Re: BUG #9447: still bug in autovacuum after upgrading to 9.1.11?

From: Alvaro Herrera <alvherre(at)2ndquadrant(dot)com>
To: pchan(at)contigo(dot)com
Cc: pgsql-bugs(at)postgresql(dot)org
Subject: Re: BUG #9447: still bug in autovacuum after upgrading to 9.1.11?
Date: 2014-03-05 18:39:11
Message-ID: 20140305183911.GO4759@eldon.alvh.no-ip.org
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-bugs

pchan(at)contigo(dot)com wrote:

> PostgreSQL version: 9.1.11
> Operating system: Linux version 2.6.18-371.3.1.el5
> Description:
>
> After upgrading to 9.1.11, I vacuum the database as instructed in the
> release note (i.e. set vacuum_freeze_table_age=0, then run vacuum). I also
> run the utility as described in this blog:
>
> http://www.depesz.com/2013/12/06/what-does-fix-vacuums-tests-to-see-whether-it-can-update-relfrozenxid-really-mean/

Uh, that article has very odd formatting and is rather hard to read and
also long, so I'm not gonna read it. It might contain errors, if you're
going from what it says.

> However, after a month upgrading to 9.1.11, I run the utility again and
> still find rows with strange xmax. The current txid in the system is
> 1922768288 but I find several rows in a table with xmax <= 5000 and I am
> sure some of them were updated after upgrade to 9.1.11. For example:

It's normal to have low numbers used in Xmax. This just means that they
used a multixact instead of a plain transaction ID. You can check that
by using the pageinspect extension and seeing the flag set in the
t_infomask field.

I'm not saying there is no bug here, only that you have provided no
evidence that there is.

--
Álvaro Herrera http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services

In response to

Browse pgsql-bugs by date

  From Date Subject
Next Message Alvaro Herrera 2014-03-05 19:22:25 Re: BUG #9223: plperlu result memory leak
Previous Message pchan 2014-03-05 18:07:58 BUG #9447: still bug in autovacuum after upgrading to 9.1.11?