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


To: Markus Stumpf <maex-lists-dns-ietf-dnsop@Space.Net>
cc: <dnsop@cafax.se>
From: Philip Hazel <ph10@cus.cam.ac.uk>
Date: Thu, 28 Feb 2002 09:09:37 +0000 (GMT)
In-Reply-To: <20020227143858.R9474@Space.Net>
Sender: owner-dnsop@cafax.se
Subject: Re: Minneapolis - agenda items please.

On Wed, 27 Feb 2002, Markus Stumpf wrote:

> I have checked with some people who are very familiar with sendmail and
> exim. According to them neither of those MTAs has any provisions to
> handle the situation effectively. They go best MX, time out and the
> backup MX. I know qmail does handle it more effectively, as it maintains
> an IP timeout table.

I am very familiar with Exim (I wrote it :-). It *does* have provisions
to handle these problems. Of course it tries the best MX first, times
out, and then goes to backup. But it remembers that the best MX timed
out, and it won't try it again until its retry time arrives. As time
passes, the retry time increases (configurable, but normally gets up to
about 6/8 hours max). Thus, after a while, most messages will go
straight to the backup, with a new attempt at the best MX every now and
again. (I guess this is the same as "an IP timeout table".)

Exim also has a provision to handle the MX->localhost->127.0.0.1
problem. You can configure it to ignore certain IP addresses that come
up in a DNS lookup. Typically, 127/8 are ignored. Exim behaves as if the
A records don't exist.


-- 
Philip Hazel            University of Cambridge Computing Service,
ph10@cus.cam.ac.uk      Cambridge, England. Phone: +44 1223 334714.


Home | Date list | Subject list