Control of Mailer Daemon
Bob Wier
which I wonder if someone can clarify.
I'm running 386BSD on a PC and have a mailing list
on it. Our domain (etsu.edu) had a nameserver which
lists the machine under one name (assigned by the net
control) and also it has a different name as an alias
(which is much more user friendly as far as people on
the mailing list are concered). My problem is that
apparantly a very few mailers look either at the path
or do a reverse name lookup when messages are bounced.
Thus, although the nameserver here has my machine
named, for example, as xx - there is also an alias
yy which is used everywhere else as far as the mailing
list is concerned. Now, if a bounce is generated back
to yy, my mailer daemon handles it just fine - BUT
some sites return it to mailer...@xx.etsu.edu.
Now, my machine thinks it's name is yy, not xx - so
it bounces the bounce since there is no such host/user
as mailer...@xx.etsu.edu (although the nameserver
tells the other machine there is).
Occasionally loops start at the rate of bounces
being exchanged 1 per minute, with about 200 bytes
of header being added each bounce. I've found that
if I kill sendmail, it stops the bounces until I start
up again. I've also tried setting the queue to non-
automatic but (here's the rub) - apparantly the bounced
messages don't get queued since there isn't any such
user/host (as far as my machine is concerned). Thus
the bounce goes back out immediately without waiting
to be queued.
Now, obviously the simplest way to fix this is to have
a consistant machine name everywhere - but the admin
here has a naming scheme they want to stick with, whereas
I have a different name which I prefer for the outside
world to use.
Alternately, if I could HOLD ALL bounce messages which come
into my machine (ie, queue 'em) that would work too.
Sorry if this isn't clear. I'd appreciate any suggestions
on this.
THANKS!
Bob Wier - mc68hc11 mailing list
======== insert usual disclaimers here ============
Bob Wier, East Texas State U., Commerce, Texas
wi...@merlin.etsu.edu (watch for address change)
Christopher Davis
BW> I've run into a problem which generates a mail loop which I wonder if
BW> someone can clarify.
BW> I'm running 386BSD on a PC and have a mailing list on it. Our domain
BW> (etsu.edu) had a nameserver which lists the machine under one name
BW> (assigned by the net control) and also it has a different name as an
BW> alias (which is much more user friendly as far as people on the
BW> mailing list are concered). [...] Thus, although the nameserver here
BW> has my machine named, for example, as xx - there is also an alias yy
BW> which is used everywhere else as far as the mailing list is concerned.
BW> Now, if a bounce is generated back to yy, my mailer daemon handles it
BW> just fine - BUT some sites return it to mailer...@xx.etsu.edu.
BW> Now, my machine thinks it's name is yy, not xx - so it bounces the
BW> bounce since there is no such host/user as mailer...@xx.etsu.edu
BW> (although the nameserver tells the other machine there is).
It would help if you gave us real names, not 'xx' and 'yy'.
The proper behavior is for your machine to recognize both names as being
valid for local delivery. How to do this depends on the version of
sendmail you're running; if it's IDA sendmail, a simple addition to the m4
file 'define(PSEUDONYMS, xx.etsu.edu)' will do.
--
* Christopher Davis * <c...@eff.org> * <c...@kei.com> * [CKD1] * MIME * RIPEM *
226 Transfer complete. 17512509 bytes received in 5.2e+02 seconds (33 Kbytes/s)
Bob Wier
> It would help if you gave us real names, not 'xx' and 'yy'.
>
> The proper behavior is for your machine to recognize both names as being
> valid for local delivery. How to do this depends on the version of
> sendmail you're running; if it's IDA sendmail, a simple addition to the m4
> file 'define(PSEUDONYMS, xx.etsu.edu)' will do.
> --
> * Christopher Davis * <c...@eff.org> * <c...@kei.com> * [CKD1] * MIME * RIPEM *
> 226 Transfer complete. 17512509 bytes received in 5.2e+02 seconds (33 Kbytes/s)
Happy to oblige - Sendmail is the Berkeley 5.67 (what's "bundled" with
the 386BSD package).
