[Mimedefang] Problems with SpamAssassin 3.1 RC1and MIMEDefang

[Nels Lindquist]

On 22 Aug 2005 at 16:57, Nels Lindquist wrote:

> On 22 Aug 2005 at 16:46, Nels Lindquist wrote:
> 
> > Everything's great as far as detection goes, but I've run into an 
> > issue with MIMEDefang and I'm wondering if any of you have 
> > experienced anything similar.  It looks like the multiplexor is 
> > unable to terminate slaves properly, and in short order there are 
> > MAX_SLAVES zombie processes, "no free slaves" and all mail is 
> > tempfailed.
> 
> Sorry to reply to my own message, but this may be related to a 
> reported SpamAssassin bug:  
> 
> http://bugzilla.spamassassin.org/show_bug.cgi?id=4518
> 
> I'm going to try SA 3.1RC1 again, but disable DCC.

I figured I'd be better off applying the patch attached to the above 
Bugzilla entry, but it didn't help.  In retrospect, since I'm using 
dccifd with a socket it's probably not the same problem as was 
triggered with dccproc.

The problem seems to be triggered when the load rises enough to spawn 
more than MX_MINIMUM slaves.  Once those excess slaves have been idle 
for a while and the multiplexor attempts to shut them down, they 
become zombies.

I assume that the same thing would happen over a longer period of 
time when slaves reach MX_REQUESTS.

I've tried disabling embedded perl (although there are no 
difficulties with it with SA 3.04) and I'll probably upgrade to the 
latest version of MIMEDefang as well.

Any other suggestions?

----
Nels Lindquist <*>
Information Systems Manager
Morningstar Air Express Inc.