| *** matrohon has joined #openstack-net-bgpvpn | 08:07 | |
| *** tmorin has joined #openstack-net-bgpvpn | 08:12 | |
| *** enikher has joined #openstack-net-bgpvpn | 08:13 | |
| doude | Hi tmorin ! | 08:27 |
|---|---|---|
| tmorin | hi doude ! | 08:31 |
| doude | tmorin: finally, I have to remove the bagpipe driver for the Juno backport | 08:31 |
| doude | the ovs agent used the callback mechanism of Neutron that appeard in release Kilo | 08:32 |
| doude | s/appeard/appeared | 08:33 |
| tmorin | do you mean: the bagpipe driver uses de neutron registry callback ? | 08:34 |
| tmorin | (it's not the ovs agent directly) | 08:34 |
| doude | I mean the ovs agent in the bagpipe service driver | 08:35 |
| tmorin | I'm not sure I understand then: which callbacks ? | 08:35 |
| doude | tmorin: oops, yes sorry, not the bagpipe ovs agent but the bagpipe driver | 08:36 |
| doude | https://github.com/openstack/networking-bgpvpn/blob/backport/juno/networking_bgpvpn/neutron/services/service_drivers/bagpipe/bagpipe.py#L22 | 08:36 |
| tmorin | ok | 08:36 |
| tmorin | yes | 08:37 |
| tmorin | in fact, even for Kilo, the registry callback was not usable for bagpipe bgpvpn driver, because the notification we needed were not implemented | 08:37 |
| doude | ha ok | 08:38 |
| tmorin | so for Kilo we use an ML2 mech_driver to feed the right events into bagpipe driver | 08:38 |
| doude | but in the case of Juni the pep8 tests fail | 08:38 |
| tmorin | this is what https://review.openstack.org/#/c/257474/ does | 08:38 |
| tmorin | with https://review.openstack.org/#/c/257474/ the bagpipe driver does not use the registry callback at all | 08:39 |
| doude | ok I missed that fix | 08:39 |
| doude | so normally I just have to backport it | 08:39 |
| doude | no, I've it with the rebase kilo on juno | 08:41 |
| doude | but https://review.openstack.org/#/c/257474/ still uses the callback mechanism ? | 08:42 |
| doude | https://github.com/openstack/networking-bgpvpn/blob/backport/kilo/networking_bgpvpn/neutron/services/service_drivers/bagpipe/ml2_mech_driver.py#L18 | 08:42 |
| tmorin | sorry, afk for a few minutes | 09:01 |
| tmorin | no, in this kilo version, the bagpipe driver does not use the callback registry | 09:02 |
| tmorin | the function function in the driver are the same than the ones used as callbacks for the liberty version | 09:03 |
| tmorin | but the driver does not try to import the callback registry to register them | 09:03 |
| tmorin | they will be called by the code in the bgpvpn_notify mech_driver | 09:03 |
| tmorin | ah | 09:04 |
| tmorin | I see what you mean | 09:04 |
| tmorin | to keep the function calls as close as possible the mech_driver imports a few constants from the registry callbacks | 09:04 |
| tmorin | e.g. https://github.com/openstack/networking-bgpvpn/blob/backport/kilo/networking_bgpvpn/neutron/services/service_drivers/bagpipe/ml2_mech_driver.py#L55 | 09:05 |
| tmorin | but this is not necessary in fact | 09:05 |
| tmorin | lines 55,56,68,69,80 and 81 can be replaced by "None," | 09:05 |
| tmorin | these parameters are not used by the callbacks in the driver | 09:06 |
| *** vthapar has joined #openstack-net-bgpvpn | 09:10 | |
| doude | tmorin: ok, so do you want I try to remove that callback dependance by replacing parameter by none ? | 09:48 |
| tmorin | yes, I think that would be a simple right solution | 10:24 |
| tmorin | another is to redefine the constants we use from the callback registry in ml2_mech_driver.py | 10:25 |
| doude | ok | 10:31 |
| doude | tmorin: re | 10:59 |
| tmorin | yep ? | 10:59 |
| doude | that works for bagpipe | 10:59 |
| doude | but now I'm facing lot of issues with UT | 10:59 |
| tmorin | what kind ? | 10:59 |
