15:00:54 <tmorin> #startmeeting bgpvpn
15:00:56 <openstack> Meeting started Tue Nov 24 15:00:54 2015 UTC and is due to finish in 60 minutes.  The chair is tmorin. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:57 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:01:00 <openstack> The meeting name has been set to 'bgpvpn'
15:01:06 <enikher> hello
15:01:15 <tmorin> hi enikher
15:01:25 <matrohon> hi
15:01:36 <tmorin> hit doude, matrohon, pcarver_, timirnich_
15:01:45 <tmorin> hi wdeclercq
15:01:51 <wdeclercq> hey tmorin
15:01:57 <doude> tmorin: hi
15:01:58 <timirnich_> salut tmorin
15:02:33 <matrohon> timirnich_, meeting en fran�ais?
15:02:44 <tmorin> guten Tag Tim
15:03:15 <tmorin> matrohon: we would have to teach utf8 to our IRC clients first :)
15:03:18 <tmorin> let's start
15:03:36 <timirnich_> :-)
15:03:39 <tmorin> #topic announcements
15:04:04 <tmorin> we had quite a few things done this week
15:04:31 <tmorin> we'll discuss drivers separately
15:04:49 <tmorin> looking at non-driver things, there is the backport works going on
15:05:06 <enikher> kilo backport seems to be ready
15:05:21 <enikher> only thing left is changing requirements.txt
15:05:30 <enikher> blocked by ci of OS
15:05:56 <tmorin> yes, we still haven't resolved this
15:06:14 <tmorin> I've exposed our issue to the openstack-dev list, including the technical committee
15:06:24 <tmorin> the feedback is reasonably good
15:06:43 <tmorin> but we aren't at a stage where we can merge changes that touch our requirements file
15:06:49 <matrohon> the discussion is ongoing also on : https://review.openstack.org/#/c/243610/
15:06:54 <tmorin> yes,
15:07:36 <tmorin> #link http://lists.openstack.org/pipermail/openstack-dev/2015-November/thread.html#79773
15:07:40 <tmorin> for the mailing list discussion
15:07:45 <matrohon> #link : http://lists.openstack.org/pipermail/openstack-dev/2015-November/078301.html
15:07:49 <tmorin> #link https://review.openstack.org/#/c/243610/
15:08:00 <matrohon> ttx annoucment ^
15:08:39 <tmorin> yes, so listening to tc tech lead, we should be able to work in our backport branches
15:08:47 <enikher> but I don't think they will accept this commit or?
15:09:03 <tmorin> but we don't have an agreement on making the problematic job being disabled for us
15:09:10 <tmorin> I don't know yet
15:09:30 <tmorin> we possibly need to offer another alternative than making the job non-voting
15:09:50 <tmorin> such as purely disabling it (to avoid having a particular rule for our project)
15:10:05 <tmorin> or such as disabling it except for stable/x branches
15:10:17 <tmorin> I'll try to propose something today
15:10:53 <enikher> the gate-networking-bgpvpn-requirements job is not listed in the commits but jenkins still gives a -1
15:11:08 <tmorin> it is not visible
15:11:14 <tmorin> unless you click "toggle CI"
15:11:20 <enikher> https://review.openstack.org/#/c/248612/
15:11:27 <matrohon> enikanorov_, it's because you're modifying the requirements.txt
15:11:27 <tmorin> in which case the diagnostic of the check requirement job is provided
15:11:54 <tmorin> this is wha you see with "toggle CI" :: gate-networking-bgpvpn-requirements http://logs.openstack.org/12/248612/7/check/gate-networking-bgpvpn-requirements/1828400/ : Incompatible requirement found; see http://docs.openstack.org/developer/requirements/ in 35s
15:11:56 <enikanorov_> matrohon:
15:11:58 <enikanorov_> :D
15:12:15 <enikher> matrohon: although the job of ci is not liseted?
15:12:51 <janscheurich> hi all, sorry for being late.
15:13:14 <matrohon> enikanorov_, oops wrong completion, sorry
15:13:45 <matrohon> enikher, this test doesn't appear like a normal job
15:14:50 <tmorin> it is badly formatted, but it is here
15:14:57 <matrohon> enikher, that's what tmorin is trying to fix with its patch  : https://review.openstack.org/#/c/243610/
15:15:20 <enikher> Yes I have seen that and also folowed the thread.
