Re: Distributing data over "spindles" even on AWS EBS, (followup to the work queue saga)

From: Justin Pryzby <pryzby(at)telsasoft(dot)com>
To: Gunther <raj(at)gusw(dot)net>
Cc: pgsql-performance(at)postgresql(dot)org
Subject: Re: Distributing data over "spindles" even on AWS EBS, (followup to the work queue saga)
Date: 2019-03-14 15:54:56
Message-ID: 20190314155456.GG6030@telsasoft.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-performance

On Wed, Mar 13, 2019 at 02:44:10PM -0400, Gunther wrote:
> You see that I already did a lot to balance IO out to many different
> tablespaces that's why there are so many volumes.

I wonder if it wouldn't be both better and much easier to have just 1 or 2
tablespaces and combine drives into a single LVM VG and do something like
lvcreate --stripes

> I am not sure my autovacuum setup is working right though. I wonder if there
> isn't some autovacuum statistics which I can query that would give me
> confidence that it is actually running?

pg_stat_all_tables

Did you do this ?
ALTER TABLE ... SET (autovacuum_vacuum_scale_factor=0.001, autovacuum_vacuum_threshold=1);

Justin

In response to

Browse pgsql-performance by date

  From Date Subject
Next Message Jeff Janes 2019-03-14 17:50:07 Re: impact of auto explain on overall performance
Previous Message Jeremy Schneider 2019-03-14 15:11:20 Re: Distributing data over "spindles" even on AWS EBS, (followup to the work queue saga)