Re: Out of shared mem on new box with more mem, 9.1.5 -> 9.1.6 - Mailing list pgsql-performance

From Maciek Sakrejda
Subject Re: Out of shared mem on new box with more mem, 9.1.5 -> 9.1.6
Date
Msg-id CAOtHd0DhwDqSNDz9kePNPHevqdyCBU7M6=--nGWXa1nJSjTMOA@mail.gmail.com
Whole thread Raw
In response to Re: Out of shared mem on new box with more mem, 9.1.5 -> 9.1.6  (Martin French <Martin.French@romaxtech.com>)
Responses Re: Out of shared mem on new box with more mem, 9.1.5 -> 9.1.6  (Martin French <Martin.French@romaxtech.com>)
List pgsql-performance
On Wed, Oct 17, 2012 at 1:53 AM, Martin French
<Martin.French@romaxtech.com> wrote:

Thanks for your response.

> What are the settings for:
> work_mem
 100MB

> maintenance_work_mem
 64MB

> How many concurrent connections are there?
~20

> Have you ran explain analyze on the query that doesn't crash (i.e the old
> box) to get the exact execution plan?

I can try that in the morning, but I didn't think this was relevant. I
know cost estimates can be off, but can the plan actually change
between a vanilla explain and an explain analyze?

> Has the DB been vacuum analyzed?

Not outside of autovacuum, no, but it's actually a former replica of
the first database (sorry I neglected to mention this earlier).


pgsql-performance by date:

Previous
From: Samuel Gendler
Date:
Subject: Re: have: seq scan - want: index scan
Next
From: Martin French
Date:
Subject: Re: Out of shared mem on new box with more mem, 9.1.5 -> 9.1.6