Horton

Re: HOT chain validation in verify_heapam()

Peter Geoghegan <pg@bowt.ie>

View thread View on postgr.es
From: Peter Geoghegan <pg@bowt.ie>
To: Andres Freund <andres@anarazel.de>
Cc: pgsql-hackers@lists.postgresql.org, Himanshu Upadhyaya <upadhyaya.himanshu@gmail.com>, Robert Haas <robertmhaas@gmail.com>, Aleksander Alekseev <aleksander@timescale.com>, PostgreSQL Hackers <pgsql-hackers@postgresql.org>
Date: 2022-11-10T02:35:12Z
Lists: pgsql-hackers 
On Wed, Nov 9, 2022 at 6:10 PM Andres Freund <andres@anarazel.de> wrote:
> And thinking about it, it'd be quite bad if the horizon worked that way. You can easily construct a workload where every single xid would "skewer" some chain, never allowing the horizon to be raised.

Your whole scenario is one involving a insert of a tuple by XID 10,
which is then updated by XID 5 -- a lower XID. Obviously that's
possible, but it's relatively rare. I have to imagine that the vast
majority of updates affect tuples inserted by an XID before the XID of
the updater.

My use of the term "skewer" was limited to updates that look like
that. So I don't know what you mean about never allowing the horizon
to be raised.

-- 
Peter Geoghegan

Commits

  1. amcheck: Generalize one of the recently-added update chain checks.

  2. amcheck: Tighten up validation of redirect line pointers.

  3. amcheck: Fix verify_heapam for tuples where xmin or xmax is 0.

  4. amcheck: Fix a few bugs in new update chain validation.

  5. Fix new test case to work on (some?) big-endian architectures.

  6. Don't test HEAP_XMAX_INVALID when freezing xmax.