[Mimedefang] Waiting for free slaves

[Andrzej Marecki]

On Thu Feb 6 16:58:01 2003 David F. Skoll wrote:

> MIMEDefang's approach is different. The slave limit is a hard limit, and it
> tempfails mail once it is reached. There is no way to wait for a slave to
> be free (without substantially rewriting the multiplexor.)

> I don't think it's a good idea to wait for a slave; if you do that and mail
> is coming in too fast, you'll have a whole pile of Sendmail processes
> waiting on the slaves, your memory consumption will increase, thrashing will
> start, and your box will grind to a halt.  If your machine is overloaded,
> it's best to say so and tempfail the mail immediately to terminate the
> waiting Sendmail process.

Just a naive question: if mail is coming in too fast and the machine runs
out of free slaves, wouldn't it be possible to keep the unscanned mail in
/var/spool/mqueue (or clientqueue). IMHO this would not create a whole pile
of Sendmail processes and grind the mailhost to a halt.

In other words: why not *accept* the mail (as if our machine was a relay)
and not force the MTA that has sent it to "try later". Let's force *our*
sendmail to try later!

I also - again naively - tried to increase confTO_CONTROL or to decrease 
confQUEUE_LA in sendmail but apparently it did not cure the immediate
tempfailing of the mail coming in too fast.

--
Andrzej

-- 
-----------------------------------------------------------------------------  
Andrzej Marecki                | 
Torun Centre for Astronomy     |   e-mail: amr at astro.uni.torun.pl
N. Copernicus University       |   WWW:    http://www.astro.uni.torun.pl
ul. Gagarina 11                |   tel: +48 56 6113032
PL-87-100 Torun, POLAND        |   fax: +48 56 6113009
-----------------------------------------------------------------------------