Hi,
Yesterday I implemented an ACL snippet that blocked all mail that did
not have a Message-ID: header present. Today, I got a lot of complaints
of customers who's mail bounced. Who would have guessed... ;-)
So I started reading up on the subject (should have done that before,
but time is precious), but I couldn't get everything 100% clear.
One example of blocked mail came from a large internet provider in The
Netherlands (@Home) that, according to the Received headers, uses Exim
4.30. This particular mail came from one of their clients and did not
have a Message-ID header.
I have always thoughtlessly assumed that Exim would add a Message-ID
header if missing, but this case led me to "control = submission". I
immediately jumped to the conclusion that I needed something like:
warn hosts = +relay_from_hosts
control = submission
on all outgoing mail relays that are under my control.
However, a quick'n'dirty test with our Exim 4.50 here at the office
learns me that a Message-ID header /is/ added when I relay a mail
through it, even though there is no "control = submission" to be found
anywhere in the config.
I read that Exim's behaviour has changed over time in this respect. How
should I look at this issue? What are recommended settings?
And a little off-topic:
RFC2822 states a Message-ID header SHOULD be present in all mails,
meaning it is not absolutely mandatory. How does the Exim community look
at this? Should I try and persuade the mentioned provider to fix up
messages from their clients?
Thanks,
Martijn Grendelman.
