Re: 8.3 beta testing suggestions welcome

From: "Kevin Grittner" <Kevin(dot)Grittner(at)wicourts(dot)gov>
To: "Greg Smith" <gsmith(at)gregsmith(dot)com>
Cc: <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: 8.3 beta testing suggestions welcome
Date: 2007-11-30 23:17:01
Message-ID: 4750458D.EE98.0025.0@wicourts.gov
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

>>> On Tue, Aug 21, 2007 at 1:12 PM, in message
<Pine(dot)GSO(dot)4(dot)64(dot)0708211316210(dot)16511(at)westnet(dot)com>, Greg Smith
<gsmith(at)gregsmith(dot)com> wrote:
> My suggestion for a starting 8.3 configuration for you would be adjusting
> these settings as follows:
>
> shared_buffers=1GB
> checkpoint_segments = 50

I'm not into the most meaningful tests yet, but I'm trying to
garner what useful information I can from the load of the 8.2.5
database into a server running the cvs tip copy taken a few hours
ago. After letting the cache settle in, while in the middle of a
COPY of a large table, I see the attached through a couple
checkpoint cycles. The pattern is pretty consistent with this
over other the other checkpoint cycles I've examined.

Note that the system can sustain over 20000 writes per second
without the disk being a bottleneck. The current tip behavior is
smoother than 8.2.x without background writer, for sure. It still
isn't as smooth as 8.2.x with an aggressive backgroud writer, at
least for loading a dump. Note how it lounges around at as low as
5500 writes per second for prolonged periods with occassional
spikes into the 55000 range, causing I/O wait time, although these
only last a few seconds. Without more rigorous tests for
comparison, I'd say that this is probably a net loss for this
operation -- I seem to recall seeing a pretty steady 20000 bo under
8.2.x.

More info to follow as testing progresses.

-Kevin

Attachment Content-Type Size
vmstat.sample1.out application/octet-stream 27.5 KB

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Usama Dar 2007-11-30 23:21:25 Re: Release Note Changes
Previous Message Tom Lane 2007-11-30 23:10:15 Re: CommandCounterIncrement versus plan caching