auto truncate/vacuum full

From: JC Praud <brutaltruth42(at)gmail(dot)com>
To: pgsql-general(at)postgresql(dot)org
Subject: auto truncate/vacuum full
Date: 2009-10-27 14:15:22
Message-ID: 8dc7b0c00910270715w89a9c52l4b4ce78b7b7d7541@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

Hi all,

I've got a weird thing on one of my databases this night:

- I've a monthly partition for storing activity logs defined as this:
- mother log table
- one child partition for each month

- Last friday I dumped the last month partition, and tried to truncate it,
which locked lots of queries because the mother table was also locked. So I
killed the TRUNCATE and performed a DELETE on my partition

- Last night the database locked. pg_log full of messages about insert into
the mother table waiting for a lock.

- After about 40min, the waiting queries acquired their locks and ran.

- At the same time, my munin graphs show a decrease of my dbase size, of
about the size of my partition, which pg_(total_)relation_size says has a
zero size now.

- No slow query mentionning truncate nor vacuum full.

- According to pg_stat_user_tables an autovacuum was run on the table just
after the unlock.

So my question are: can the autovacuum daemon perform vacuum full ? Or
another internal postgres process ? Could it come from the TRUNCATE I run
and canceled 4 days before ?

I'm running postgresql-8.3.7

Regards,

--
JC
Ph'nglui mglw'nafh Cthulhu n'gah Bill R'lyeh Wgah'nagl fhtagn!

Responses

Browse pgsql-general by date

  From Date Subject
Next Message Joshua Berry 2009-10-27 14:27:17 Absolute value of intervals
Previous Message Merlin Moncure 2009-10-27 14:07:27 Re: Implementing Frontend/Backend Protocol TCP/IP