From: | Jim Nasby <Jim(dot)Nasby(at)BlueTreble(dot)com> |
---|---|
To: | Peter Eisentraut <peter_e(at)gmx(dot)net>, pgsql-hackers <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: statistics for shared catalogs not updated when autovacuum is off |
Date: | 2016-02-02 01:38:32 |
Message-ID: | 56B00898.4010409@BlueTreble.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
On 2/1/16 7:20 PM, Peter Eisentraut wrote:
> That's probably right. Even with autovacuum on, the statistics for
> shared catalogs do not appear as updated right away. That is, if you
> run VACUUM and then look at pg_stat_sys_tables right away, you will see
> the stats for shared catalogs to be slightly out of date until the
> minutely autovacuum check causes them to update.
>
> So the problem exists in general, but the autovacuum launcher papers
> over it every minute.
I suspect the issue is in backend_read_statsfile(). Presumably the if
just needs a call to AutoVacuumingActive() added:
> /*
> * Autovacuum launcher wants stats about all databases, but a shallow read
> * is sufficient.
> */
> if (IsAutoVacuumLauncherProcess())
> pgStatDBHash = pgstat_read_statsfiles(InvalidOid, false, false);
> else
> pgStatDBHash = pgstat_read_statsfiles(MyDatabaseId, false, true);
The interesting thing is that we always start the launcher one time, to
protect against wraparound, but apparently that path doesn't call
anything that calls backend_read_statsfile() (which is static).
--
Jim Nasby, Data Architect, Blue Treble Consulting, Austin TX
Experts in Analytics, Data Architecture and PostgreSQL
Data in Trouble? Get it in Treble! http://BlueTreble.com
From | Date | Subject | |
---|---|---|---|
Next Message | David G. Johnston | 2016-02-02 01:42:19 | Re: "using previous checkpoint record at" maybe not the greatest idea? |
Previous Message | Euler Taveira | 2016-02-02 01:37:34 | Re: Raising the checkpoint_timeout limit |