Re: set autovacuum=off - Mailing list pgsql-performance

From Alessandro Gagliardi
Subject Re: set autovacuum=off
Date
Msg-id CAAB3BB+PKcm0i0s7yz00r8G_6u5dtV8y7W9wW2kD0H0svjbG-A@mail.gmail.com
Whole thread Raw
In response to Re: set autovacuum=off  (Peter van Hardenberg <pvh@pvh.ca>)
Responses Re: set autovacuum=off  (Robert Haas <robertmhaas@gmail.com>)
List pgsql-performance
I'm unable to make sense of pg_locks. The vast majority are locktype='transactionid', mode='ExclusiveLock', granted=t. There are some 'relation' locks with mode='RowExclusiveLock' and fewer with 'AccessShareLock'. I have no idea what I should be looking for here.

On Thu, Feb 23, 2012 at 10:42 AM, Peter van Hardenberg <pvh@pvh.ca> wrote:
On Thu, Feb 23, 2012 at 10:38 AM, Alessandro Gagliardi
<alessandro@path.com> wrote:
> around the same time as disabling auto-vacuum, so that could account for the
> coincidental speed up). I'm not sure what else I could be doing wrong. It's
> definitely better than it was a few days ago, but I still see "LOG:
> duration: 77.315 ms statement: COMMIT" every minute or two.
>

Have you considered that you may have lock contention? Sampling
pg_locks may be illuminating; based on your description the lock
contention would be intermittent, so I wouldn't trust an n=1 test.

-p

--
Peter van Hardenberg
San Francisco, California
"Everything was beautiful, and nothing hurt." -- Kurt Vonnegut

pgsql-performance by date:

Previous
From: Tom Lane
Date:
Subject: Re: Disable-spinlocks while compiling postgres 9.1 for ARM Cortex A8
Next
From: Alessandro Gagliardi
Date:
Subject: Re: set autovacuum=off