>> PID USERNAME PRI NICE SIZE RES STATE TIME WCPU CPU COMMAND
>> 55651 mysql 8 12 138M 33524K nanslp 0:21 0.00% 0.00% mysqld
>> 55649 mysql 20 14 138M 33524K pause 0:21 0.00% 0.00% mysqld
>> 55866 mysql 4 14 138M 33524K sbwait 0:12 0.00% 0.00% mysqld
> Ya, since you're using linuxthreads, these are all really one process
> with one single 138MB address space; note that SIZE and RES are
> identical all the way down.
>> 76746 squid 96 0 90756K 38016K select 4:16 0.00% 0.00% squid
>> 56725 pmx4 96 0 36524K 34908K select 0:01 0.00% 0.00% perl
>> 56724 pmx4 96 0 36172K 34560K select 0:00 0.00% 0.00% perl
> Try running "ps axlm", which will show all the processes sorted by
> memory usage. If there are more of those perl scripts running, they
> may be a contributing factor. Apache with script modules (perl/php
> etc) can also suck up lots of memory if you get lots of hits at once.
You learn something new every day. It seems the culprit actually was our
anti-spam engines + apache/mod_fastcgi.
We're down to 100MB Swap used and 100MB free MEM - big difference. We'll
need to get a upgrade though, we must be able to carry the load...
Thanks for the help