From: Tom Lane <tgl@sss.pgh.pa.us>
To: Robert Haas <robertmhaas@gmail.com>
cc: Daniel Shelepanov <deniel1495@mail.ru>,
        PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>,
        Andres Freund <andres@anarazel.de>
Subject: Re: collect_corrupt_items_vacuum.patch
In-reply-to: 
 <CA+TgmoZtw7yNZogHFrgatHnvc2rOX9hoRRq=GtUBZx9mxh86Vg@mail.gmail.com>
References: <1649062270.289865713@f403.i.mail.ru>
 <CA+TgmoZtw7yNZogHFrgatHnvc2rOX9hoRRq=GtUBZx9mxh86Vg@mail.gmail.com>
Comments: In-reply-to Robert Haas <robertmhaas@gmail.com>
	message dated "Wed, 27 Jul 2022 17:50:46 -0400"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <3107108.1658958956.1@sss.pgh.pa.us>
Date: Wed, 27 Jul 2022 17:55:56 -0400
Message-ID: <3107109.1658958956@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/3107109.1658958956%40sss.pgh.pa.us>
Precedence: bulk

Robert Haas <robertmhaas@gmail.com> writes:
> In reality, the oldest all-visible XID cannot move backward, but
> ComputeXidHorizons() lets it move backward, because it's intended for
> use by a caller who wants to mark pages all-visible, and it's only
> concerned with making sure that the value is old enough to be safe.

Right.

> And that's a problem for the way that pg_visibility is (mis-)using it.

> To say that another way, ComputeXidHorizons() is perfectly fine with
> returning a value that is older than the true answer, as long as it
> never returns a value that is newer than the new answer. pg_visibility
> wants the opposite. Here, a value that is newer than the true value
> can't do worse than hide corruption, which is sort of OK, but a value
> that's older than the true value can report corruption where none
> exists, which is very bad.

Maybe we need a different function for pg_visibility to call?
If we want ComputeXidHorizons to serve both these purposes, then it
has to always deliver exactly the right answer, which seems like
a definition that will be hard and expensive to achieve.

			regards, tom lane