[Mimedefang] config for big servers-- problem found

[Charles Mount]

I found the cause of the problem below.
I use the milter version of TrendMicro antivirus followed by MimeDefang
with spamassassin.   It is not possible to configure TrendMicro to NOT send
a warning message to the recipent targeted by a virus.  I use
MimeDefang/spamassassin to block those messages.   Mimedefang is set to
quarantine rather than delete messages.   With the recent huge peak in
virus activity I got a huge number of quarantined messages.
I had been using --
find /var/spool/MD-Quarantine -depth -mtime +5 -exec rm -r {} \;
running daily from a crontab to purge old messages from the quarantine.
With the huge number of quarantined messages that purge script failed.
This led to MIMEDefang not being able to write to the quarantine directory.
Messages that should have been quarantined would sit five minutes and give
Error from multiplexor: ERR Filter timed out - system may be overloaded
(consider increasing busy timeout)
I did not have a disk full problem, it was a number of directorys problem.
I have the quarantine directory cleaned out and we are past this virus
storm.   I think I can keep this from happening again.   However, if
someone else has solved a similar problem, please let me know.


===============
I have four SUN 880s with eight processors and 16 GB.   These are running
Solaris 8 with Sendmail Switch 3 and Mimedefang 2.51.
One of the four seems to have a slightly higher load and gives these
errors.  I have not seen load average above 3.5.


Error from multiplexor: ERR Filter timed out - system may be overloaded
(consider increasing busy timeout)
Error from multiplexor: error: No free slaves
Milter (mimedefang): timeout before data read
Milter (mimedefang): write(Q) returned -1, expected 5: Broken pipe
Milter (mimedefang): to error state
Milter (mimedefang): error connecting to filter: Connection refused by
/var/spool/MIMEDefang/mimedefang.sock

In the startup script I have

ulimit -n 1024

MAX_SLAVES=50
MIN_SLAVES=25
BUSY_TIMEOUT=300
IDLE_TIMEOUT=300