On 14 Sep 2018, at 12:59, Jim Fenton wrote:
In addition to the comment from Viktor a couple of weeks ago, I
received a suggestion off-list the other day that is worthy of
discussion. The submitter did not want to join the mailing list but
did agree to the terms of the Note Well.
The suggestion:
I had the idea, that it would be useful to see the status of the
requested REQUIRETLS service level in the "Received" trace header,
which is inserted by the MTA using simple comments. Having this, you
could see at which point in message flow the REQUIRETLS was inserted
(and lost), as well as detection of mis-behaving MTAs. And it gives
the recipient of such a message more confidence because he can see
what really happened. For this purpose, adding such information into
the trace header must be mandatory for each MTA supporting Require
TLS.
Examples:
Received: from example.com (example.com [127.0.0.1])
by relay.com (Postfix) with ESMTP
(using REQUIRETLS)
for<[email protected]>; Thu, 6 Sep 2018 11:46:34 +0200 (CEST)
or
Received: from example.com (example.com [127.0.0.1])
by relay.com (Postfix) with ESMTP
(using RequireTLS:No)
for<[email protected]>; Thu, 6 Sep 2018 11:46:34 +0200 (CEST)
These strings should be well-defined by RFC.
I am under the impression that the contents of the Received header
field are (in practice, at least) somewhat ad hoc, at least with
respect to transport characteristics like the use of TLS. So I'm not
sure how well-defined these can be, but we could at least suggest the
annotation of whether the message was received with the RequireTLS
option included. This would be primarily for diagnostic purposes,
because of course any MTA along the path could falsify that
information.
I'm less convinced of the value in the RequireTLS:no case, because
that is already expressed in a header field, and there isn't any
characteristic of the incoming SMTP session to note.
Any other thoughts on whether this would be a valuable addition?
Received has a well-defined extensible syntax as of RFC 2822. Comments
are not an extensibility mechanism and the contents of comments have no
interoperable semantic value and should never be standardized if there's
a reasonable non-comment alternative syntax.
Since Received is a trace header I think it's more important to record
what actually happened with respect to TLS rather than to record policy
that impacted what happened. There's already a standard way to record
the fact TLS was used via the "WITH clause":
https://www.iana.org/assignments/mail-parameters/mail-parameters.xhtml#mail-parameters-5
and a standard way to record the cipher suite:
https://tools.ietf.org/html/rfc8314#section-4.3
If the WG deems it important to record 'requiretls' policy, I'd prefer
that be done by adding new options to the "WITH" clause of Received for
the sake of consistency with prior work.
- Chris
_______________________________________________
Uta mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/uta
