Re: pg15b2: large objects lost on upgrade

Peter Geoghegan <pg@bowt.ie>

From: Peter Geoghegan <pg@bowt.ie>

To: Tom Lane <tgl@sss.pgh.pa.us>

Cc: Andres Freund <andres@anarazel.de>, Robert Haas <robertmhaas@gmail.com>, "Jonathan S. Katz" <jkatz@postgresql.org>, Noah Misch <noah@leadboat.com>, Andrew Dunstan <andrew@dunslane.net>, Bruce Momjian <bruce@momjian.us>, Michael Paquier <michael@paquier.xyz>, Justin Pryzby <pryzby@telsasoft.com>, "pgsql-hackers@postgresql.org" <pgsql-hackers@postgresql.org>, Shruthi Gowda <gowdashru@gmail.com>

Date: 2022-08-03T20:42:22Z

Lists: pgsql-hackers

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

Revert recent changes to 002_pg_upgrade.pl.
- 6f7e7d0c482d 15.0 landed
- 87e22f675fd8 16.0 landed
Revise test case added in 43746996399541ecb5c7b188725a5f097c15ceae.
- d92f2bc0dae3 15.0 landed
- 212bdc0cbc32 16.0 landed
Use TRUNCATE to preserve relfilenode for pg_largeobject + index.
- bbe08b8869bd 16.0 landed
- 4ab5dae9472c 15.0 landed
Preserve relfilenode of pg_largeobject and its index across pg_upgrade.
- a2996478c32d 15.0 landed
- d498e052b4b8 16.0 landed
Have VACUUM warn on relfrozenxid "in the future".
- e83ebfe6d767 15.0 cited
Set relfrozenxid to oldest extant XID seen by VACUUM.
- 0b018fabaaba 15.0 cited
pg_upgrade: Preserve relfilenodes and tablespace OIDs.
- 9a974cbcba00 15.0 cited
Perform a lot more sanity checks when freezing tuples.
- 699bf7d05c68 11.0 cited

On Wed, Aug 3, 2022 at 1:34 PM Tom Lane <tgl@sss.pgh.pa.us> wrote:
> That doesn't seem like it'd be all that thorough: we expect VACUUM
> to skip pages whenever possible.  I'm also a bit concerned about
> the expense, though admittedly this test is ridiculously expensive
> already.

I bet the SKIP_PAGES_THRESHOLD stuff will be enough to make VACUUM
visit every heap page in practice for a test case like this. That is
all it takes to be able to safely advance relfrozenxid to whatever the
oldest extant XID happened to be. However, I'm no fan of the
SKIP_PAGES_THRESHOLD behavior, and already have plans to get rid of it
-- so I wouldn't rely on that continuing to be true forever.

It's probably not really necessary to have that kind of coverage in
this particular test case. VACUUM will complain about weird
relfrozenxid values in a large variety of contexts, even without
assertions enabled. Mostly I was just saying: if we really do need
test coverage of relfrozenxid in this context, then VACUUM is probably
the way to go.

-- 
Peter Geoghegan