Minutes for NETCONF at interim-2014-netconf-3
minutes-interim-2014-netconf-3-2

Meeting start: October 6, 2014, 17:00 UTC

Meeting end: October 6, 2014, 19:05 UTC

Virtual meeting attendees:

* Kent Watsen
* Juergen Schoenwaelder
* Andy Bierman
* Mahesh Jethhanandani
* Mehmet Ersue (WG co-chair)

===================
#netconf-wg Meeting
===================

Meeting started by kwatsen at 17:19:54 UTC. The full logs are available
at netconf-wg/2014/netconf-wg.2014-10-06-17.19.log.html .

Additional information on the issue discussion is available on netconf-wg
github: https://github.com/netconf-wg/restconf/issues and
https://github.com/netconf-wg/server-model/issues

Meeting summary
-------------------

* Agenda Bashing (kwatsen, 17:20:04)
* meta meeting discussion (kwatsen, 17:21:03)
* need key participants (kwatsen, 17:21:21)
* focus is to go over issues as tracked by GitHub (kwatsen, 17:21:36)
* Who are key contributors for the meeting to happen? How are these
"key contributors" defined? (mj, 17:22:17)
* I wish we had more time to discuss, how about 80% of any mailing
list participants in drafts being discussed? (kwatsen, 17:23:40)

* RESTCONF Issue #2 (kwatsen, 17:24:45)
* LINK: https://github.com/netconf-wg/restconf/issues/2 (kwatsen,
17:25:28)
* kent says that auth mechanism may drive this (kwatsen, 17:27:15)
* juergen says for TLS, would want long-lived sessions too (kwatsen,
17:27:38)
* sounds like there is agreement to try to make it show up as a
session - we need to find how to satisfy the requirement (kwatsen,
17:30:36)
* looks like an agreement (Mehmet, 17:31:06)
* Juergen: schema announcement should be protocol independent
(Mehmet, 17:47:42)
* juergen says reason we're discussing 6022 is so RESTCONF clients can
discover module's supported (kwatsen, 17:48:30)
* there is a duplication concerning capabilities (Mehmet, 17:53:22)
* Monitoring can be kept protocol specific (Mehmet, 17:54:03)
* My requirement is that the same schema information is available to
both NETCONF and RESTCONF clients (JS__, 18:13:07)
* factor out module list into separate module (abierman, 18:13:19)
* netconf capability and restconf capability list are different
(abierman, 18:13:42)
* module list is mandatory to implement in restconf (abierman,
18:14:32)
* A generic session data model may be a separate piece of work.
(JS__, 18:14:45)
* this to address a catch-22 situation (kwatsen, 18:14:50)
* do not open RFC 6022 (abierman, 18:14:51)
* how to show restconf sessions in /netconf-state/sessions (abierman,
18:15:40)
* add identity restconf-tls (similar to netconf-tls) for transport
(abierman, 18:15:58)
* could not define any 6022 extensions now and define
restconf-specific extensions later (abierman, 18:16:57)
* what about streams monitoring? (abierman, 18:17:30)
* defer all session monitoring to a future draft (abierman, 18:19:16)

* RESTCONF Issue #7 (kwatsen, 18:21:14)
* there was a call on this list (kwatsen, 18:22:08)
* two weeks ago it was ~ 11-9 (almost a split) (kwatsen, 18:22:32)
* 11 for option 'D', 9 for option 'A' (kwatsen, 18:24:28)
* Mehmet says we should discuss at next IETF meeting (kwatsen,
18:24:50)
* Andy concerned it will delay the draft more (kwatsen, 18:25:09)
* Juergen says chairs should like to define a rough consensus
statement (kwatsen, 18:25:54)
* ACTION: Mehmet to discuss with Bert (kwatsen, 18:26:04)

* RESTCONF Issue #9 (kwatsen, 18:26:37)
* Juergen notes that netconf-tls uses certificates (kwatsen,
18:28:45)
* Kent says we should try that option first, to see how well it works
for REST (kwatsen, 18:29:19)
* Juergen suggests we find other REST protocols and copy them
(kwatsen, 18:29:56)
* Andy asks if RESTCONF has requirement to extract username (Juergen
confirms - for NACM) (kwatsen, 18:30:33)
* RFC 7285 says use HTTP digest for auth (abierman, 18:31:50)
* LINK: https://tools.ietf.org/html/rfc7285 (kwatsen, 18:31:59)

* Wrapping up RESTCONF (kwatsen, 18:38:23)
* charter dates have been exceeded (kwatsen, 18:38:53)
* fixed deadlines aren't helpful (kwatsen, 18:39:05)
* need to prevent feature-creep (kwatsen, 18:39:13)

* server-model #7 (Mehmet, 18:43:34)
* issue #7 Do we need enable/disable knobs (Mehmet, 18:44:24)
* no strong objections. So the issue is closed. (Mehmet, 18:45:42)
* Bert has started consensus call. (Mehmet, 18:45:58)

* server-model #10 (Mehmet, 18:46:00)
* Add support for Restconf? (Mehmet, 18:46:20)
* better to have separate modules so do not need to make everything
purely optional (abierman, 18:53:34)
* andy/kent say another model should be created specific for restconf
(kwatsen, 18:54:23)

* server model issue #12 (kwatsen, 18:56:11)
* Do you guys hear me? (Mehmet, 18:57:16)
* Andy and Juergen think that lot's of groupings make harder to read
(kwatsen, 19:05:48)
* Andy mentions that ODL also uses lots of groupings (kwatsen,
19:06:21)
* kent to collect more opinions (kwatsen, 19:07:10)

Meeting ended at 19:07:15 UTC.

Action items, by person
-----------------------

* Mehmet
* Mehmet to discuss with Bert on the result of the consensus call for "RESTCONF
modularity" and propose a way forward.

People present (lines said)
---------------------------

* kwatsen (41)
* abierman (11)
* Mehmet (11)
* meetbot (4)
* JS__ (2)
* mj (1)