[Wylug-help] Exim/Mailman unrouteable mail domain "xxxx.xxxx.xxx"

[Dave Fisher]

Hi all,

I run a very small mailing list from home for a group of friends, but
unfortunately a proportion of the messages are being consistently
bounced with the error message: unrouteable mail domain "xxxx.xxxx.xxx"

Googling turns up quite a bit of relevant, interesting but ultimately
unhelpful stuff, e.g.

  http://www.exim.org/pipermail/exim-users/Week-of-Mon-20030407/052077.html

The problem is obviously triggered by dns lookup failures, but the
symptoms are just too severe (failures on up to one third of recipient
addresses) and too anomalous to ignore.

It is entirely possible that the recipients' ISPs (inc BT and Freeserve)
have sporadic and temporary problems with their nameservers getting out
of synch.  Unfortunately, that doesn't really explain several phenomena:

1) Why people at the failing addresses swear that they never have mail
delivery problems from any other source.

2) Why individual mail messages from me to the affected recipients
always get through whereas a very high proprtion of list mail fails.

3) Why manual dnsdb lookups with dig always succeed (e.g. dig
xxx.xxx.xxx mx, dig xxx.xxx.xxx ns, etc.) but other tools which use
namelookups (e.g. ping, traceroute, nslookup, etc) produce 'unknown
host' or 'No answer'.

N.B.  I don't think my nameserver config is at fault, since the
behaviour doesn't change no matter who's name server I use, although
there may be some other part of the lookup process that I've configured
wrongly.

Anyone got any ideas?  Could Exim be doing something differently with
list messages rather than ordinary mail?

I notice, for example, that an unusually high proportion (almost all) of
the reports I've found on Google relate to the same software combination
I run, i.e.  Debian/Exim/Mailman/Mutt.  I realise that this could be a
red herring or merely an artefact of my search strategies, but the
pattern is striking.

Dave