15:02:25 <tmorin> #startmeeting bgpvpn
15:02:26 <openstack> Meeting started Tue Sep  5 15:02:25 2017 UTC and is due to finish in 60 minutes.  The chair is tmorin. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:02:27 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:02:29 <openstack> The meeting name has been set to 'bgpvpn'
15:02:38 <tmorin> hi everyone !
15:02:47 <tmorin> let's see who is around...
15:02:48 <pcarver> hi
15:02:55 <tmorin> hi pcarver !
15:03:04 <pcarver> I'm just back from vacation and not at all caught up on email or anything
15:03:21 <tmorin> pcarver: ok, that's fine :)
15:03:40 <tmorin> matrohon: hi.... around maybe ?
15:04:05 <tmorin> doude: hi... coming to bgpvpn weekly perhaps ?
15:04:15 <matrohon> hey tmorin !
15:04:25 <tmorin> hey :)
15:05:46 <tmorin> ok, I propose that we start
15:06:18 <tmorin> here are the things I had in mind as agenda
15:06:35 <doude> Hi tmorin
15:06:35 <tmorin> 1) pike release
15:06:43 <tmorin> hi doude ! nice to see you !
15:06:53 <tmorin> 2) routes-control API
15:07:03 <tmorin> 3) tempest tests
15:07:17 <tmorin> 4) horizon fixes in flight
15:08:00 <tmorin> 5) one driver framework evolution (bug 1709647)
15:08:02 <openstack> bug 1709647 in networking-bgpvpn "delete_net_assoc_precommit should be done before deleting bgpvpn net assoc" [Undecided,In progress] https://launchpad.net/bugs/1709647 - Assigned to Thomas Morin (tmmorin-orange)
15:08:16 <tmorin> 6) bagpipe driver work in progress (OVO & RPCs)
15:08:30 <tmorin> pcarver, matrohon, doude: other topics you'd see ?
15:08:56 <pcarver> nothing for me
15:08:56 <tmorin> #topic pike release
15:09:02 <tmorin> ok
15:09:14 <doude> Old Contrail driver deprecation
15:09:15 <matrohon> no, fortunately
15:09:25 <tmorin> doude: ok !
15:09:40 <tmorin> matrohon: :)
15:09:58 <tmorin> nothing much to say on bgpvpn pike release
15:10:16 <tmorin> one thing though: this time we didn't have anything to do
15:10:33 <tmorin> everything was handled by the release team as part of the openstack release cycle
15:10:59 <tmorin> which is nice I find
15:11:22 <tmorin> #topic routes-control API
15:11:51 <tmorin> now that we've started a new cycle, its the perfect time for working on new features
15:12:23 <tmorin> this one has one change which would be ready to review: https://review.openstack.org/#/c/481125
15:12:42 <tmorin> so pcarver, doude, matrohon, your reviews would be much welcome
15:13:05 <matrohon> tmorin: ack
15:13:27 <doude> ok
15:14:18 <tmorin> #topic tempest tests
15:14:25 <tmorin> #undo
15:14:26 <openstack> Removing item from minutes: #topic tempest tests
15:14:46 <tmorin> copy paste issue :-/
15:15:08 <tmorin> cool if you can both review matrohon and doude , thanks !
15:15:35 <matrohon> not soon but I'll do
15:15:52 <tmorin> matrohon: ok, still nice, any rough idea when ?
15:16:26 <matrohon> not before sept the 15th, sorry
15:17:15 <tmorin> matrohon: mid-sept is still fine, and the other pieces can start before this lands
15:17:17 <tmorin> next steps will be : CLI support and neutron/ovs bagpipe driver
15:17:30 <doude> tmorin: in BGPVPNDriverRCBase, the 'RC' is for 'Route Control'?
15:17:37 <tmorin> doude: yes
15:18:36 <tmorin> I thought that introducing a different base class was a nice way to allow the framework to know whether or not to advertise support for the API extension
15:18:40 <tmorin> I'm volunteering for the bagpipe driver evolutions
15:19:06 <tmorin> I was wondering if perhaps doude you would be ready to contribute the CLI support ? ...
