[ On Monday, February 8, 1999 at 11:11:34 (+0000), Paul Mansfield wrote: ]
> Subject: Re: [EXIM] SMTP timeouts?
>
> I saw this problem once with a network that was dropping all packets over a
> certain size (due to a Cisco bug^H^H^Hfeature), which mean that the first part
> of the SMTP transaction worked, but when the body of the mail started flowing,
> the main data bits never got through the network. This can also happen if a
> machine has a broken TCP stack and won't reassemble fragmented datagrams.
There's also the possiblity someone is filtering ICMP which can
adversely affect attempts by remote hosts to use Path-MTU-discovery.
I've seen this a *lot*, far too often in fact, because I have an MRU of
1006 on my PPP link. Usually such filtering is "caused" by
inexperienced firewall administrators being too draconian because they
really don't understand what ICMP is and how TCP/IP really works.
Sometimes though it's because a site uses a "layer 3/4 re-director" for
load balancing or fault tolerance and not all such devices "Do The Right
Thing" with ICMP either (i.e. their designers similarly over-looked the
importance of ICMP for correct functioning of TCP). Sometimes PMTUD can
be disabled on the remote host to temporarily "fix" the problem (for
example on NetBSD it is "sysctl -w net.inet.ip.mtudisc=0").
--
Greg A. Woods
+1 416 218-0098 VE3TCP <gwoods@???> <robohack!woods>
Planix, Inc. <woods@???>; Secrets of the Weird <woods@???>
--
*** Exim information can be found at
http://www.exim.org/ ***
