On Wed, Jun 17, 2015 at 10:29 AM, John Levine <[email protected]> wrote:
> >I think I remember a rule that says that delivery to a domain's > >A record SHOULD NOT happen, if an MX exists. > > > >Does that kind of rule really exists ? Or do I mis-remember ? > > You misremember. It's MUST NOT. Unfortunately, as others have > noted there is still mail software written by people who did > not read the spec. > > The relevant part of the spec is section 5.1 in RFC 5321, which > is also in RFC 2821. > > If one or more MX RRs are found for a given name, SMTP systems MUST > NOT utilize any address RRs associated with that name unless they are > located using the MX RRs; the "implicit MX" rule above applies only > if there are no MX records present. If MX records are present, but > none of them are usable, this situation MUST be reported as an error. > I can certainly imagine the scenario where an MX lookup fails and so the code precedes to the A resolution. The MX lookup failure should force the whole thing to retry, but strange results do occur. I would expect if that's the case, the fallback should be fairly rare though maybe somewhat persistent based on caching (ENODATA is often cached at SOA TTL which is often 1d or more, for example). If the code always falls back to A after exhausting MX... that's just broken. Brandon
_______________________________________________ mailop mailing list [email protected] http://chilli.nosignal.org/mailman/listinfo/mailop