15:15:51 <matrohon> let's keep discussing this on review
15:16:34 <tmorin> ok
15:16:44 <matrohon> I recall taht we wo'nt be able to merge if requirements job are failing
15:16:54 <matrohon> We already tried by the past
15:16:57 <tmorin> matrohon: yes, indeed
15:17:23 <tmorin> ok, another thing since last week is that enikher has started to work on tempest testing
15:17:29 <matrohon> that's the main issue, even with the ttx's proposal of maintaining our own branches
15:17:34 <tmorin> sorry :)
15:17:52 <matrohon> ok let's move forward
15:17:55 <tmorin> ok :)
15:18:00 <matrohon> #topic tempest
15:18:05 <enikher> yes
15:18:10 <tmorin> thanks enikher for starting this work
15:18:15 <matrohon> enikher, +1
15:18:28 <matrohon> can you sum up the situation enikher
15:18:33 <enikher> the first test is already running - but we have a issue with the lcient
15:18:37 <enikher> client
15:18:40 <tmorin> the next step is review/merge, and then to create a jenkins/zuul config using this job
15:19:01 <enikher> Here is the commit: https://review.openstack.org/#/c/246509/
15:19:08 <tmorin> enikher: what command to you use to simply run the one tempest case you added ?
15:19:11 <matrohon> #link : https://review.openstack.org/#/c/246509/
15:19:17 <enikher> I have added now a tempest client, but I don't like that.
15:19:56 <enikher> python -m testtools.run networking_bgpvpn_tempest.tests.api.test_bgpvpn.BgpvpnTest.test_create_bgpvpn
15:20:30 <matrohon> enikher, you said that every project is using it's own client?
15:20:36 <enikher> you have to install tempest first. I use ./run_tempest.sh -N -s
15:20:59 <enikher> nova/neutron/manila
15:21:09 <enikher> have all own clients in tempest as well.
15:21:34 <enikher> manila has even tow in the same repo. one for tempest plugin, one for cli
15:21:48 <tmorin> enikher: I suggest we first investigate the pros/cons of each approach, and then see if we switch to using the bgpvpn python-neutronclient extension
15:21:48 <enikher> tempest is a mess :-)
15:21:53 <matrohon> seems really messy
15:21:55 <matrohon> +1
15:22:25 <matrohon> I heard that several project get rid of tempest and use rally instead
15:22:46 <tmorin> I'm volunteering to work on the zuul/jenkins config needed to create a tempest job leveraging the new plugin
15:23:06 <matrohon> tmorin : our gate guru!
15:23:27 <enikher> I can do some work to check if we can include python-neutronclient
15:23:44 <wdeclercq> I thought rally is mainly used for multiple threads/concurrency and stuff
15:23:44 <enikher> tmorin: do you have a good wiki how to do that?
15:24:37 <boris-42> matrohon: wdeclercq it can be used for functional testing as well
15:24:42 <tmorin> enikher: no that I know about
15:24:48 <matrohon> wdeclercq, that was its main target, but rally is trying to bypass tempest when its first implemention was relying on it
15:24:49 <boris-42> matrohon: wdeclercq just remove concurrency and you'll get functioanl test
15:25:13 <wdeclercq> okay then :)
15:25:22 <boris-42> matrohon: wdeclercq so basically first implementation didn't rely on tempest.. it was seperated framework since begging-)
15:25:36 <matrohon> boris-42, ok thanks
15:25:39 <tmorin> enikher: looking at existing jobs and taking "inspiration" is what I do
15:25:40 <boris-42> )
15:26:29 <matrohon> neutron also improved its own functional test framework, with full-stack
15:26:50 <enikher> tmorin: can you send me a link to exsisting? I am new to jenkins
15:27:14 <tmorin> enikher: I'm fairly new too, matrohon was teasing me when saying guru
15:27:17 <pcarver> enikher: it's the project-config repo
15:27:43 <enikher> so there is no jenkins website?
15:27:45 <pcarver> https://review.openstack.org/#/admin/projects/openstack-infra/project-config
15:27:45 <tmorin> enikher: this change is an example https://review.openstack.org/#/c/243725/
15:28:05 <tmorin> enikher:  fungi had a good blog post, but I can't find it now
15:28:20 <pcarver> enikher: I'm not aware of one. What I've done in the past is clone the project-config repo, edit files and push a git-review just like any code change
15:28:24 <enikher> ok, we can discuss that later.
