* Matt Magoffin (postgresql.org@msqr.us) wrote:
> We've been having persistent out-of-memory errors occur in our production
> 8.3 deployment, which is now running 8.3.5. I'm not sure the query here is
> the cause of the problem, but this is our most-recent example which
> triggered an out-of-memory error for us.
erm.. How much memory do you have in the system?
> Perhaps our configuration needs tweaking somewhere...
>
> shared_buffers = 4000MB
I hope you've got a fair bit more than 4G of memory if you're going to
use 4G for shared buffers... Once that memory is dedicated to shared
buffers it's not going to be available for other usage.
> #temp_buffers = 8MB
> work_mem = 128MB
> maintenance_work_mem = 256MB
> #max_stack_depth = 2MB
Those are bumped up from the defaults, but don't look too bad to me. Do
realize that 'work_mem' is the amount PG is allowed to use for each
action and that complex queries can end up using many times that amount
(but I doubt that's the real issue here..).
Thanks,
Stephen