| *** gianpietro has joined #openstack-kuryr | 00:19 | |
| *** gianpietro has quit IRC | 00:52 | |
| *** gianpietro has joined #openstack-kuryr | 01:16 | |
| *** phuoc has joined #openstack-kuryr | 01:41 | |
| *** gianpietro has quit IRC | 01:50 | |
| *** premsankar has quit IRC | 02:58 | |
| *** hongbin has joined #openstack-kuryr | 02:58 | |
| *** janki has joined #openstack-kuryr | 04:30 | |
| *** hongbin has quit IRC | 05:07 | |
| *** yboaron_ has joined #openstack-kuryr | 05:32 | |
| *** phuoc_ has joined #openstack-kuryr | 05:32 | |
| *** phuoc has quit IRC | 05:33 | |
| *** links has joined #openstack-kuryr | 05:56 | |
| *** irenab has quit IRC | 07:00 | |
| *** gkadam has joined #openstack-kuryr | 07:02 | |
| *** ccamposr has joined #openstack-kuryr | 07:07 | |
| *** ccamposr__ has joined #openstack-kuryr | 07:08 | |
| *** ccamposr has quit IRC | 07:12 | |
| *** pcaruana has joined #openstack-kuryr | 07:20 | |
| *** pcaruana has quit IRC | 07:34 | |
| *** pcaruana has joined #openstack-kuryr | 07:40 | |
| *** snapiri has joined #openstack-kuryr | 07:45 | |
| *** maysams has joined #openstack-kuryr | 07:47 | |
| *** aperevalov has joined #openstack-kuryr | 07:49 | |
| *** celebdor has joined #openstack-kuryr | 08:29 | |
| *** yboaron_ has quit IRC | 08:51 | |
| *** yboaron_ has joined #openstack-kuryr | 08:52 | |
| *** links has quit IRC | 09:26 | |
| *** yboaron_ has quit IRC | 09:27 | |
| *** garyloug has joined #openstack-kuryr | 09:27 | |
| *** yboaron_ has joined #openstack-kuryr | 09:28 | |
| *** snapiri has quit IRC | 09:36 | |
| *** snapiri has joined #openstack-kuryr | 09:45 | |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-tempest-plugin master: Enhance svc namespace isolation with LoadBalancer type svc https://review.openstack.org/618751 | 09:58 |
|---|---|---|
| celebdor | yboaron_: where is the select? | 10:14 |
| yboaron_ | celebdor, https://github.com/kubernetes-client/python-base/blob/master/stream/ws_client.py#L169 | 10:15 |
| celebdor | thanks | 10:15 |
| yboaron_ | celebdor, the run_forever and read_stdout and read_stderr | 10:17 |
| celebdor | ok | 10:17 |
| yboaron_ | celebdor, call to this update function | 10:17 |
| celebdor | I was just thinking if the fact that we do it from several threads at the same time could be the issue | 10:17 |
| openstackgerrit | Yossi Boaron proposed openstack/kuryr-tempest-plugin master: Add timeout to calls for ws_client.py functions https://review.openstack.org/618982 | 10:22 |
| openstackgerrit | Genadi Chereshnya proposed openstack/kuryr-tempest-plugin master: Check the number of neutron ports after controller restart https://review.openstack.org/618650 | 10:24 |
| dulek | celebdor: Hey, quick question about https://review.openstack.org/#/c/617552/. | 10:26 |
| dulek | celebdor: Which is - why do we care? :D | 10:26 |
| celebdor | xD | 10:27 |
| celebdor | well, I don't think that anybody would use a controller withough https, so better to have it as default in the gates | 10:27 |
| dulek | celebdor: At the moment all administrative stuff we do in the plugin is through that endpoint that doesn't require any auth (localhost:8080). | 10:29 |
| dulek | celebdor: Then the endpoint is ignored. I don't see that as a serious problem. If that patch proves to be hard to get right, I don't think we need to invest a lot of time here. | 10:30 |
| celebdor | I'll give it one more try | 10:30 |
| celebdor | also, I moved it closer to the upstream docs regarding the cert generation | 10:31 |
| celebdor | since the make-cert.sh is not on the k8s repo for a very long time | 10:31 |
| dulek | celebdor: Okay, but it only dies due to some code related to hyperkube being wrong. So it seems not a big deal to fix it. | 10:31 |
| dulek | celebdor: Yeah, since 1.6 IIRC. ;) | 10:31 |
| celebdor | I thought it was 1.8 | 10:31 |
| celebdor | but yes, ancient | 10:31 |
| *** gianpietro has joined #openstack-kuryr | 10:37 | |
| celebdor | maysams: ltomasbo: when the quota is infinite, isn't it -1? https://review.openstack.org/#/c/618655/4/kuryr_kubernetes/controller/handlers/namespace.py | 10:44 |
| celebdor | if that is the case, won't this check be wrong? | 10:44 |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: devstack: use https to access the k8s cluster https://review.openstack.org/617552 | 10:52 |
| ltomasbo | celebdor, checking... | 10:52 |
| ltomasbo | celebdor, yep, you are right, if -1 we should just return True | 10:53 |
| ltomasbo | maysams, ^^ | 10:53 |
| celebdor | I'm catching details lately | 10:54 |
| celebdor | celebdor++ | 10:54 |
| celebdor | :-) | 10:54 |
| ltomasbo | :) | 10:54 |
| celebdor | ltomasbo: regarding checking quota very often with neutron... | 10:56 |
| celebdor | maybe we should just cache it | 10:57 |
| ltomasbo | yep, catching it for 1 min or something should work... | 10:58 |
| ltomasbo | celebdor, my point was that calling neutron (several times) every 3-5 seconds... you know... | 10:59 |
| maysams | you're right celebdor | 10:59 |
| maysams | thanks for pointing that out | 10:59 |
| celebdor | you're most welcome | 11:00 |
| celebdor | ltomasbo: I wish they would just implement https://github.com/kubernetes/kubernetes/issues/37218 | 11:00 |
