[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Simple] MSRP-ACM Discussion Summary
(As individual)
The discussion thread on MSRP-ACM seems to be dying out, and I don't
think we've reached closure on most of it. In order to jump start it,
I am going to attempt to summarize the issues that I think are still
open. I will start separate threads for each issue. Please reply to
_this_ message to comment on whether I have offered a fair summary, or
mischaracterized or omitted anything material. Please reply to the
specific issue threads to discuss the issues themselves.
1) We have consensus that the transport address connection parts
(section 4.2 of draft-ietf-simple-msrp-acm-00) must allow backwards
compatibility with a peer that uses an RFC4976 MSRP relay. We have had
one proposal on how to accomplish this, which involved updating 4975
to allow a middlebox to modify the authority part of an MSRP(S) URI in
the SDP "a=path" property.
We further determined that this proposal may interfere with the use of
MSRPS URIs (i.e. TLS) in the case where said middlebox is transparent
to TLS. It modified an MSRPS URI to point to itself, but transparently
relays the TLS handshake to a relay (or other MSRP device that uses a
TLS server cert.). The modified URI no longer matches the server cert,
so things break.
We've had some discussion about using the comedia-TLS fingerprint
mechanism for certificate matching rather than SubjectAltName matching
to get around this issue. This discussion is far from "baked". We've
also had some objection (primarily from Jon) to doing work around a
new class of MSRP TLS intermediary without better describing that
intermediary, as well as concerns (primarily from Ben ) about
standardizing MSRP mechanisms to work with non-standardized
middleboxes without a better understanding of them.
2) We have little or no objection to the usage of the comedia "setup"
attribute (section 4.1 of draft-ietf-simple-msrp-acm-00. )
Is this a fair summary? Have I mischaracterized or missed anything?