15:07:18 <tmorin> #startmeeting bgpvpn
15:07:18 <openstack> Meeting started Tue May 10 15:07:18 2016 UTC and is due to finish in 60 minutes.  The chair is tmorin. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:07:19 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:07:23 <openstack> The meeting name has been set to 'bgpvpn'
15:07:37 <tmorin> too long between two meetings and the good habits are lost ;-)
15:07:39 <tmorin> hi everyone
15:07:41 <tmorin> doude
15:07:44 <tmorin> hi doude
15:07:47 <tmorin> hi paul
15:07:48 <pcarver> hi
15:07:52 <tmorin> hi matrohon
15:07:57 <tmorin> hi timirnich
15:08:08 <matrohon> hi
15:08:21 <doude> hi
15:09:15 <pabelanger> o/
15:09:31 <matrohon> pabelanger, hi newcomers :)
15:09:48 <matrohon> pabelanger, interested in bgpvpn?
15:10:05 <pabelanger> matrohon: no, tripleo-ci :)
15:10:50 <matrohon> pabelanger, it seems you're in the wrong room!
15:11:01 <pabelanger> matrohon: I was going to just say that, sorry for the noise
15:11:15 <matrohon> tmorin, we have several topic to discuss today
15:11:21 * tmorin trying to see who else would join
15:13:05 <tmorin> let's start
15:13:18 <tmorin> go ahead matrohon you seem to have an agenda in mind :)
15:13:40 <matrohon> the main topic is the neutron-stadium evolution
15:14:04 <tmorin> yes, and we'll discuss what we'll do for newton afterwards
15:14:05 <matrohon> https://review.openstack.org/#/c/312199/
15:14:20 <tmorin> the part that relates to bgpvpn currently says:
15:14:41 <tmorin> * networking-bgpvpn: the core team will have to:
15:14:41 <tmorin> * adopt neutron-api, and neutron-lib;
15:14:41 <tmorin> * complete docs, end-to-end testing, and demonstrate ability to upgrade;
15:14:41 <tmorin> * move client side code to python-neutronclient and switch to OSC plugin;
15:15:34 <tmorin> the things that is hard to envision yet is neutron-api : the neutron-api project isn't fully defined yet
15:15:56 <matrohon> well, I don't know if "adopting neutron-api" means we need to discuss about our current API
15:17:15 <tmorin> my assumption would be that what we have today would be an agreed starting point, and that only future evolutions would be coordinated via neutron-api
15:17:17 <matrohon> neutron-lib is already leveraged by bgpvpn, but we have to run our UT on neutron-lib changes
15:17:23 <tmorin> but I don't know
15:17:35 <matrohon> tmorin, ^I want to understand it this way too :)
15:17:53 <tmorin> neutron-lib: we depend on it, but there are a few things we still import from neutron and that we should import from neutron-lib directly
15:18:22 <matrohon> tmorin, it need improvement at the bgpvpn side so
15:18:36 <tmorin> matrohon: on neutron-api, we can't assess the impact on us until the project is defined --> hard to upvote/downvote https://review.openstack.org/#/c/312199/ without knowing what neutron-api means
15:19:07 <tmorin> yes, fully adopting neutron-lib will require a few changes, but mostly trivial changes I think
15:19:35 <tmorin> on neutron-api: I'll comment on the change about "we can't assess the impact on us until the neutron-aip project is better defined"
15:20:01 <matrohon> tmorin, I'm not sure we will be able to completely get rid of our neutron dependency, though
15:20:12 <matrohon> tmorin, +1
15:20:45 <tmorin> no, I don't think we willl be able to, at least until "everything required for a service plugin" moves in neutron-lib, but I don't know if it is a part of the plan
15:20:54 <tmorin> this will be incremental I think
15:21:43 <matrohon> another part that you didn't mentioned is the ability to upgrade seamlessly, which mean adopting oslo versioned object
15:22:05 <tmorin> another thing to discuss will be "demonstrate ability to upgrade": easy for the API/DB part (alembic), but more work for the bagpipe driver (how to handle evolution of the content of RPC)
15:22:20 <tmorin> ^ yes
15:22:29 <matrohon> neutron-lib plans to provide an api to help manipulating ovo for neutron
15:22:42 <tmorin> but I haven't looked close enough to know if versionedobject gives the full answer
15:23:01 <matrohon> tmorin, that's what I understood at austin
15:23:01 <tmorin> matrohon: would you have a link to share on this  ?
