Date: Sat, 18 Sep 2021 16:35:51 -0500
From: Justin Pryzby <pryzby@telsasoft.com>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Bruce Momjian <bruce@momjian.us>, pgsql-hackers@postgresql.org
Subject: Re: PG 14 release notes, first draft
Message-ID: <20210918213551.GV831@telsasoft.com>
References: <20210918204605.GU831@telsasoft.com>
 <1500489.1631999739@sss.pgh.pa.us>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <1500489.1631999739@sss.pgh.pa.us>
User-Agent: Mutt/1.9.4 (2018-02-28)
Archived-At: 
 <https://www.postgresql.org/message-id/20210918213551.GV831%40telsasoft.com>
Precedence: bulk

On Sat, Sep 18, 2021 at 05:15:39PM -0400, Tom Lane wrote:
> Justin Pryzby <pryzby@telsasoft.com> writes:
> > I think the release notes for the autovacuum item (which was first reverted and
> > then partially un-reverted) should say something like "Partitioned tables are
> > now included in pg_stat_all_tables":
> > | e1efc5b465 Keep stats up to date for partitioned tables
> 
> Hmm.  If I'm reading the commit message properly, the actual change there
> is not that, but that analyze count and last analyze time are now tracked
> correctly for partitioned tables.  Might be worth mentioning, not sure.

The reverted patch to autoanalyze included partitioned tables in
pg_stat_all_tables, and the revert specifically avoided changing that (to avoid
a catbump).  But last_analyzed and analyze_count were always shown as "0".  So
the e1 commit addresses that by tracking that information and showing correct
value instead of always 0.

The relevant portion starts here:
https://www.postgresql.org/message-id/flat/202108161700.d4eh6a7n2lki%40alvherre.pgsql#b2e426eb19dbbddee0adf9bb1bcbbcf1

I suggest that this *should* be included in the release notes, since I
specifically requested that partitioned tables be included in 2018.

> In 20200418050815(dot)GE26953(at)telsasoft(dot)com I wrote:
> |This patch includes partitioned tables in pg_stat_*_tables, which is great; I
> |complained awhile ago that they were missing [0].  It might be useful if that
> |part was split out into a separate 0001 patch (?).
> | [0] https://www.postgresql.org/message-id/20180601221428.GU5164%40telsasoft.com

Also, I've patched my analyze script to use that field (same as for
nonpartitioned tables) rather than needing to do a subquery involving
max(last_analyzed) of the partitions.  Since it's still needed to manually
analyze parent tables.

-- 
Justin