Agenda for the NETMOD 119 WG Session

https://datatracker.ietf.org/meeting/119/materials/agenda-119-netmod
https://datatracker.ietf.org/meeting/119/session/netmod

Session:

Thursday, March 21, 2024
13:00-15:00 Brisbane (Australia - Eastern Time)
03:00-05:00 UTC
23:00-01:00 Wednesday March 20 America - Eastern Time
https://www.timeanddate.com/worldclock/converter.html?iso=20240321T030000&p1=47

Room: M2 https://datatracker.ietf.org/meeting/119/floor-plan?room=m2

WG Chairs:

Lou Berger (lberger at labs dot net)
Kent Watsen (kent plus ietf at watsen dot net)

WG Secretary

Jason Sterne (jason dot sterne at nokia dot com)

Available During Session:

MeetEcho: https://meetings.conf.meetecho.com/ietf119/?session=31987
Onsite tool:
https://meetings.conf.meetecho.com/onsite119/?session=31987
Audio Only: https://mp3.conf.meetecho.com/ietf119/31987.m3u

Available During and After Session:

Notes: https://notes.ietf.org/notes-ietf-119-netmod?both
Slides: https://datatracker.ietf.org/meeting/119/session/netmod
Zulip (chat):
https://zulip.ietf.org/#narrow/stream/126-netmod/topic/ietf-119
Drafts (TGZ):
https://datatracker.ietf.org/meeting/119/agenda/netmod-drafts.tgz
Drafts (PDF):
https://datatracker.ietf.org/meeting/119/agenda/netmod-drafts.pdf
Datatracker: https://datatracker.ietf.org/group/netmod/about/
ICS: https://datatracker.ietf.org/meeting/119/session/31987.ics

Available After Session:

Recording: http://www.meetecho.com/ietf119/recordings#NETMOD
Jabber Logs: https://www.ietf.org/jabber/logs/netmod

1) Session Intro & WG Status (10 min)

Presenter: Chairs

Notes:

Lou Berger: Giving an intro. Need more reviewers for acl-extensions
module.
Lou Berger: Change WG policy - want authors to provide a status update
for every WG adopted document.

Mahesh Jethanandani: Thank you to Rob for his service.
Also a Syslog update - had dependency on client/server sutie of drafts.
Authors can probably update syslog drafts.

Kent Watsen: Syslog draft has normtive dependencies on Crypto-types &
TLS client server. But only on client side of it (not server side) so
we're good.

Mahesh Jethanandani: Perhaps other chairs would like to adopt a similar
position to WG documents (new policy to have status update for every WG
doc).

Joe Clarke: I agree with the new policy. Great idea. There is a -32
syslog live, and we think that should be mostly good to go.

Scott Mansfield: Comment about no liaisons, but put a liaision in the
chat window (https://datatracker.ietf.org/liaison/1890/). We have a
group in the ITU that gets together monthly to discuss YANG related
topics with various groups (ITU, IETF, IEEE, etc). If you are interested
in that coordination then look at chat and ask Scott for further
details.

Lou Berger: Thank you Rob, looking forward to contributing as a WG
participant.

Chartered items:

2) Common Interface Extension and Sub-interface VLAN YANG Data Models (10 min)

Start: 13:11
Presenter: Scott Mansfield
Draft:
https://datatracker.ietf.org/doc/draft-ietf-netmod-intf-ext-yang-13
Draft:
https://datatracker.ietf.org/doc/draft-ietf-netmod-sub-intf-vlan-model-10

Notes:

Rob Wilton: There was someone who wanted to leverage this. Should we try
to include their use case?

Scott Mansfield: We should take to the list, and look to extend this.
Their use case is a good example.

Kent Watsen: That answers my question which was next steps. We'll take
it to the list.

3) YANG Versioning (30 min)

Start: 13:14
Presenter: Rob Wilton

a - YANG Versioning Update

Start: 13:17
Presenter: Rob Wilton
Draft:
https://datatracker.ietf.org/doc/draft-ietf-netmod-yang-module-versioning-11

Notes:

Lou Berger: About filenames, what does the document say in the end about
them

Rob Wilton: It doesn't say anything about them. The section was deleted.

Lou Berger: We may end up with a de-facto standard and that may not be
good.

Mahesh Jethanandani: Agree with Lou that we should try to resolve the
issue of filename rather than deferring it. Even if the concensus is
rough.

Balazs Lengyel: As a company that already uses this filename notation, I
think it would be very useful. But we've been debating for too long so
we need to finish this.

Rob Wilton: Per tried tooling changes to pyand and it was a trivial 2
line change.

Joe Clarke: This filename thing would be in Semver (not Module
Versioning)

Kent Watsen: Wouldn't you have to modify this document first (for
filenames) before LC?

Rob Wilton: No. The filename issue would be in Semver.

Kent Watsen: We can certainly take it to last call after this meeting.

Lou Berger: I was hoping we could do LC on the 2 docs together.

Rob Wilton: Either way is fine with us.

