[Mimedefang] Timeout before data read

[Joseph Brennan]

--On Wednesday, August 23, 2006 17:14 -0500 Kayne Kruse 
<kkruse at pointone.com> wrote:


> One thing to also look at is what your load average is.  Mine would
> start timing out on my slower box when it had ~10 concurrent mimedefang
> processes going.  Adjusting the Milter timeout like David says and a
> combination of load average throttling via sendmail.mc/cf can keep those
> more controlled.
>
> So for me 10 mimedefang procs usually slammed the box to about 12 load
> ave.  Throttle sendmail to 10 load average and let it try to finish
> running.  Usually after 2-4 mimedefang slaves free up, things are back
> to 0.
>
> KK


We're much larger-scale than that.  However it is good to know that
the problem has to do with sendmail waiting for the milter.  Nothing
we do should take that long per message, or so I thought.  That tells
me what to look into.  Thanks, and thanks to David too.

Joseph Brennan
Columbia University Information Technology