[Mimedefang] Timeout before data read

David F. Skoll dfs at roaringpenguin.com
Tue Dec 4 21:03:10 EST 2007


Kevin A. McGrail wrote:
> md-mx-ctrl busyslaves
> 0 B 1381 (last=scan) (scan /var/spool/MIMEDefang/mdefang-lB4Ljqqq006025)

> [root at intel1 IMAP]# grep lB4Ljqqq006025 /var/log/maillog
> Dec  4 16:49:54 intel1 sendmail[6025]: lB4Ljqqq006025: Milter
> (mimedefang): timeout before data read
> Dec  4 16:49:54 intel1 sendmail[6025]: lB4Ljqqq006025: Milter
> (mimedefang): to error state
> Dec  4 16:49:54 intel1 sendmail[6025]: lB4Ljqqq006025: Milter: data,
> reject=451 4.3.2 Please try again later
> Dec  4 16:49:54 intel1 sendmail[6025]: lB4Ljqqq006025:
> to=<treasurer at asanv.org>, delay=00:04:00, pri=130583, stat=Please try
> again later

> What's curious is that the mimedefang process was still hanging around
> and I run with some pretty decent timeouts:

Here's what's happening: Sendmail is timing out the call into milter
after 4 minutes (the Read timeout: R:4m).  However, the Perl scanner
is blissfully unaware of this and keeps working.

To avoid (or mitigate) this, you should set your MX_BUSY timeout (the
argument to the "-b" option of mimedefang-multiplexor) to only
slightly more than the R: or E: timeout in the INPUT_MAIL_FILTER line.
That way, both Sendmail and MIMEDefang more-or-less agree on when it's
appropriate to give up.

Now, as to *why* it's taking 4 minutes to scan the message, that's a whole
other mystery to explore.

Regards,

David.




More information about the MIMEDefang mailing list