15:28:41 <enikher> learning me is not the main topic of that meeting :-)
15:28:47 <tmorin> enikher: look at the change above, you'll have a first idea
15:28:52 <tmorin> next topic then :)
15:28:54 <enikher> ok thanks
15:29:02 <tmorin> #topic router association
15:29:20 <matrohon> tmorin are you talking about : http://www.joinfu.com/2014/01/understanding-the-openstack-ci-system/
15:29:36 <tmorin> matrohon: yes, this is a very good blog post
15:29:44 <tmorin> tmorin: good highlevel starting point
15:29:47 <fungi> tmorin: yeah, that's asselin, not me
15:29:51 <matrohon> let's keep on trying to use tempest for the moment, since the work is bootstrapped
15:29:53 * fungi doesn't "blog"
15:30:00 <tmorin> fungi: hah, sorry :)
15:30:38 <fungi> i stand corrected, that's jaypipes' blog
15:30:38 <tmorin> so, back to router association
15:31:05 <tmorin> fungi: ah, thanks for the precision
15:31:25 <tmorin> back to router associations...
15:31:53 <tmorin> thanks wdeclercq for the contribution (both API, client and reference driver)
15:32:01 <matrohon> wdeclercq, +1000
15:32:06 <tmorin> yes :)
15:32:16 <tmorin> we have to review it
15:32:17 <wdeclercq> you're welcome :)
15:32:25 <matrohon> #link : https://review.openstack.org/#/c/249230/
15:32:58 <tmorin> it looks pretty clean
15:33:02 <wdeclercq> 1 thing I'll mention. if I run this "neutron-db-manage --subproject networking-bgpvpn --database-connection sqlite:// check_migration" which is inside the tox thing
15:33:13 <tmorin> I think we'll try to merge before we do a release
15:33:14 <wdeclercq> it doesn't seem to output my added DB change
15:33:40 <wdeclercq> however, I can drop the neutron database and db-manage will create it with the router association table, so I guess it's fine
15:34:17 <tmorin> wdeclercq: we have to investigate what check_migration is supposed to check, and why it does not consume all alembic changes
15:36:34 <tmorin> wdeclercq: we don't have any automated testing yet, I'm assuming you have done manual testing, in particular on the bagpipe driver ?
15:37:13 <wdeclercq> yeah, that's how I ended up seeing those port_update notification issues of L3 like I mentioned last week
15:37:39 <tmorin> I'll do some as well
15:37:44 <tmorin> wdeclercq: yes
15:39:43 <tmorin> next topic ?
15:39:52 <tmorin> #topic bagpipe driver
15:40:14 <tmorin> wdeclercq (see above) found a bug, which we discussed together
15:40:30 <tmorin> #link https://launchpad.net/bugs/1517480
15:40:30 <openstack> Launchpad bug 1517480 in bgpvpn "port-delete fails" [Critical,In progress] - Assigned to Thomas Morin (tmmorin-orange)
15:40:43 <tmorin> I pushed a fix today
15:41:05 <tmorin> along with a vastly improved unit test to detect issues related to neutron registry callbacks
15:41:10 <tmorin> #link https://review.openstack.org/#/c/249081/
15:41:43 <tmorin> by the way, wdeclercq, it would be nice if the router association code could have a similar unit test
15:42:07 <wdeclercq> Okay, I'll see what has to be done to be in line with networks
15:42:11 <matrohon> tmorin : great improvments, I had not much to say on the first patch
15:42:12 <tmorin> thanks wdeclercq for finding this bug, that was the opportunity to improve the unit tests
15:42:21 <matrohon> tmorin : I'll review new one soon
15:43:00 <tmorin> matrohon: thanks
15:43:04 <tmorin> next topic ?
15:43:11 <matrohon> so we're postpoing our first release to have router associations in?
