Mailscanner and memory resources

Rose, Bobby brose at MED.WAYNE.EDU
Fri Jan 23 17:28:33 GMT 2004

It's the larger size.  Res runs about 50 to 107megs.  I had to reboot
this morning because MailScanner ran out of fork space and I developed a
queue backup.  But people are starting develop their own rules and this
is going to be more prevalent so SA can get larger.  I even stopped
using bayes because it was being too slow and I still use RBLs on the
sendmail side so SA and MS don't mess with those.  So the only SA stuff
running is the local stuff. 

-----Original Message-----
From: MailScanner mailing list [mailto:MAILSCANNER at JISCMAIL.AC.UK] On
Behalf Of Julian Field
Sent: Friday, January 23, 2004 11:53 AM
Subject: Re: Mailscanner and memory resources

At 16:38 23/01/2004, you wrote:
>  I've started using some of the SA ruleset like tripwire, bigevil and 
>such.  In doing so, I'm finding that the size of the MailScanner 
>processes increase dramatically because each MailScanner process is 
>essentially it's own SA process and loads all the configs and rules it 
>needs.  Wouldn't it be better to just have MailScanner make calls to 
>spamd thus reducing the amount of resources?

In my view, no. You are running far more extra rulesets than most people
do. Is the 119Mb per process the resident size (RSS in "top"), or just
(larger) "size" figure quoted in "top"? It's only the RSS figure that

Calling spamd would be slower and, more importantly, would rely on you
having a daemon running, which introduces a whole host of reliability
and recovery problems. I would have to write a whole system to handle
memory leaks (and other resource leaks) in spamd, and be able to detect
when it stops working properly and restart it. That's a very messy
process, and is the same reason I don't support the daemonised versions
of the virus scanning engines.

>   On my system with 2gigs of
>Ram and running 10 MailScanner processes, each process is 119megs with 

Julian Field
MailScanner thanks transtec Computers for their support

PGP footprint: EE81 D763 3DB0 0BFD E1DC 7222 11F6 5947 1415 B654

More information about the MailScanner mailing list