Re: Changing the state of data checksums in a running cluster

From: Daniel Gustafsson <daniel(at)yesql(dot)se>
To: PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org>
Cc: Zsolt Parragi <zsolt(dot)parragi(at)percona(dot)com>, Alexander Lakhin <exclusion(at)gmail(dot)com>, Michael Paquier <michael(at)paquier(dot)xyz>, Tomas Vondra <tomas(at)vondra(dot)me>, Andres Freund <andres(at)anarazel(dot)de>, Heikki Linnakangas <hlinnaka(at)iki(dot)fi>, Bernd Helmle <mailings(at)oopsware(dot)de>, Michael Banck <mbanck(at)gmx(dot)net>
Subject: Re: Changing the state of data checksums in a running cluster
Date: 2026-08-28 11:45:10
Message-ID: 9A5B206A-DA3B-48AD-AAC0-A7EF22E332BB@yesql.se
Views: Whole Thread | Raw Message | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Attached is a WIP revert of online checksums after the recent discussions [0].
It's not a 100% revert, though close to it, as it keeps a few smallers fixes
which have merit on their own IMHO, the commit message outline these. This was
a first pass to get started on this, but it requires a little more study to
make sure it's actually as complete and correct as it should be. It passes a
check-world locally on macOS and Linux which seems like a good place to submit
to -hackers for referece.

While the fate of this commit may not have been decided yet, I think it's best
to prepare for all scenarios since creating a revert of a feature which has
been in the tree for some time also takes a bit of time.

If it is reverted I argue for only reverting in 19 and leaving it in master to
ensure it can be worked on. If no one fixes the open item in a timely manner
then we can look at reverting from master as well, but I think optimizing for
the best case scenario makes sense.

--
Daniel Gustafsson

[0] https://postgr.es/m/CA+Tgmob9NY6m0YNFTQ4nFH2d0iC9SQRruDYxfndGKKzh8OC80w@mail.gmail.com

Attachment Content-Type Size
v1-0001-Revert-online-data-checksums-transitions.patch application/octet-stream 288.8 KB

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Alexander Lakhin 2026-08-28 12:00:00 Re: Internal error codes triggered by regression tests and user queries, take 2
Previous Message Zsolt Parragi 2026-08-28 11:16:55 Re: Offline data checksum changes can cause incorrect checksum state on standbys