[Top] [All Lists]

Re: current usage of AAAA implicit MX?

2008-04-04 13:44:17

--On Friday, 04 April, 2008 13:06 -0400 Tony Hansen
<tony(_at_)att(_dot_)com> wrote:

I got the following message and received permission to repost
it here.

      Tony Hansen

From: "Steinar H. Gunderson" <sesse(_at_)google(_dot_)com>


Harald sent this on to me, as I've set up and help administer
the a real-world e-mail setup with IPv6 support. I'm not sure
how much I can help you with the other points, but I can tell
you that there is at least one site accepting e-mail with an
AAAA record only -- my home machine. :-)

Apart from that, I don't think you can remove delivering
e-mail directly to AAAA records; it's already supported in
MTAs out there, and you can be pretty sure it's actually being
used. Besides, if you sent mail to foo(_at_)example(_dot_)org, and had both A and AAAA records (but no MX), should
you really not try delivering on IPv6?


While this is useful, it also illustrates that we still haven't
managed to make clear what problem we are talking about.  If you
sent mail to foo(_at_)example(_dot_)org, and "both A and AAAA records (but
no MX)" are present, then

(i) Under the scenario in which an implicit MX is generated only
for A RRs,  IN MX 0
would be implied.

(ii) Under the scenario in which an implicit MX is generated for
either A or AAAA RRs,  IN MX 0
would be implied -- exactly the same.

Once  IN MX 0
exists, whether explicit or derived from the presence of the A
RR, it would be, IMO, a serious misunderstanding of the protocol
to claim that delivery to the AAAA interface is prohibited.  

I don't think that is at issue here.  I think the _only_ issue
is whether or not, if a given domain name has no MX record(s)
_and_ no A record(s), but has at least one AAAA record, whether
that implies that a mail client should synthesize the implied MX
and then consider it as a mail-delivery destination.

The confusion about what is actually the question is beginning
to concern me more than the question itself.  I know this isn't
your fault, but we need to figure out how to explain this better.


/* Steinar */
-- Software Engineer, Google Norway

<Prev in Thread] Current Thread [Next in Thread>