spamass-milt-list
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: spamass-milter stops forking new childs?


From: Ron Snyder
Subject: RE: spamass-milter stops forking new childs?
Date: Tue, 28 Jan 2003 08:10:07 -0800

> Yes, I used your restart-script (with kill -KILL cause the 
> process which was
> haning usually doesn't close with kill -TERM) in november 
> 2002 on a other
> server with ~120'000 mails/day, but it restarted spamassasin 
> up to 3 times a

I was experiencing similar "yuckage" (with only 5000 messages/day) on
openBSD 2.7, and put the following in my /etc/rc.local:

( while true ; do logger -p mail.err starting spamassassin milter;
/usr/local/sbin/start-spamass-milter ; rm /var/run/spamass.sock ; done) &

where file/usr/local/sibn/start-spamass-milter is this:
#!/bin/sh
/usr/local/sbin/spamass-milter -d 2 -p unix:/var/run/spamass.sock \
          2>> /var/log/spamass-errors; echo `date` >> \
          /var/log/spamass-errors ; sleep 1

(wrapped for readability-- it's actually all on one line)

This worked for me because the process was dumping core most of the time.
Occasionally it would hang and just not read from the pipe, and I'd have to
manually kill it (this only happened about once every 3 weeks).

(I do have to say that to narrow down the problem I moved my mail filtering
to a RH Linux 8.0 and all of my problems with the milter seem to have gone
away.)

> Anyone ever tried to create a redudant spamass-milter? I 
> could setup 3 IP
> aliases on the spamassassin-milter server and create a host 
> with 3 IN A
> records. Is this a stupid Idea? :-)

I think you'll see only limited benefit.  We've got three MX records for our
domain, and would frequently see evidence of MTAs that would attempt the
delivery once, our milter would crash so the sending MTA would get a 4xx
error, and the sending MTA would just requeue the message to try again later
rather than attempt delivery to one of the other MX machines.  I think you'd
see similar behavior with your proposed configuration.  (BTW, I didn't
bother to read the RFCs to determine if there was a "correct" behavior wrt
to immediately attempting delivery to an alternate MX upon receiving a 4xx.)

Of those senders that were most diligent about attempting delivery to a
different MX, the mail seemed to be mostly spam. :)

-ron




reply via email to

[Prev in Thread] Current Thread [Next in Thread]