[Mimedefang] Mimedefang crashing randomly

[M Jerome Garrett]

I am having a strange problem w/a Mimedefang install.  Mimedefang, ClamAV,
and Spamassassin are all on the system.  This is my 5th server to do this
identical set up on but this one is the only one having problems.  

The milter will work for a while and then decide that it doesn't want to
play anymore and mimdefang will stop.

Here's the maillog.

Feb 21 21:08:14 gateway sendmail[5864]: l1M38DVs005864: from=<>, size=28174,
class=0, nrcpts=1, msgid=<courier.45DB868F.000065FD at mail1.permonline.ru>,
bodytype=8BITMIME, proto=ESMTP, daemon=Daemon0,
relay=mail.peckandassociates.com [70.250.240.76] (may be forged)
Feb 21 21:08:14 gateway sendmail[5864]: l1M38DVs005864: Milter add: header:
X-Virus-Scanned: ClamAV 0.88.7/2624/Wed Feb 21 18:04:38 2007 on
gateway.peckandassociates.com
Feb 21 21:08:14 gateway sendmail[5864]: l1M38DVs005864: Milter add: header:
X-Virus-Status: Clean
Feb 21 21:08:14 gateway sendmail[5864]: l1M38DVs005864:
milter_read(mimedefang): cmd read returned 0, expecting 5
Feb 21 21:08:14 gateway sendmail[5864]: l1M38DVs005864: Milter (mimedefang):
to error state
Feb 21 21:08:14 gateway sendmail[5864]: l1M38DVs005864: Milter: data,
reject=451 4.3.2 Please try again later
Feb 21 21:08:14 gateway sendmail[5864]: l1M38DVs005864:
to=<oxy at peckandassociates.com>, delay=00:00:01, pri=58174, stat=Please try
again later
Feb 21 21:11:14 gateway sendmail[5881]: l1M3BEvJ005881: Milter (mimedefang):
error connecting to filter: Connection refused by
/var/spool/MIMEDefang/mimedefang.sock
Feb 21 21:11:14 gateway sendmail[5881]: l1M3BEvJ005881: Milter (mimedefang):
to error state
Feb 21 21:11:14 gateway sendmail[5881]: l1M3BEvJ005881: Milter:
initialization failed, temp failing commands

I can restart the mimdedefang service and it will start running again until
this unknown event happens.  In the mean time the filter works just like it
is supposed to.

Any ideas on what "milter_read(mimedefang): cmd read returned 0, expecting
5" means?

The server has a gig of memory and I've tried lowering the number of slaves
to 2 running and 1 stopped for a total of 3 and that still didn't help
anything.  

M. Jerome Garrett
Technology Solutions, Information System Consultant
M.S. in Telecommunications Management
M.S. in Management Information Systems