On 30/03/2010 11:27 PM, Lothar Bongartz wrote:
> For this reason the database is stalling from time to time and I have to
> restart the server.
Are those stalls accompanied by any mention of checkpoints in the logs?
The default PostgreSQL configuration isn't tuned for larger workloads -
it's very minimalist. You probably need to increase your
checkpoint_segments . It may also be a good idea to look at shared_buffers .
Before just re-starting the server, you should look at `iostat',
`vmstat', `top', `iotop' or other suitable performance monitoring tools
and see if the server's actually DOING anything. You'll probably find
it's busy with disk I/O and not really stalled at all.
--
Craig Ringer