Discussion about filename in the chat:
Andy Bierman: revision-label in filename is NBC change to YANG 1.1 so
needs to wait for YANG-next

Robert Wilton: Hi Andy, I guess you mean an NBC change on tooling. We
wouldn't be taking away the old revision date scheme so that would still
work.

Italo Busi: I think that as soon as revision labels are going to be
used, there will be the need to deliver different versions of the same
YANG model in one day. This could not be done because the revision-date
MUST still be unique in YANG 1.1. I therefore prefer to complete this
work and to start YANG-next quickly rather than fixing a naming issue

b - NETMOD YANG Semver Update

Start: 13:32
Presenter: Joe Clarke
Draft:
https://datatracker.ietf.org/doc/draft-ietf-netmod-yang-semver-14

Notes:

Joe Clarke: Next steps was going to be WGLC, but may need to update on
the filenaming.

Lou Berger: We should try and have a focused consensus call on
filenaming, and then more the documents to WG LC.

James Cumming: Are we planing to augment into modules-state? (in
addition to YANG library)

Joe Clarke: I can't remember whether we did it. Q. in NETCONF was how
much longer do we keep updating deprecated subtree. Personal opinion is
just do it as it is easy. But I don't think we addressed it yet.

Kent Watsen (as contibutor): Want to address the elephant in the room.
Expect module versioning draft is likely to be less contentious. Can you
help explain why this YANG Semver draft won't cause problems?

Joe Clarke: (...explains how YANG Semver works...)

Kent Watsen: Are you saying that it is opt-in? There is a graceful
fallback to existing mechanisms?

Joe Clarke: Yes. If tools don't understand the new extensions, they will
make decisions as per before. With the new extensions, problems may be
flagged up front vs hit later in the process.

Maria Matějka: You are introducing a versioning scheme with a string may
be hard for tools.

Joe Clarke: Introducing a string with some semantic meaning to how it is
constructed. It does have some structure.

Maria Matějka: Questioning whether a structure within a structure is
okay.

Joe Clarke: Didn't want to make it overly complicated. Wanted to use the
industry Semver rules. Didn't want to create lots of new concepts, but
wanted something recognizable. But we needed branching support.

Kent Watsen: Please take this to the list.

Rob Wilton: Regular semver allows extra bits at the end so it is
effectively a string anyway.
Mahesh(in chat): I think what Maria is saying is that why we are using
string in the version field (for the semver portion of the field)?

Maria Matějka (in chat): yes, the question is, if the string is actually
"major.minor.patch_whatever", then why this isn't like grouping semver
{
uint32 major;
uint32 minor;
uint32 patch;
string whatever;
}

Non-Chartered items:

4) Validating anydata in YANG Library context (10 min)

Start: 13:56
Presenter: Ahmed Elhassany
Draft:
https://datatracker.ietf.org/doc/draft-aelhassany-anydata-validation-00

Notes:

Alex Huang Feng: This is very interesting work. You have two conditions.
The first one is restricting YANG. I am supportive of that (stating that
anydata can be validated using YANG rules). The second one, the open
issue about complete vs incomplete. I have some mixed feelings. At what
point we are stating that anydata is valid, even if there is mandatory
statement. Could add the xpath or subtree as a parameter of the method.
Or two flags that were proposed in a thread. One way of validating
complete data and one way of validating incomplete data. Solution
without modifying the rules of YANG.

Benoit Claise: Mentionded that anydata is used in multiple places. Is
this solution just for YANG Push telemetry, or is that something wider.

Ahmed Elhassany: Solution is to solve any of them, but YANG Push is the
primary use case.

Balazs Lengyel: In 3GPP we have situations where we store interesting
attributes, but we don't know in advance what that will be. The complete
vs incomplete sounds like validation against candidate configuration.

Jean Quilbeuf: You might want to specify the path of the anydata, you
may want to have a mapping from the anydata node to the YANG library.

Ahmed Elhassany: If I understand your point, even if you have multiple
anydata leaves you will still be able to figure out where they came from
and be able to validate them.

Jean Quilbeuf: If you want to validate, for instance, your anydata1
contains ietf interfaces and anydata2 contains yang library, you want to
specify that so if you put yang library in anydata1 you have an error.

Ahmed Elhassany: We can restrict anydata to really specific data. Not
trying to restrict which data goes in. But if there is data that goes
in, we can validate it against a model.

Lou Berger: Let's take this to the list to continue.

5) Philatelist and YANG time-series db (10 min)

Start: 14:07
Presenter: Jan Lindblad
Draft:
https://datatracker.ietf.org/doc/draft-lindblad-tlm-philatelist-00
Draft: https://datatracker.ietf.org/doc/draft-kll-yang-label-tsdb-00

Notes:

Thomas Graf: should document be informational?

Jan Lindblad: we have a YANG so it could be normative

Thomas Graf: We have an NMOP WG and I've been doing this as
informational document, but then the protocol changes are going into
NETMOD or NETCONF

Lou Berger: We need to take this discussion to the list

7) YANG Full Embed (10 min)

Start: 14:17
Presenter: Jean Quilbeuf or Benoit Claise
Draft:
https://datatracker.ietf.org/doc/draft-jouqui-netmod-yang-full-include-01

Notes:

Balazs Lengyel: This would be very interesting for 3GPP YANG modules,
where have a basic problem where the root of our YANG modules can either
be at the sub-network level or the management element level. Second
question, you have just put a single YANG module, but usually it would
need to be a tree or hierarchy of modules.

