Hi all,
Long story short here. I need to convince a business partner's admin that
we are correct in sending their users bounce messages with a null envelope
sender.
The partner is using Exchange and they are relaying a bounce message from
us to one of their internal non-M$ systems and rewriting the envelope
sender to <unknown>. M$ acknowledges the problem, it is documented. The
partner admins will refute the M$ document (it has happened before!) and
continue to assert that our MTA is to fault for this. I hate doing other
people's work.
Now, we all know that's the correct way to do it, but I went to RFC821 to
back me up. RFC821 states that this is "permissible". Fair enough - if it's
"permissible" then a compliant MTA that receives such a message should be
able to deliver or relay as appropriate. Permissible implies to me that you
may choose to use that feature, but you MUST support it when used by
another MTA delivering a message to yours. I think my argument is sound
here.
My real question is the reference in 821 (page 15) that the sender
(apparantly header sender) is server-SMTP. Now typically this is
mailer-daemon. Does this RFC really say that I should be using some form of
`cat /var/qmail/control/me`-SMTP instead? I really don't think it matters
in practice, but I want to have my ducks in a row when I take my position.
Bottom line: Should I change my bouncefrom if I want to stand by the RFC
100%?
Josh Tibbs
Kendle International
