Tom,
You hit the nail on the head with what we did. We did two things and it
made a world of difference.
We moved from RAID 5 SCSII drives to our EMC SAN RAID 10 and adjusted
the checkpoint segments from 15 to 30.
The bottleneck disappeared totally and actually have never seen better
performance.
Two questions:
What are the implications to further increasing the checkpoint so say
40?
Also how does 8.0's background-writer feature work and what are going to
benefits?
Brian
-----Original Message-----
From: Tom Lane [mailto:tgl@sss.pgh.pa.us]
Sent: Tuesday, October 26, 2004 5:59 PM
To: Brian Maguire
Cc: pgsql-general@postgresql.org
Subject: Re: [GENERAL] what could cause inserts getting queued up and db
locking??
Brian Maguire <bmaguire@vantage.com> wrote:
>> We though there might be locking, but noticed that there were not any
>> queries in wait mode indicating that no statements were blocked by
>> another statement's lock.
In that case it's not a locking problem, but just a resource-saturation
problem. I'm wondering if you are maxing out your disk drives'
throughput.
Are the slowdowns correlated with checkpoints? (Watch to see if there
is a postmaster child process spawned for checkpointing when it
happens.) Fooling with checkpoint intervals might help some, though
I suspect the only real answer will be 8.0's background-writer feature.
regards, tom lane