At 00:33 27-03-2008, Sabahattin Gucukoglu wrote:
To address the problem that begs for the rule's removal, however: how
about a fake MX that rejects everything as the draft-rfc2821-bis allows
with the 554 code on connection opening? How about not minding that the
If I'm not mistaken, example.com did that before giving up on the
idea. Even if a device can be assigned an AAAA RR, it doesn't mean
that it can run an SMTP service.
Most people won't bother to run a minimal SMTP service as it requires
resources and can have a negative impact on other running
services. The way they address the problem is by adding a MX RR
pointing to an invalid host.
sending host is the one with the problem, and not you (not nice, so it'd
be good to sort it out)? How about using a known-nonexistent (EG:
definitely.invalid) host as your MX target (has more bad effects than
known behaviour but at least it works - and when the rules don't allow us
to do better than that for now, that's all we can hope for)? This "MX 0
." sounds good as well, because it only requires time for implementers to
get the idea and implement it while it's a bit of bother in the meantime.
That's what some people do. MX RRs would have to be defined for all
the hosts with AAAA RRs then.
Regards,
-sm