hackers@postgresql.org>
Date: 2020-07-13T21:28:25Z
Lists: pgsql-hackers 
Greetings,

* Robert Haas (robertmhaas@gmail.com) wrote:
> - Do people think it would me smart/good/useful to include something
> like this in PostgreSQL?

Absolutely, yes.

> - If so, how? I would propose a new contrib module that we back-patch
> all the way, because the VACUUM errors were back-patched all the way,
> and there seems to be no advantage in making people wait 5 years for a
> new version that has some kind of tooling in this area.

While I agree that this would be a good and useful new contrib module to
have, I don't think it would be appropriate to back-patch it into PG
formally.

Unfortunately, that gets into the discussion that's cropped up on a few
other threads of late- that we don't have a good place to put extensions
which are well maintained/recommended by core PG hackers, and which are
able to work with lots of different versions of PG, and are versioned
and released independently of PG (and, ideally, built for all the
versions of PG that we distribute through our packages).

Given the lack of such a place today, I'd at least suggest starting with
proposing it as a new contrib module for v14.

> - Any ideas for additional things we should include, or improvements
> on the sketch above?

Not right off-hand, but will think about it, there could certainly be a
lot of very interesting tools in such a toolbox.

Thanks!

Stephen

Commits

  1. Fix wrong data table horizon computation during backend startup.

  2. Centralize horizon determination for temp tables, fixing bug due to skew.

  3. pg_surgery: Try to stabilize regression tests.

  4. New contrib module, pg_surgery, with heap surgery functions.

  5. Set cutoff xmin more aggressively when vacuuming a temporary table.

  6. snapshot scalability: Don't compute global horizons while building snapshots.

  7. Introduce vacuum errcontext to display additional information.