[Mimedefang] Logging CPU Time

[cam]

> It often takes a few minutes till I get the high CPU load message by nagios
> notifications. When I login the load is back to 0.5 again and I can't see
> anything special when running "ps auxf" or "top". The mailserver scans
> ~100'000 mails a day so only the knowledge of high CPU load event start time
> doesn't help to find the mail causing this by checking sendmail's logfiles.
> 
> There must be some special kind of mail which makes MIMEDefang run high CPU
> load for a short time on my machine.


I know this response may seem like a "that's too obvious" but have you 
checked all the crontabs?  (and i mean all... not just /var/spool/cron/*)

/etc/crontab?
/etc/cron.{hourly daily weekly monthly}

based on your load of 55 that you mention, i would personally be looking 
for something that does massive amounts of disk access for short periods 
of time, like updatedb (the whereis database), find based queries, tools 
used to move failed messages from on queue to another, etc.


in my experience disk access is about teh only thing that drive the load 
up repeatedly on a box of that nature.

cam
bcp1 at craqskeep.com