On 2 Feb 2021, at 8:46, Tim Coote wrote:
Before I pollute the mailing list with all the gory details. Is this
a known/expected/ever seen behaviour?
SRS rewriting should not operate on ANY header addresses. It is
designed, like SPF, to operate on the SMTP envelope sender.
It is normal for automated messages such as bounces to have a null
envelope sender, so it is possible to pass a null address to
postsrsd.
My abuse of email jargon and misunderstanding. Sorry. I must be wrong
about Message-id, then. I’ll check.
So, if there’s no sender in the envelope, the email cannot be
automatically forwarded to some mail providers (notably gmail in this
case), and the user of the forward address is unaware of such
response?
I may be misunderstanding your question, since I am certain that GMail
accepts and delivers legitimate bounce messages with null envelope
senders.
In the case of this source, it’s an automated reply, so the end
email address is unaware of the successful or failed completion of the
message exchange.
Is that how forwarding’s supposed to work?
I think a solid example of the problem with all relevant log entries
could help.
Traditionally, forwarding via .forward files or aliases is done using
the same envelope sender as the forwarded message originally used on
arrival. SRS only exists to modify envelope senders so that forwarding
MTAs do not cause forwarded mail to violate SPF, which authenticates
sender address domains based on the SMTP client's IP address. Because
the forwarding MTA is acting as a SMTP client when passing along the
message, SRS is used to replace the original sender with one in the
forwarder's domain, but there's no need to do that in the special case
of the null envelope sender, which has no domain part to check with SPF.
--
Bill Cole
b...@scconsult.com or billc...@apache.org
(AKA @grumpybozo and many *@billmail.scconsult.com addresses)
Not Currently Available For Hire
