On Tue, Feb 2, 2010 at 12:11 PM, Rob <rclemley@yahoo.com> wrote:
> pg 8.3.9, Debian Etch, 8gb ram, quadcore xeon, megaraid (more details at end)
> ~240 active databases, 800+ db connections via tcp.
>
> Everything goes along fairly well, load average from 0.5 to 4.0. Disk
> IO is writing about 12-20 MB every 4 or 5 seconds. Cache memory about
> 4gb. Then under load, we see swapping and then context switch storm and
> then oom-killer.
SNIP
> postgresql.conf:
> max_connections = 1100
> work_mem = 32MB
32MB * 1000 = 32,000MB... And that's if you max out connections and
they each only do 1 sort. If you're running many queries that run > 1
sorts it'll happen a lot sooner.
Either drop max connections or work_mem is what I'd do to start with.
If you have one or two reporting apps that need it higher, then set it
higher for just those connections / users.