15:24:03 <matrohon> https://etherpad.openstack.org/p/newton-neutron-lib-next-steps
15:24:25 <matrohon> http://lists.openstack.org/pipermail/openstack-dev/2016-May/093876.html
15:24:35 <matrohon> ^ is more accurate
15:25:31 <tmorin> thanks
15:25:52 * tmorin will need to do a bit of reading :)
15:26:13 <tmorin> other items are:
15:26:16 <tmorin> doc
15:26:36 <tmorin> we are not starting from zero, but we know we can do better, this looks doable
15:27:00 <matrohon> so if we want to stay in the neutron-stadium, we have to highly prioritize those effort, considering the few developer we have on bgpvpn
15:27:09 <tmorin> end-to-end testing: we have made good progress, and we'll need more tempest test to have a significant coverage
15:27:26 <pcarver> tmorin: I still have the API docs on my todo list. After Tokyo I tried to figure it out but the docs on how to write API docs were a mess. It looks like that's improving.
15:28:27 <tmorin> pcarver: ok, it will be great if you can progress on that
15:28:34 <enikher> hey
15:28:47 <matrohon> armax is waking up, we can ask him about neutron-api!
15:28:53 <tmorin> hi enikher
15:29:00 <matrohon> hi enikher
15:29:20 <enikher> hey, I have 10 min left :-)
15:29:37 <tmorin> enikher: part of the neutron stadium requirements consist in end2end test coverage
15:29:50 <tmorin> I wanted to ask you if you had plans to write more tempest tests
15:30:22 <enikher> not right now
15:30:51 <enikher> we had the plans for the next release but at the moment cause gluon work I don't have much time for that
15:31:30 <enikher> however it should be quite stright forward to write some. What would help is a list of testcases we want to see
15:31:32 * tmorin :(
15:32:07 <enikher> If we have that list, implementing is quite easy
15:32:09 <tmorin> enikher: the two base test cases described in OPNFV SDNVPN (VM to VM pings) would be a good starting point
15:32:14 <enikher> I can help with that
15:32:41 <enikher> ok, these are bit tricky
15:32:51 <enikher> api testing is much easier
15:33:06 <enikher> but sure that end2end testing would be very nice
15:33:18 <enikher> the testing framework has to be enriched for that
15:33:36 <tmorin> enikher: that would get us closer to the goal, and this will help OPNFV side as well, won't it ?
15:33:40 <enikher> I will not have time in the next month but maybe after that
15:33:50 <enikher> yes it will
15:34:33 <tmorin> enikher: ok...
15:34:59 <enikher> Let me come back to you in June. when is the deadline?
15:35:36 <tmorin> no hard deadline, at least in the newton cycle for stadium things (as I understand)
15:36:14 <tmorin> but this is part of the things that would be nice to land for newton anyways, independently of what we'd need to comply with for the neutron stadium
15:36:18 <tmorin> last item of stadium "compliancy list" would be "move client side code to python-neutronclient and switch to OSC plugin"
15:36:52 <tmorin> it's certainly easy to do technically, just "one more thing" to add to the list
15:37:04 <tmorin> we'll see when/if/how the neutron stadium proposal is accepted
15:37:48 <tmorin> ok, next topic ?
15:38:04 <tmorin> matrohon, had you something specific in mind ?
15:38:43 <tmorin> our discussion in Austin on what we can target for Newton are in the pad at https://etherpad.openstack.org/p/newton-bgpvpn
15:39:19 <matrohon> no  specific thing, just to mention that it would a bad idea to start working on other topics than the one to stay in the neutron stadium
15:40:41 <enikher> sorry I have to go.
15:40:48 <tmorin> bye enikher
15:41:05 <tmorin> matrohon: really ? I'm not sure: the things to do to stay in the stadium wouldn't have to be done in the newton cycle
15:41:26 <matrohon> we have till ocata-1!
15:42:38 <tmorin> yes, so we can balance what we target in newton: some things to get closer to stadium requirements, some things to extend "featues"
15:42:42 <tmorin> features
15:43:59 <tmorin> my take is that we should do the E-VPN part, and start the API work on port assoc/static routes/aggregation now
15:44:37 <tmorin> not sure yet how E-VPN is high on the list, but we can see
15:55:50 <tmorin> ok, what else to cover ?
15:55:57 <tmorin> done for today it seems
15:56:17 <tmorin> more on what-to-do-for-newton and who-does-what next week!
15:57:53 <tmorin> bye everyone
15:59:27 <tmorin> #endmeeting