Anyone know why this happens?

Jeff A. Earickson jaearick at COLBY.EDU
Sat May 3 12:27:11 IST 2003

   The remote side is whacking the connection before the data transfer
completes.  This is not a DNS issue, because the DNS lookup are
already done and the connection has been made by this point.  I ran
into this with another site using Novell Groupwise email and some
firewall setup I've never heard of.  The problem was at their end.
They discovered that the MTU setting on their firewall was smaller
than their MTU on their network.  As packets went out thru the
firewall, they were fragmented.  All that packet fragmentation and
reassembly slowed the connection to a crawl.  In general, most mail
software has timeout settings (sendmail has a lot of them, see the
Bat book), and if a crawly data connection takes longer than one
of these timeout settings, then the mail software thinks the connection
is dead and whacks it.

If you are seeing the EOM message in your logs, then the problem is
at the remote end.  Either their mailer timeouts are way too small
(mailer misconfigured), their network connection is way slow (eg,
fragmented packets), or their firewall is screwing up the connection.
Since firewall boxes are the new thing, I always ask remote sites
to check the config on them first.

--- Jeff Earickson

On Fri, 2 May 2003, Mike Kercher wrote:

> Date: Fri, 2 May 2003 13:57:01 -0500
> From: Mike Kercher <mike at CAMAROSS.NET>
> Reply-To: MailScanner mailing list <MAILSCANNER at JISCMAIL.AC.UK>
> Subject: Anyone know why this happens?
> May  2 13:57:48 genesis sendmail[27561]: h42IoYo27561: collect: premature EOM:
> Error 0
> May  2 13:57:48 genesis sendmail[27561]: h42IoYo27561: collect: unexpected close
> on connection from, sender=<trudy at>: Error 0
> Is this due to a DNS timeout or something?
> Mike

More information about the MailScanner mailing list