Hector Santos wrote:
In regards to RFC2821, section 5, saying:
There seems to be a conflict between this and the following itemized rule in
5.3.4 Reliable Mail Transmission
(1) Multiple MX Records -- these contain a preference
indication that should be used in sorting. If there are
multiple destinations with the same preference and there
is no clear reason to favor one (e.g., by address
preference), then the sender-SMTP SHOULD pick one at
random to spread the load across multiple mail exchanges
for a specific organization; note that this is a
refinement of the procedure in [DNS:3].
RFC2821 seems to suggest to include all equal preference MX records, while
RFC1123 seems to suggest, unless I am reading it wrong, to choose 1 (at
random) of the equal preference MX records.
I think RFC 1123 is ambiguous here. You are right that it says to
choose one. But note that it does not say what to do next -- if the
first one picked fails -- and that is the question. Your
interpretation can follow from what it says.
One possible interpretation (which I believe is yours) is:
1. to skip over all the remaining destinations with the same
preference (that is to pick one -- and only one -- destination with
But other interpretations can also follow, such as:
2. to pick another one at random from the remaining destinations with
the same preference as the first one that failed (that is to pick
first one at random, and then another one at random if the first fails).