Re: Large empty table, balanced INSERTs and DELETEs, not being vacuumed

From: Michael Paquier <michael(dot)paquier(at)gmail(dot)com>
To: Vick Khera <vivek(at)khera(dot)org>
Cc: Jason Dusek <jason(dot)dusek(at)gmail(dot)com>, "pgsql-general(at)postgresql(dot)org" <pgsql-general(at)postgresql(dot)org>
Subject: Re: Large empty table, balanced INSERTs and DELETEs, not being vacuumed
Date: 2016-10-22 07:11:46
Message-ID: CAB7nPqR3w4qjDTgvu2BMumRxOQHr3BT4WLYHma+7OzsVGGyj+Q@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

On Sat, Oct 22, 2016 at 6:02 AM, Vick Khera <vivek(at)khera(dot)org> wrote:
> On Fri, Oct 21, 2016 at 4:53 PM, Jason Dusek <jason(dot)dusek(at)gmail(dot)com> wrote:
>> This is really only a temporary fix, though. We can have a cron job running
>> in the background running TRUNCATE ONLY ... but this seems like the kind of
>> thing that auto-vacuuming should have handled for us, before the problem got
>> “too large”. Are there auto-vacuum settings that we can set, globally or on
>> the table, to address this situation?
>
> Did auto-vacuum actually succeed in vacuuming this table? Check your
> logs. You may need to make auto vacuum more log-verbose first.

Yeah. If you are using 9.5 or newer versions, you could set
log_autovacuum_min_duration to 0 for this relation and avoid a lot of
noise in your logs. pg_stat_user_tables and pg_stat_all_tables also
contain information regarding the last time autovacuum has been run on
a relation.
--
Michael

In response to

Browse pgsql-general by date

  From Date Subject
Next Message Alexander Farber 2016-10-22 11:12:13 Re: Selecting records with highest timestamp - for a join
Previous Message Michael Paquier 2016-10-22 07:04:15 Re: Replication rolling back to normal.