Thanks to all who replied, mostly confirming my own thoughts, but adding a few wrinkles I hadn't thought of or realised. I was pretty sure already that lack of ram/heavy swapping was the issue. Now thanks to a member of this list who emailed me offlist I have 256M of RAM instad of 64M, and the machine seems to be performing much much better. It hasn't had enough time in the new configuaration to test the theory that torrentflux may have a memory leak, but I should know by tomorrow morning.
Thanks again. On Fri, 03 Mar 2006 13:26:28 +1300 Volker Kuhlmann <[EMAIL PROTECTED]> wrote: > Clear case of an I/O-bound machine: <10% CPU used, >90% time wasted > waiting for I/O. > > uptime counts processes ready-to-run as "load", hence the big numbers > when the CPU seemingly does nothing. > > Also, 60M RAM, 250M swap used - the thing's creaking at the seams. > It's unresponsive because it can't swap in your command shell... (or > anything else, for that matter). > Safe bet it's swapping itself to death. > > More RAM would help most, after that you're into faster disks. Those > apaches are seriously overloading the box. You ought to limit the number > of apache child processes to say 5 at most, as the hardware sure ain't > doing more than that. Just total up the VM used (though some of it is I > think shared between apache processes). > > Volker > > -- > Volker Kuhlmann is possibly list0570 with the domain in > header > http://volker.dnsalias.net/ Please do not CC list postings to me.
