Skip to main content

Minutes IETF101: sipcore
minutes-101-sipcore-00

Meeting Minutes Session Initiation Protocol Core (sipcore) WG
Date and time 2018-03-22 09:30
Title Minutes IETF101: sipcore
State Active
Other versions plain text
Last updated 2018-04-13

minutes-101-sipcore-00
sipcore 101 notes

2018 March 22, Thursday, 0930-1000
Viscount, Hilton London Metropole

Chairs: Brian Rosen and Jean Mahoney, with Russ Housley helping out since Jean
could not attend.

-----------------------------------------------------------------------
5 m Administrivia
Presenter: Brian Rosen
Slides:
https://datatracker.ietf.org/meeting/101/materials/slides-101-sipcore-chair-slides-00

Jabber scribe: Matt Miller

Notes: Jean Mahoney (from the Meetecho recording)

Brian Rosen: Most drafts are in LC and LC is finished. We are here to discuss
session timers.

Rifaat Shekh-Yusef: Please review draft-ietf-sipcore-sip-authn and provide
feedback.

Brian: We would like to clear the current SIPCORE drafts. We need reviews for
these drafts. Please send comments to the ML or say the draft is ready to go if
you have no comments.

Ben Campbell: A WGLC without comments is not a successful WGLC.

-----------------------------------------------------------------------
30 m draft-ietf-sipcore-session-timer-race-00
Presenter: Christer Holmberg
Slides:
https://datatracker.ietf.org/meeting/101/materials/slides-101-sipcore-session-timer-race-condition-01

Paul Kyzivat: When an update is sent by the application server, it's relative
to the request. UAC/UAS is a transaction thing - it doesn't change the problem.

Christer: Then sending a 480 is wrong.

Paul: Correct, but if the ---

Jonathan Rosenberg: I remember this. It matches what Paul is saying - it's
transactional.

Christer: I'm going to move ahead to the Blame slide (slide 12). The RFC
doesn't cover mid-dialog requests. Slide 14 (The Solution (Draft)) - we need to
change the draft text. Focus on ongoing transaction to avoid glares.

Brian: Is everyone ok with the approach? Is the problem well stated?

Roman Shpount: The problem is well stated, but not happy with the solution.
Just insist on consistent behavior.

Christer: The aim is to minimize impacts on the UAs. It's not feasible to
update them all.

Brian: UA to UA - they don't do session timers.

Roland Jesske: The focus is well defined, going in the right direction. I
support. Document the proxy behavior well.

Paul: Claiming a distinction between AS and UAS is a red herring. There is a
difference between the UA and proxy.

Brian: What do you think of the draft?

Paul: Having the update not carry SE can work, but doesn't feel right. Require
consistency. Need to explore implications.

Christer: Take it to the list.

Brian: Just want to agree on the problem and understand the solution space. Any
other comments? Want to push this along.

Roman: AS and UAS are the same. Will have some UA impact any way you go.

Christer: Agree. Just have a mindset to minimize impact.

Roman: With proxies - very limited things can be done there. Won't be backwards
compatible.

Brian: Have discussion on list. Pick a direction. Get the doc done.

Christer: We can have informal calls, or an interim meeting is fine.

Brian: Informal calls are not a working group thing. They would be a working
group proposal.