Jean Quilbeuf: For tree of module, we can have a module embedded in a
module embedded in a module.

Scott Mansfield: This is solving a problem that we have seen in a
liaison in broadband forum. To the chairs: if you are ready/interested
then please reach out.

Alex Huang Feng: Very interesting work, and I support. Are you allowed
to include parts of a YANG module or do you embed the full tree.

Jean Quilbeuf: This is covered on the next slide, point 3. This is a
good question. It is something we are considering.

James Cumming: Very interesting work. Any thoughts about how to deal
with leaf-refs.

Jean Quilbeuf: Leaf-ref from within the embedded part can only target
embedded part. The module embedding another module could have leaf-ref
to embedded part.

Rob Wilton: Is this too big of a change for YANG, does it need to be
done in a new version of YANG? Might also want to look at the YANG
packages work that also investigated some of these issues.

Kent Watsen: +1 support for this work, much better than schema mount.

From the chat:

Mahesh Jethanandani: @Jan, not to put you on the spot, but doesn't
tail-f have a solution that kind of solves the problem that Jean is
talking about? How does it align with what Jean is proposing.

Jan Lindblad: @Mahesh, the solution we have is more similar to
schema-mount, where you don't know what YANG modules are mounted at
design time. Recently, I have stumbled over the same/similar need for
something like full-include, so I think this is interesting to discuss,
even if I'm also a bit worried about the deeper implications of this
addition.

8) Applying COSE Signatures for YANG Data Provenance (10 min)

Start: 14:28
Presenter: Diego R. Lopez
Draft:
https://datatracker.ietf.org/doc/draft-lopez-opsawg-yang-provenance-02

Notes:

Kent Watsen: It sounds like you are still experimenting on this. I
suggest to bring it back and give an update when you are ready and we
can guage interest.

Diego Lopez: Yes, that sounds good.

9) DTN Management Architecture (10 min)

Start: 14:36
Presenter: Ed Birrane
Draft: https://datatracker.ietf.org/doc/draft-ietf-dtn-dtnma-12

Notes:

Mahesh Jethanandani: Brought in the NETMOD WG to look at the language
and protocol perspective. On a separate note, looking at the Venn
diagram makes it look like there is no overlap.

Ed Birrane: We need to make that bigger - agree.
Kent Watsen: Very interest work. Reason why it is here to see if YANG as
is, is suitable.

Ed Birrane: We have 2 documents in the DTN WG on how we can build and
model this system with unrestricted YANG. The question is "Are we doing
it right?"

Kent Watsen: Both RPCs and Action both have input and output. Ideally
would have a way to indicate that output statements aren't allowed.

Kent Watsen: Comment on the chat to also take it to YANG Doctors.

10) A Common YANG Data Model for Scheduling (10 min)

Start: 14:40
Presenter: Qiufang Ma
Draft:
https://datatracker.ietf.org/doc/draft-ma-opsawg-schedule-yang-04

Notes:

Kent Watsen: This idea has been presented in the past, and the comments
back then were "Why do we need scheduling on the device while controller
could push changes to the device?". IMO the reason this is necessary
might be for autonomous systems that are not controlled. I think this is
useful.

Joe Clarke: The ACL work is close to last call, and wanted to some
quickish indication of what to do.

First poll: Who is interested in this topic?

Result: good support, almost half present

Second poll: who thinks this is a good foundation for our work on this topic?

Second poll results: a suprsing number given late addition to agenda.

Lou Berger: For a late addition there is actually pretty good support.
Chairs will discuss offline, but it looks like we are likely to run an
adoption poll.

On the chat:

Vishnu Beeram:
https://datatracker.ietf.org/doc/draft-liu-netmod-yang-schedule/ -- this
is the old ietf-schedules draft that was discussed in the past..

6) A YANG model for Device Power Management (10 min)

Start: 14:54
Presenter: Tony Li
Draft: https://datatracker.ietf.org/doc/draft-li-ivy-power-01

Notes:

Kent Watsen: Matthew Palmer and I have a patent with Juniper in this
area. Power based routing, based on how green it is. If power is solar,
then prefer that over other energy sources. Do it in layers.

Mahesh Jethanandani: I think that there is enough community interest to
at least have a BOF. I hope that the interested parties will bring a BOF
proposal for IETF 120.

Qin Wu: For this draft, we are interested, we have a similar draft in
Ivy. When you disable one of the components, do you need to disable the
other dependent coponents?

Balazs Lengyel: There was some workshop in the IETF about energy a few
months ago. I think Jan Lindblad was there. It would be interesting to
hear what their comments were and what they proposed for such topics.

Tony Li: Dependencies is to avoid turning off things that are still in
use. i.e. Don't turn off switch fabric if there are line cards still
using it.

Unallocated time: 0 min