workload on daedalus

[Greg Ames <gr...@remulak.net>]

I noticed that apache.org httpd response time was sluggish.  We were running at
MaxClients 600 consistently, so I bumped MaxClients to 800 and did a graceful
restart.  That seems to help.  Looking at server-status, there definately are a
lot of downloads for 2.0.35 happening.  Assuming that's the cause and that it
will be temporary, I changed MaxClients back to 700 in the config file to be
conservative.

top shows that CPU time is getting eaten more than I'm used to seeing.  The load
averages are somewhat higher than normal too.  The big indiviual users look like
named, qmail, and something called multilog:

  PID USERNAME     PRI NICE  SIZE    RES STATE  C   TIME   WCPU    CPU COMMAND
  386 qmailr        62   0  1320K   676K RUN    0 727:01 10.25% 10.25% qmail-rs
  115 bind           2   0 20668K 18360K select 1 742:46  9.08%  9.08% named
  382 qmails         2   0  1164K   468K select 1 728:36  5.08%  5.08% qmail-se
  383 qmaill        -6   0   892K   284K piperd 1 303:12  3.32%  3.32% multilog

I don't know an easy way to see the total CPU time that all the httpd's are
using.  Any ideas would be appreciated.

Greg