> I've read the first link but not the second link. Your hardware could be undersized, but it is
> hard to say without knowing the actual number of concurrent connections. Also, what exactly are
> you experiencing when you say that your server is not able to keep up? What does a vmstat 1 10
> show while you are experiencing slow preformance. Do you have sufficently aggressive autovaccum
> setting? Also, have you adjusted any of your postgresql.conf parameters?
I greatly increased the shmmax to approx 448MB (machine has 521K installed)
max_connections = 30
I've changed the shared_buffers to 32768.
work_mem to 2048.
maintenance_work_mem to 32768
max_fsm_page to 40_000
and the last one, that I'm not too sure about, is
effective_cache_size to 52_000 (which is the max I can set anything at)
So, those are all the settings.
It's generally running with about 16K of swap in use, but I can't say what it is
that is swapped out.
But top typically shows these processes are taking up less than 55% of the
available RAM for each process and typically has the CPU load up around 2.10.
vmstat is pretty quiet. Nothing out of the ordinary that would say it's running
poorly. si/so are 0 99% of the time.
So, is this good? bad? dunno?