15:43:39 <tmorin> let's review the change first
15:44:14 <matrohon> aprt from bagpipe know of our current driver implement it
15:44:14 <tmorin> if, as it seems, not much additional work is needed to merge it, then yes, I think we can release after
15:44:23 <tmorin> matrohon: correct
15:44:48 <matrohon> sorry for my poor english, too speed at writing :(
15:45:14 <matrohon> ok let's review first
15:45:20 <tmorin> we still need to check what behavior we have when trying to do a router association on a non supporting driver
15:45:31 <tmorin> silently doing nothing is possibly the current behavior
15:46:40 <matrohon> let's think about it
15:46:48 <tmorin> maybe we should default to having the router association method in BGPVPNDriverBase raise an Exception
15:47:02 <tmorin> so that we have something sane for non-ready drivers
15:47:11 <tmorin> yes, let's think about it
15:47:24 <matrohon> #topic driver status
15:47:47 <tmorin> #topic ODL driver
15:47:55 <tmorin> since we already covered bagpipe
15:48:11 <tmorin> the ODL driver has been submitted for review in networking-bgpvpn
15:48:25 <tmorin> #link https://review.openstack.org/#/c/247096/
15:48:49 <tmorin> the change is waiting for a new PS after a first review
15:49:17 <matrohon> I gave my +1 on the light implementation in the odl repo : https://review.openstack.org/#/c/231321/
15:49:27 <tmorin> the amount of work left is reasonably low
15:49:45 <tmorin> I'm waiting for cleaner exception being raised before giving a +1
15:49:56 <tmorin> the amount of work left is low enough
15:50:02 <matrohon> vthapar wanted to have a first implementation ready in odl repo, and cooked a finest one in bgpvpn repo
15:50:15 <matrohon> I'm fine with that
15:50:20 <tmorin> why not
15:50:50 <tmorin> but I think only a reduced amount of work is needed to have something satifsying to merge in networking-bgpvpn
15:51:12 <tmorin> doable before a first release if vishal finds the time to work on it, I'd think
15:51:12 <matrohon> he had some issue with mock
15:51:26 <tmorin> matrohon: what kind ?
15:51:53 <matrohon> and he also start to implement the revert of a error in bgpvpn deletion
15:52:30 <matrohon> for the deletion : he wants to rely on a ODL framework that is not currently available
15:52:51 <matrohon> tmorin : I don't really, I have to test it locally
15:53:34 <tmorin> yes, but I think we understand that reverting is complex to do with ODL and that it can be refined later (does not seem specific to BGPVPN in fact, AFAIU)
15:53:41 <janscheurich> Can't we use the driver from networking-odl in the first release if that is good enough and does not have unmet external dependencies?
15:54:07 <matrohon> janscheurich, that's the idea
15:54:32 <janscheurich> But then we should have that patch merged in networking-bgpvpn, or not?
15:54:33 <tmorin> janscheurich: the driver from networking-odl would be usable with the first release of net-bgpvpn, but would not be "in" the first release
15:54:39 <matrohon> our driver API shouldn't change, unless we have a very very good reason to modify it
15:55:32 <janscheurich> OK, if that is what you have agreed with vishal
15:56:03 <matrohon> I think its idea is to keep on working in the odl repo
15:56:16 <matrohon> but having a beta release in the bgpvpn repo
15:56:57 <matrohon> contrail?
15:57:01 <matrohon> doude?
15:57:03 <tmorin> I think we can plan to have both versions repo mentioned in the odl driver doc that will sit in networking-bgpvpn/doc/source/odl
15:57:06 <matrohon> only 3 min left
15:57:21 <tmorin> I forgot to mention it in the announcements
15:57:33 <doude> matrohon: yes
15:57:42 <doude> ?
15:58:00 <matrohon> #annoucement : the contrail driver has merged
15:58:01 <tmorin> the OpenContrail driver merged last week !
15:58:12 <tmorin> that's it :)
15:58:26 <tmorin> anything to add on the next steps doude ?
15:58:26 <matrohon> #link : https://review.openstack.org/#/c/202806/.
15:58:56 <tmorin> wdeclercq: anything to announce on nuage side, about a nuage driver ?
15:59:12 <doude> tmorin: for the contrail driver ?
15:59:23 <tmorin> doude: yes
15:59:24 <wdeclercq> hm no, I actually have no news of that. I can ask around and inform next week
15:59:44 <doude> tmorin: I'm still working to add the bgpvpn resource to the contrail data model
15:59:47 <tmorin> before closing, I want to really insist that we are really happy to see new contributors coming, making networking-bgpvpn more diverse and making good progress
15:59:55 <doude> then I'll update the driver to use it
16:00:06 <tmorin> doude: ok
16:00:17 <matrohon> tmoin : +100000
16:00:22 <matrohon> tmorin : +100000
16:00:25 <doude> for the moment the work is on the contrail side
16:00:47 <tmorin> ok, we have to free the floor
16:01:05 <tmorin> this was a very good week for new contribut(ion|or)s
16:01:11 <matrohon> thanks everybody
16:01:12 <tmorin> keep up the good work !  ;)
16:01:16 <tmorin> thanks everyone
16:01:22 <enikher> thanks and bye :-)
16:01:26 <janscheurich> bye
16:01:27 <wdeclercq> byebye o/
16:01:31 <tmorin> #endmeeting