15:19:22 <tmorin> :-D
15:19:35 <doude> :)
15:20:10 <doude> I cannot confirm now, can we re-discuss that in the next meeting?
15:20:28 <tmorin> yes, of course :)
15:20:31 <tmorin> thanks!
15:20:39 <tmorin> next topic ?
15:20:40 <doude> but should be not too hard to do now the base is there
15:21:12 <tmorin> doude: yes, although some additional things will be needed to allow updates of association resources
15:21:33 <tmorin> ok, next topic
15:21:38 <doude> yes
15:21:43 <tmorin> #topic tempest tests
15:22:05 <tmorin> a few additional tempest tests were contributed this summer by some colleagues of us
15:22:17 <tmorin> and there is more in the pipe
15:22:31 <tmorin> in particular: https://review.openstack.org/#/c/496674/
15:22:59 <tmorin> and work in progress (not yet pushed) to cover roughly the same set of tests that have been done in OPNFV
15:24:16 <tmorin> one bug was found (today) in the change merged last week: https://review.openstack.org/#/c/480142/ results in the test VM pinging itself rather than the remote VM...
15:24:22 <tmorin> this will be fixed soon ...
15:24:56 <tmorin> with the idea of improving the test case, to check that the ping fails before the BGPVPN association is made
15:25:07 <tmorin> to avoid false positives
15:25:17 <tmorin> any comments ?
15:26:08 <tmorin> doude: we had discussed about having this test suite run against OpenContrail in OpenContrail CI... is this something you have discussed internally ?
15:26:54 <doude> no
15:28:10 <tmorin> doude: is this something never done for neutron integration, or just not systematically done ?
15:29:42 <doude> Contrail CI actually run 'classic' neutron tests
15:31:19 <tmorin> it's seems there would be value in running the other test suites for service plugins for which opencontrail has a driver
15:32:12 <doude> do you want to integrate that test in the bgpvpn gates?
15:32:51 <pcarver> We're in the planning stages of a new CI for OpenContrail. Part of that could include OpenStack CI, but I don't know exactly how that would work yet.
15:33:03 <tmorin> doude: why not ? but having the suite run in opencontrail CI would work as well and would tests changes in OC driver
15:33:17 <doude> sure
15:33:32 <tmorin> doude: both would make sense in fact
15:36:25 <tmorin> ok, we can pursue the discussions in next meetings
15:36:35 <tmorin> next topic ?
15:36:52 <tmorin> #topic horizon fixes in flight
15:37:00 <tmorin> we have two horizon-related topics in flight
15:37:43 <tmorin> https://review.openstack.org/482849
15:37:48 <tmorin> and https://review.openstack.org/484791
15:38:08 <tmorin> and (three!) https://review.openstack.org/481141
15:38:12 <tmorin> they haven't landed yet
15:38:34 <tmorin> one of them is triggering a bug in a tempest job, very surprisingly
15:38:40 <tmorin> we haven't found the solution yet
15:39:09 <tmorin> https://review.openstack.org/#/c/484791 is the one   (change to use bgpvpn policy file in horizon, to decide what to expose to the user)
15:39:20 <tmorin> comments on these ?
15:39:29 <tmorin> next topic ?
15:39:52 <tmorin> #topic 5) driver framework pre commit evolution
15:40:06 <tmorin> this is about bug 1709647
15:40:08 <openstack> bug 1709647 in networking-bgpvpn "delete_net_assoc_precommit should be done before deleting bgpvpn net assoc" [Undecided,In progress] https://launchpad.net/bugs/1709647 - Assigned to Thomas Morin (tmmorin-orange)
15:40:45 <tmorin> a intern working with me on evolving bagpipe driver to use OVO noticed this issue
15:41:12 <tmorin> the problem is that we delete in the DB before the delete pre_commit hook, which prevents the driver from finding the resource in the DB
15:41:38 <tmorin> and finding the resource in the DB is needed to produce OVObjects
15:42:03 <tmorin> since this change modifies the driver framework we need to agree
15:42:51 <tmorin> the ODL driver uses the pre commit hook, but not in a way that would make it problematic to call the hook before the DB delete
15:43:07 <tmorin> I haven't checked other drivers
15:43:28 <tmorin> doude: do you know the answer for contrail driver ?  (use of delete pre_commit hook ?)
