Minutes IETF124: snac: Tue 14:30
minutes-124-snac-202511041430-02

SNAC IETF 124
Tuesday 04 November 2025, 09:30-11:00, Montreal
Where: Agora, https://meetings.conf.meetecho.com/ietf124/?session=34801

Chairs: Darren Dukes & David Lou
Note takers: Tim Winters, Bob Hinden, Esko Dijk

Agenda
- Administrative & Agenda Bashing (5min, Chairs)
- WG Status
- Multi-AIL （30min, Esko Dijk）
- Any other Business (AoB)

Administrative & Agenda Bashing (5min, Chairs)
Ted Lemon is asking about having time to present on the state machine
work he has been working on towards the end. This is added to the agenda.

WG Status Update
Closed 20+ issues in 5 interim meetings in September/October. 30 Items are
still under review. Reviewed updated Milestones, no-one objected to sending
those to the AD. Eric Vyncke (AD) suggested the milestones could be reviewed as
early December. snac-router-flag-03 is waiting in RFC editor queue (MISSREF)
for SNAC simple.

State Machine Changes (Ted Lemon)
Ted reviewed state machine changes. The working copy is at
https://author-tools.ietf.org/api/export/a530d761-ffab-41a4-9bcc-b26dbb49a420/draft-ietf-snac-simple.html
See Section 7. The general idea is to compose multiple "simpler" state machines
instead of 1 or 2 complex ones. Discussed that an (ASCII art) state graph is
not a good idea for this document.

Ted: December 1 deadline may be optimistic.
Lorenzo Colitti: detailed comment on 7.4.5.1;
Ted: it's incorrect to fix DHCP-PD, consider how it should work. Engineer
wanted to shared the DUID. That's not the intention. So we should clarify this
more - DUID per BR (SNAC router) to be unique. Ted: it would be nice if
prefixes could be shared, but it's a hard problem. Doing it over stub network
(Thread e.g.) is not the solution. Eric: ASCII art could help for diagrams.
Leading text 7.x says it's not normative. Correct? Ted: it's useful to call
these items out in the state machine section. The FSM description is intended
to help implementers. Not doing ASCII art. Eric: we need to make it informative
as it is mainly local implementation. Otherwise, it freezes the spec. Ted
agrees. Daren: Asked which parts of state machine are normative. Ted: Doesn't
have an easy answer, there are several alternatives. Lorenzo: Section 5 is not
fully clear what implementer needs to do. So Section 7 is needed to be
interoperable at this moment. If we add all normative things, then in the end
Section 7 / state machines could be moved to Appendix. Bob: we had other docs
(ND, VRRP, etc.) that ran into the same issues. we should be able to find
solutions and keep the state machines in the document. Tim: echo what Bob says.
ND has a lovely appendix (pointing to the normative text). State machines
easier to understand for implementer. Lorenzo: agree - could be state machine
or otherwise. Ted: so I'll update the document, do the work, then the people
making these comments could check if it's underspecified or not. I'll take a
pass and ask this question during WGLC again.

Multi-AIL （30min, Esko Dijk）
(presented slides)
https://datatracker.ietf.org/meeting/124/materials/slides-124-snac-snac-simple-multi-ail-situations-00
Stuart Cheshire: Asked if this is a problem that needs to be fixed. Not sure
this is directly related to SNAC. Esko: Will be clearer on next slides.
Lorenzo: Asked about another specific case. Issues relating to device
connecting to multiple networks but in the same AIL. Lorenzo: Slide 6. Why do
you want to do this. Will not work well. Don't do anything. Also asked about if
hosts need to advertise global addresses. Tim Winters: Feels like we will want
a better solution for multiple AILs later. Not sure we understand this well
enough. Ted Lemon: Also doesn't think this is a problem for SNAC-Simple. Could
be done later in SNAC-Complex. Maybe we could do something to detect broken
configurations. More important for Thread. Lorenzo: Suggest some text that says
what is in or out of bounds. Mechanism to choose one AIL. SNAC Simple doesn't
work support multiple AILs. Stuart: Agree with Lorenzo, don't want to solve all
problem we can think of. We should make failures clear and understandable.
Eric: As AD, wants to get SNAC simple shipped. Suggested simple text for
current document how the system will fail  in such a case, and this requires
standard/interoperation. Multiple AILs be done later in snac-complex even with
multiple uplink, so keep the work, but outside the scope of snac-simple.

Any other business
Eric: Reminer that NomCom feed back is still open. Please do that.
Esko: How to progress open issues. State machine biggest issue.
Editorial change scan be fixed after w.g. last call. Suggest sending email
about state machine changes to the list.