Re: System overload / context switching / oom, 8.3

On 2/2/2010 1:11 PM, Rob wrote:
>
> postgresql.conf:
> max_connections = 1100
> shared_buffers = 800MB
> max_prepared_transactions = 0
> work_mem = 32MB
> maintenance_work_mem = 64MB
> max_fsm_pages = 3300000
> max_fsm_relations = 10000
> vacuum_cost_delay = 50ms
> bgwriter_delay = 150ms
> bgwriter_lru_maxpages = 250
> bgwriter_lru_multiplier = 2.5
> wal_buffers = 8MB
> checkpoint_segments = 32
> checkpoint_timeout = 5min
> checkpoint_completion_target = 0.9
> effective_cache_size = 5000MB
> default_statistics_target = 100
> log_min_duration_statement = 1000
> log_checkpoints = on
> log_connections = on
> log_disconnections = on
> log_temp_files = 0
> track_counts = on
> autovacuum = on
> log_autovacuum_min_duration = 0


Ok, seriously this time.

 > work_mem = 32MB
 > maintenance_work_mem = 64MB


if you have lots and lots of connections, you might need to cut these down?

 > effective_cache_size = 5000MB

I see your running a 32bit, but with bigmem support, but still, one
process is limited to 4gig.  You'd make better use of all that ram if
you switched to 64bit.  And this cache, I think, would be limited to 4gig.

The oom-killer is kicking in, at some point, so somebody is using too
much ram.  There should be messages or logs or something, right?  (I've
never enabled the oom stuff so dont know much about it).  But the log
messages might be helpful.

Also, do you know what the oom max memory usage is set to?  You said:
"oom_adj -17.  vm_overcommit_memory set to 2, but at this time
vm_overcommit_ratio was still at 50 (has since been changed to 90,
should this be 100?)"

but I have no idea what that means.

-Andy