* Matt Magoffin (postgresql.org@xxxxxxx) wrote: > Yep, we've got 16GB to work with here. I should have also mentioned the > architecture in my original post, sorry. SELECT version() returns this: > > PostgreSQL 8.3.5 on x86_64-unknown-linux-gnu, compiled by GCC gcc (GCC) > 4.1.2 20070626 (Red Hat 4.1.2-14) Does the result from 'free' look reasonable on this box? How many PG processes are running, generally? Do you see the PG process running this query growing to fill the available memory? That query really didn't look like it was chewing up all that much memory to me.. Have you tried reducing your work_mem setting to see if that makes the errors go away? It might cause a different plan which might be less efficient for some queries, not sure if you have anything which would be dramatically affected by that.. Do the row counts in the plan match up with what you'd expect? Can you provide output from 'explain analyze' on a successful run? Thanks, Stephen
Attachment:
signature.asc
Description: Digital signature