| doude | which depends on Neutron UT framework that did not exist in juno | 11:00 |
| doude | like: neutron.tests.unit.db.test_db_base_plugin_v2 | 11:00 |
| tmorin | I see | 11:00 |
| doude | or: neutron.tests.unit.api.v2.test_base | 11:00 |
| tmorin | I was kindof expecting this kind of issue | 11:00 |
| tmorin | hence my comment yesterday | 11:01 |
| doude | yes | 11:01 |
| tmorin | I have to go for lunch | 11:01 |
| doude | you're right | 11:01 |
| doude | ok | 11:01 |
| tmorin | right now I don't have any better idea than "let's accept having less UT for this backport" | 11:01 |
| doude | all right I do that | 11:01 |
| tmorin | because a backport is probably not worth investing time in UTs | 11:02 |
| doude | true | 11:02 |
| tmorin | let's take this option | 11:02 |
| tmorin | I have to go | 11:02 |
| tmorin | see you later today | 11:02 |
| doude | enjoy your meal tmorin | 11:03 |
| *** vthapar has quit IRC | 11:39 | |
| *** vthapar has joined #openstack-net-bgpvpn | 11:44 | |
| doude | tmorin: re | 15:09 |
| doude | I've got issue to push all commit in review on the backport/juno branch | 15:10 |
| tmorin | yep | 15:10 |
| tmorin | which issue ? | 15:10 |
| doude | when I try to push reviews, git-review wants to push all commits of bgpvpn in review | 15:14 |
| doude | it does not start after the last commit of the backport/juno branch | 15:16 |
| tmorin | then you probably should try to do one "git review" for each commit | 15:16 |
| doude | arf | 15:16 |
| tmorin | assuming you have commits c1,c2,c3,c4, I think you should do: | 15:16 |
| tmorin | git checkout -b juno_backport (from current backport/juno HEAD) | 15:17 |
| tmorin | git cherry-pick c1 | 15:17 |
| tmorin | git review backport/juno | 15:17 |
| tmorin | git cherry-pick c2 | 15:17 |
| doude | I remark the branch backport/juno branch on the repo review.openstack.org | 15:17 |
| tmorin | git review backport/juno | 15:17 |
| tmorin | etc. | 15:17 |
| doude | if I do 'git review backport/juno' | 15:18 |
| doude | I get 'The branch 'backport/juno' does not exist on the given remote' | 15:18 |
| tmorin | hmm surprising | 15:19 |
| tmorin | what does "git branch -a" say ? | 15:19 |
| tmorin | did you do a fresh "git fetch gerrit" ? | 15:19 |
| tmorin | assuming gerrit is the name of your remote for review.openstack.org | 15:20 |
| doude | tmorin: oops, I just released my remote gerrit pointed to neutron repo :( | 15:22 |
| doude | sorry for the noise | 15:22 |
| tmorin | ah, ok | 15:22 |
| tmorin | all good then :) | 15:22 |
| tmorin | np | 15:22 |
| doude | tmorin: sorry, seems good now | 15:24 |
| doude | I'll pussh 23 review | 15:24 |
| tmorin | :D | 15:25 |
| tmorin | doude: about https://bugs.launchpad.net/bgpvpn/+bug/1531905 | 16:18 |
| openstack | Launchpad bug 1531905 in bgpvpn "juno: Cannot load the BaGPipe driver" [Undecided,New] | 16:18 |
| tmorin | is this bug here simply to explain the additional change you will had to ml2_mech_driver.py to use None instead of the constants from the callback registry module ? | 16:19 |
| doude | failyes | 16:38 |
| doude | yes | 16:38 |
| *** matrohon has quit IRC | 17:45 | |
| *** vthapar has quit IRC | 17:54 | |
| *** enikher has left #openstack-net-bgpvpn | 20:03 | |
| *** openstackgerrit has quit IRC | 20:23 | |
| *** openstackgerrit has joined #openstack-net-bgpvpn | 20:24 | |
| *** matrohon has joined #openstack-net-bgpvpn | 20:39 | |
| *** matrohon has quit IRC | 20:58 | |
| *** matrohon has joined #openstack-net-bgpvpn | 21:03 | |
| *** matrohon has quit IRC | 21:37 | |
| *** matrohon has joined #openstack-net-bgpvpn | 21:38 | |
| *** matrohon has quit IRC | 22:12 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!