MailScanner cron job?

[Eric Sandquist]

On my system, I started using MailScanner yesterday...  Seemed to work ok
for a while... Server load w/o is .60 - .80...  After starting it,
activating the cron and restarting sendmail for queue.in and queue... load
jumped to 1.2-1.7... still acceptable... Only scanning for virii...
SpamAssassin is running through procmail for individual users with
spamc/spamd since system wide scanning on this machine nearly killed it in
the past...

I have 5 child-processes set for Mail Scanner...  In about an hour server
load had exceeded 17.0-22.0... ouch.. not acceptable... was barely able to
get back in and kill things off, and that was only after a reboot...

Took another 10-20 minutes to settle back down to normal...

Would reducing the child-processes to 1 stop this from happening?

Is there any way to set this up for specific users or to exclude specific
users/accounts/aliases???

I run a few mail-list discussion groups here and the list management
software scans for virii, so they don't really need it... just the normal
users...

Thanks..

Eric
Systems Engineer

-----Original Message-----
From: MailScanner mailing list [mailto:MAILSCANNER at JISCMAIL.AC.UK]On
Behalf Of Chris W. Parker
Sent: Wednesday, June 04, 2003 1:12 PM
To: MAILSCANNER at JISCMAIL.AC.UK
Subject: Re: MailScanner cron job?


Tony Finch <mailto:dot at DOTAT.AT> wrote:

> You probably have a Max Children setting that's too high. Unlike
> Apache (whose child worker processes don't do anything when the
> machine is idle, and will happily page out), MailScanner is
> continuously active scanning the incoming queue for new messages.
> Also unlike Apache, MailScanner's child processes are big and don't
> share much of their memory -- on my setup each child uses 20MB. I
> would run with Max Children = 2 on your machine.

In fact I've moved it down to 1 and everything has quieted down now
(actually it quieted down yesterday around 3pm). I think maybe that was the
problem as it was the only significant change I made.


Chris.