On 8/13/25 19:42, Al Iverson wrote:
Indeed, Microsoft is requiring SPF to pass. It doesn't have to align,
but it does have to pass. Which means a scenario you would expect to
result in delivery -- DKIM pass, DMARC pass (because of DKIM pass and
alignment), SPF fail (because of forwarding) -- will now be rejected.
So, yeah, I think you're reading the requirements right. If you're a
bulk sender, SPF, DKIM and DMARC all have to pass. Doesn't mean that
both SPF and DKIM have to align. They encourage/recommend it, but do
not mandate that both must align. It does seem to put a crimp in the
ability to forward mail to destinations at MS's consumer mailbox
domains.
IMHO, this is fully unrelated to the "low TTLs"/DNS timeout issues
others are mentioning here. Also a problem, but a different problem.
I agree, this is not about alignment. But after spending way too much
time looking through logs I find this is not as consistent as I
initially thought it was. For the same sender domain I can sometimes
forward and sometimes not and when it fails it's always with SPF=fail
(which should _always_ fail since we're forwarding), DKIM=pass,
DMARC=pass. So it appears MS requires all checks to pass for large
senders _sometimes_ but not other times. Or the reason for these rejects
are not what the error says they are.
The behaviour also varies by sender domain. Most notably forwarding
mails from facebookmail.com (definitely a large sender) never fails to
deliver. For other sender domains it varies between 25-50% failures
(mail.instagram.com has a ~50% failure rate so no special rules for
facebook there...). If I treat the error as temporary and just retry at
our normal retry intervals I seem to get a ~10% success rate to
eventually deliver the messages (though I haven't run that test for
long enough to get any definite final statistics on it). That still
means that ~20-45% of all forwards to Microsoft end up bouncing in the
end though, which is pretty terrible. (Also my mail queue is getting
huger by the second)
--
BR/Mvh. Dan Malm, Systems Engineer, one.com
_______________________________________________
mailop mailing list
[email protected]
https://list.mailop.org/listinfo/mailop
