[Mimedefang] Server crashes

[Alan Madill]

Late reply.

The system had so many files in the /var/spool/mqueue directory 
that "rm -f /var/spool/mqueue/*" would fail.  (find worked)  Sendmail 
was leaving 6 month old files there.

I was convinced it was a hardware problem.  I even set up a MRTG 
page to watch the load average etc. 
http://mrtg.hwy16.com/sysindex.html
The load average page at week 34 is where I cleaned the queue 
out.

Once the files were deleted the system runs fine.  I now run 
tmpwatch with a 7 day timeout on the queue directory.

It was a pathologically-overloaded server.

> 51000 sounds huge to me for a mail queue.  You definitely need to setup a
> slow-mail-queue and figure out why it's growing so large.  Also 1GB sounds
> pretty low for what seems to be a VERY busy server.
> 
> I documented the way I implemented DFS' recommendation's here
> http://www.pccc.com/downloads/sendmail/current-8.12.X/untarred/contrib/slowqueue/
> 
> Regards,
> KAM
> 
> > Aug 12 01:01:38 setup sendmail[31135]: grew WorkList for
> > /var/spool/mqueue to 51000
> >
> > It died here.  Is that an overly large value for the worklist?
-- 
Alan Madill - Aspen House Systems
250 567-4200