Horton

Re: Tracking last scan time

Michael Paquier <michael@paquier.xyz>

View thread View on postgr.es
From: Michael Paquier <michael@paquier.xyz>
To: Dave Page <dpage@pgadmin.org>
Cc: Robert Treat <rob@xzilla.net>, Andres Freund <andres@anarazel.de>, Matthias van de Meent <boekewurm+postgres@gmail.com>, PostgreSQL Developers <pgsql-hackers@lists.postgresql.org>, Vik Fearing <vik@postgresfriends.org>
Date: 2022-11-07T07:54:07Z
Lists: pgsql-hackers

Attachments

On Thu, Nov 03, 2022 at 04:44:16PM -0400, Dave Page wrote:
> Here's a patch to fix this issue. Many thanks to Peter Eisentraut who
> figured it out in a few minutes after I spent far too long looking down
> rabbit holes in entirely the wrong place.

FWIW, all the other areas of pgstatfuncs.c manipulate timestamptz
fields with a style like the attached.  That's a nit, still per the
role of consistency with the surroundings..

Anyway, it seems to me that a regression test is in order before a
scan happens just after the relation creation, and the same problem
shows up with last_idx_scan.
--
Michael

Commits

  1. Fix initialization of pg_stat_get_lastscan()

  2. pgstat: Track time of the last scan of a relation

  3. Have GetCurrentTransactionStopTimestamp() set xactStopTimestamp if unset