This is a discussion on Re: SMTP Reject Correctness - VMS ; This is from John Covert: Michael had asked: >So, the question is this: Is there some SMTP netiquette/RFC that I've >violated by rejecting sites that don't honor the MX record and am I >therefore losing legitimate email? and Jeremy had ...
This is from John Covert:
Michael had asked:
>So, the question is this: Is there some SMTP netiquette/RFC that I've
>violated by rejecting sites that don't honor the MX record and am I
>therefore losing legitimate email?
and Jeremy had correctly replied:
>I don't believe so.
The RFC is clear that if there are any MX records, the A record should
not be used for mail delivery. The A record is only to be used for mail
if there are no MX records at all.
However, TCPIP for VMS, at least through V5.5, incorrectly will first
try the MX records and if none of them are currently reachable, go on
to the A record. This is definitely a violation of the RFC, and results
in occasional random failures for various reasons, such as Michael's
configuration (if his appliance pointed to by the MX records were not
reachable) or in the case where email is handled by a different system
than www traffic.
There is a correction in the works for a future version, and I have been
trying for several years to convince the developer and project leader
that the new default should correct this bug, but at the moment the
new default is the old, incorrect behaviour, and a system manager
would have to edit SMTP.CONFIG to fix it. I'm still trying to convince
the folks, but I'm now an independent consultant, no longer in the VMS
group, and have limited influence. This list is not the place for any further
discussion of TCPIP for VMS; I only mention this in reply to Michael's
question about whether legitimate senders could be running afoul of