On Nov 9, 2004, at 2:01 PM, Shane | SkinnyCorp wrote:

Thanks in advance for anything you can do to help.

The real issue is this, we have THE SAME queries taking anywhere from .001 - 90.0 seconds... the server is using 98% of the available RAM at all times (because of the persistant connections via php), and I don't know what to do. Every time I change a

I'd recommend strongly ditching the use of pconnect and use pgpool + regular connect. It is a terrific combination that provides pool connections like how you'd think they shoudl work (a pool of N connections to PG shared by Y processes instead of a 1:1 mapping).

curiously, have you noticed any pattern to the slowdown?
It could be induced by a checkpoint or vacuum.

Are you swapping at all?

Are your PHP scripts leaking at all, etc.?

Your load average is high, how does your CPU idle look (if load is high, and the cpus are pretty idle that is an indicator of being IO bound).

good luck.


---------------------------(end of broadcast)--------------------------- TIP 4: Don't 'kill -9' the postmaster

Reply via email to