#ifndef lint
static char sccsid[] = "@(#)version.c 5.67 (Berkeley) 5/10/91";
#endif /* not lint */
char Version[] = "5.67";
and here is the entry in the nameserver:
CSCI-WierPC A 192.43.199.82
CSCI-WierPC HINFO 386 BSDUNIX
Hipp CNAME CSCI-WierPC.etsu.edu
Now, for mailing list purposes, it's a WHOLE lot better
to use hipp rather than the computer center's name...
And here is part of the bounce: (emphasis mine, naturally)
Return-Path: <wi...@merlin.etsu.EDU>
^^^^^^^^^^^^^^^^^^^^^ merlin is my smtp host
Received: from ecc.tased.edu.au (zaphod.ecc.tased.au) by
slick.clare.tased.edu.a
u (4.1/SMI-4.1)
id AA06229; Sat, 6 Mar 93 11:02:36 EST
Received: from diemen.utas.edu.au by ecc.tased.edu.au (PMDF #3100 ) id
<01GVHFTMZ...@ecc.tased.edu.au>; Sat, 6 Mar 1993 10:06:02 +1000
Received: from merlin.etsu.edu by diemen.utas.edu.au with SMTP id AA04393
(5.65b/IDA-1.4.3 for DA...@asgard.clare.tased.edu.au); Sat,
6 Mar 93 10:05:32 +1100
Received: from hipp.etsu.edu by merlin.etsu.edu with SMTP (16.6/16.2) id
AA02634; Fri, 5 Mar 93 17:05:09 -0600
Received: from merlin.etsu.edu by hipp.etsu.edu (AA01878); Fri,
5 Mar 93 19:01:31 -0600
Received: by merlin.etsu.edu (16.6/16.2) id AA02626; Fri,
5 Mar 93 17:04:39 -0600
Date: 05 Mar 1993 17:04:38 -0600 (CST)
From: "Dr. Robert R. Wier" <wi...@merlin.etsu.EDU>
Subject: test message
To: mc68...@CSCI-WierPC.etsu.EDU
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
I can't see where it came up with this unless it did a reverse
look-up or possibly from a PATH entry??? In any case, my machine knows
itself
as hipp, not CSCI, etc.
Message-Id: <930306010...@hipp.etsu.edu>
X-Envelope-To: DA...@asgard.clare.tased.edu.au
Content-Transfer-Encoding: 7BIT
Mailer: Elm [revision: 66.25]
=======
THANKS!
Christopher Davis
ckd> The proper behavior is for your machine to recognize both names as
ckd> being valid for local delivery. How to do this depends on the
ckd> version of sendmail you're running; if it's IDA sendmail, a simple
ckd> addition to the m4 file 'define(PSEUDONYMS, xx.etsu.edu)' will do.
BW> Happy to oblige - Sendmail is the Berkeley 5.67 (what's "bundled" with
BW> the 386BSD package).
OK, you'll probably need to change your sendmail.cf. Remember to rebuild
the .fc file if you use it, and also to kill and restart the daemon.
BW> CSCI-WierPC A 192.43.199.82
BW> CSCI-WierPC HINFO 386 BSDUNIX
BW> Hipp CNAME CSCI-WierPC.etsu.edu
BW> Now, for mailing list purposes, it's a WHOLE lot better
BW> to use hipp rather than the computer center's name...
Right, but a standards-conformant mailer will resolve the CNAME, so you'll
get mail for us...@CSCI-WierPC.etsu.edu.
BW> To: mc68...@CSCI-WierPC.etsu.EDU
BW> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
BW> I can't see where it came up with this unless it did a reverse look-up
BW> or possibly from a PATH entry??? In any case, my machine knows itself
BW> as hipp, not CSCI, etc.
It's from a canonicalization pass in the address rewriting.
Make sure all possible names for your machine are listed on the Cw line(s) in
sendmail.cf (I *think* that's the one; it is in my IDA-generated .cf file
anyway):
Cwhipp hipp.etsu.edu localhost localhost.etsu.edu
Cwcsci-wierpc csci-wierpc.etsu.edu
I'm sure Neil Rickert will correct me if I messed up here :)