Re: 9.2.3 upgrade reduced pgbench performance by 60%

From: Jeff Janes <jeff(dot)janes(at)gmail(dot)com>
To: Colin Currie <colin(at)marketsworld(dot)com>
Cc: "pgsql-performance(at)postgresql(dot)org" <pgsql-performance(at)postgresql(dot)org>
Subject: Re: 9.2.3 upgrade reduced pgbench performance by 60%
Date: 2013-03-31 17:16:50
Message-ID: CAMkU=1ycPhXvEKzz8urr8-ih91UXsv3CC9sGMBxpiXMZA-KmCA@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-performance

On Monday, March 25, 2013, Colin Currie wrote:

> Hi,
>
> I recently upgraded PostgreSQL from 9.0.12 to 9.2.3 on a test server to
> compare performance. I'm using pgbench to measure which results in around a
> 60% reduction.
>
> The non-default configuration remains identical between versions except
> archive_command (different location) and custom_variable_classes (no longer
> supported) and are detailed are below. Is there some updated default
> configuration that I'm missing? Perhaps it's because of the new cascading
> replication feature? I've tried tweaking the memory settings to no avail.
>
> The Linux server is on a cloud and has 4GB RAM and 2 CPUs and the same
> server is running both master and slave (these are separate in production).

What does your recovery.conf look like? What if you don't run the slave at
all, then how do they compare?

Cheers,

Jeff

In response to

Browse pgsql-performance by date

  From Date Subject
Next Message Tom Lane 2013-04-01 00:52:59 Re: query plan estimate
Previous Message Kelphet Xiong 2013-03-31 04:00:02 Re: Question about postmaster's CPU usage