15:45:20 <doude> OC don't use hooks as it only uses the class BGPVPNDriverBase
15:45:22 <tmorin> doude: sorry, my question is dumb
15:45:29 <tmorin> doude: yes, just realized...
15:45:34 <doude> https://github.com/openstack/networking-bgpvpn/blob/master/networking_bgpvpn/neutron/services/service_drivers/driver_api.py#L23
15:45:38 <tmorin> easy answer ...
15:46:19 <tmorin> ODL seems to be fine as well: use the delete precommit hook but not in a way that depends on the DB in a particular way it seems
15:48:28 <tmorin> nuage seems to be in the same situation than contrail: https://github.com/nuagenetworks/nuage-openstack-neutron/blob/master/nuage_neutron/bgpvpn/services/service_drivers/driver.py
15:48:44 <tmorin> ok I think the decision should be easy
15:49:26 <tmorin> doude, matrohon: if you can have a look at https://review.openstack.org/#/c/499943/ and confirm that you are happy with it ...?
15:49:59 <doude> ok
15:50:10 <tmorin> thanks
15:50:29 <tmorin> #topic Old Contrail driver deprecation
15:50:46 <tmorin> doude: I'll let you talk about this one .. ?
15:51:24 <doude> no too much to say, just I have to take that one
15:51:44 <tmorin> doude: ok, agreed
15:51:53 <doude> as we already discuss in the a precedent meeting, deprecate it in Pike and remove in Queen
15:52:22 <tmorin> doude: ok, I guess you plan to add a deprecation warning and have the doc point to the new driver ?
15:52:32 <doude> yes
15:52:37 <tmorin> ok, sounds good
15:52:51 <doude> push that in master and backport it in Pike (as I'm late :( )
15:52:51 <tmorin> #topic bagpipe driver evolution
15:53:00 <tmorin> #undo
15:53:01 <openstack> Removing item from minutes: #topic bagpipe driver evolution
15:53:20 <tmorin> I'm not sure that the stable team will let us backport that
15:54:00 <tmorin> unless we consider the old driver really not good, up to the point of considering it a bug to not mark it deprecated
15:54:34 <tmorin> we can do that: if so, doude, we can fill a bug ?
15:54:59 <doude> yes, I don't remember if one was already opened
15:55:12 <tmorin> doude: I don't remember
15:55:27 * doude look into launchpad
15:55:35 <tmorin> #topic bagpipe driver evolution
15:55:48 <tmorin> just to mention that because we're late :
15:56:20 <tmorin> an intern working with me has been doing nice work to use OVO-based RPCs instead of the custom RPCs we currently use
15:56:44 <tmorin> most of the corresponding changes are in gerrit now
15:57:10 <tmorin> it will be a good basis to extend the driver, in particular to implement the routes-control blueprint
15:58:05 <tmorin> and we'll have an RPC machinery and agent extension design much closer to what exists in other places
15:58:23 <tmorin> #topic open discussion
15:58:27 <tmorin> anything to add ?
15:58:30 <tmorin> PTG perhaps
15:59:16 <tmorin> who is coming to Denver ?
15:59:40 <pcarver> I'm not going to be able to make it
15:59:44 <tmorin> looking at https://etherpad.openstack.org/p/neutron-queens-ptg only me it seems :-/
15:59:49 <tmorin> pcarver: :-(
16:00:19 <tmorin> pcarver: not much people from n8g-sfc either..
16:00:24 <doude> I'm not going to
16:00:52 <tmorin> doude: yep... :-/
16:00:56 <tmorin> wops
16:01:00 <tmorin> I see we have to leave the floor
16:01:08 <tmorin> thanks everyone
16:01:18 <doude> bye o/
16:01:22 <tmorin> I won't chair the meeting next week (flight to denver)
16:01:25 <tmorin> bye o/
16:01:34 <tmorin> perhaps matrohon will (?)
16:01:38 <tmorin> #endmeeting