Scott Marlowe wrote:
Your entire disk io subsystem is a pair of hard drives. I'm assuming
software RAID.
Correct.
The time that this query takes is not the issue, rather it is the impact
that it has on the server - effectively killing it for the 40 seconds due to
the heavy disk access.
You either need to invest more into your drive subsystem so it can
handle parallel load better, or you need to create a slave db with
slony or londiste so that the ugly queries hit the slave.
____
I take your point Scott. But short of the hardware upgrade, is there a
way to effectively throttle this query? If I reduce maintenance_work_mem
then the database dump/restore is slower but there is less overall
impact on the server. Is there some equivalent parameter on the server
to throttle general queries? It would be unfortunate if all queries
slowed down a bit, but a better outcome than having the entire server
hang for 40 seconds.
--
Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general
