Michael Kohl wrote:
> HDD: 2x 120 GB OCZ Vertex 2 SSD; RAID 1
>
As a general warning here, as far as I know the regular Vertex 2 SSD
doesn't cache writes properly for database use. It's possible to have a
crash that leaves the database corrupted, if the drive has writes queued
up in its cache. The Vertex 2 Pro resolves this issue with a supercap,
you may have a model with concerns here. See
http://wiki.postgresql.org/wiki/Reliable_Writes for more information.
In addition to the log_checkpoints suggestion already made, I'd also
recommend turning on log_lock_waits and log_temp_files on your server.
All three of those--checkpoints, locks, and unexpected temp file
use--can cause the sort of issue you're seeing. Well, not locks so much
given you're seeing heavy disk I/O, but it's good to start logging those
issues before they get bad, too.
--
Greg Smith 2ndQuadrant US greg@2ndQuadrant.com Baltimore, MD
PostgreSQL Training, Services, and 24x7 Support www.2ndQuadrant.us
"PostgreSQL 9.0 High Performance": http://www.2ndQuadrant.com/books