| celebdor | so that we could check different things at different intervals | 11:00 |
| celebdor | otherwise we need to put the different intervals logic in our health manager | 11:00 |
| celebdor | which is also an option that may be better than caching | 11:00 |
| ltomasbo | yep | 11:01 |
| maysams | would like this too ltomasbo | 11:01 |
| maysams | but this is for liveness probe, right now we would need for readiness | 11:02 |
| celebdor | readiness is fine to check often, since it is only checked at the beginning, isn't it maysams ? | 11:03 |
| maysams | celebdor: what do mean with the beginning? the readiness probe is performed in the same frenquency as the liveness, I believe | 11:05 |
| celebdor | I thought it's only done until it is found ready | 11:08 |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: devstack: use https to access the k8s cluster https://review.openstack.org/617552 | 11:10 |
| celebdor | but I'm probably wrong | 11:13 |
| maysams | celebdor: the readiness probe is constantly performed | 11:13 |
| celebdor | ok | 11:14 |
| maysams | just as the liveness | 11:14 |
| celebdor | :-) | 11:14 |
| maysams | :) | 11:14 |
| celebdor | so caching everywhere | 11:14 |
| celebdor | or just have our health manager be able to check different things at different intervals | 11:14 |
| *** tosky has joined #openstack-kuryr | 11:16 | |
| dmellado | dulek: tosky o/ | 11:17 |
| dulek | tosky, dmellado: Okay, so my concern is that if we import octavia-tempest-plugin, then we need to enforce stable interface of that client. | 11:17 |
| dmellado | dulek: should totally be there | 11:17 |
| dulek | This means running kuryr jobs on octavia-tempest-plugin, which makes Octavia people less happy. | 11:17 |
| dmellado | it was enforced by gmann and the folks last cycle IIRC | 11:17 |
| dulek | dmellado: Oh, really? | 11:17 |
| dmellado | gmann: could you confirm that? | 11:17 |
| dulek | Okay, so I'm probably simply misinformed then. | 11:18 |
| dmellado | nir will be joining soon too | 11:18 |
| *** nmagnezi has joined #openstack-kuryr | 11:18 | |
| celebdor | openstacksdk, openstacksdk, openstacksdk | 11:18 |
| celebdor | openstacksdk | 11:18 |
| celebdor | openstacksdk | 11:18 |
| tosky | dulek: if octavia-tempest-plugin uses the stable API for plugins, and kuryr-tempest-plugin does the same, the client should be magically available | 11:18 |
| celebdor | openstacksdk | 11:18 |
| tosky | openstacksdk? Is it a different discussion? | 11:19 |
| dulek | tosky: celebdor wanted to rewrite our tempest plugin to use openstacksdk. | 11:19 |
| tosky | no | 11:19 |
| maysams | celebdor: okay, thanks | 11:19 |
| tosky | sorry for the short answer, but it's simply no | 11:19 |
| dmellado | no, that's why we have the stable API | 11:20 |
| dmellado | no openstacksdk | 11:20 |
| dmellado | that'd be a terrible hack | 11:20 |
| tosky | there is a reason for tempest client to be independent | 11:20 |
| dmellado | not to rely on the client itself, basically | 11:20 |
| dulek | tosky: Sure, but if we don't test it, then it'll break pretty often, subtle changes are casually ignored by human reviewers. So in that case we need kuryr jobs running on octavia-tempest-plugin. | 11:20 |
| dmellado | so it can be tested | 11:20 |
| dmellado | dulek: so we have some options | 11:20 |
| dmellado | 1) we add these two to the gate, make sure they do use stable API | 11:20 |
| dmellado | and just live with that | 11:20 |
| tosky | dulek: you just need to gate changes to service clients, not to the tests | 11:20 |
| tosky | so a specific path | 11:21 |
| tosky | and not something happening frequently - once written, service clients are usually complete | 11:21 |
| dulek | tosky: Hm. Yeah, that makes sense to me, haven't came up with that idea. | 11:21 |
| dmellado | 2) import the plugin directly | 11:21 |
| dulek | tosky: So it's a bit like having that code in library, just it sits in your repo. It's fine with me. | 11:21 |
| dmellado | dulek: if you recall gmann email re: congress | 11:21 |
| dmellado | that's what they suggest | 11:21 |
| tosky | it's not my repo, but that's fine :) | 11:22 |
| dulek | tosky: Sure, sure. ;) | 11:22 |
| tosky | I'm working on a different project, but I've been following the evolution of tempest plugins since its inception, hence my suggestions | 11:22 |
| dulek | Wait, moment. | 11:22 |
| celebdor | dmellado: why is it a terrible hack? | 11:23 |
| dulek | So we still need an "okay" from Octavia to add those Kuryr gates. :P | 11:23 |
| celebdor | I don't understand why the tests can't just use a normal client to reach openstack projects | 11:23 |
| dulek | celebdor: That's because tempest should test API, not the clients. Client may hide errors. | 11:23 |
| dulek | celebdor: Bugs I mean. | 11:23 |
| gmann | dulek: dmellado commented on patch | 11:23 |
| celebdor | dulek: openstacksdk is giving you a requests sessiont | 11:24 |
| dulek | nmagnezi: Is there "okay" for adding a kuryr-kubernetes job to octavia-tempest-plugin that will be run each time client stuff is changed? | 11:24 |
| celebdor | *session | 11:24 |
| dmellado | celebdor: because that'd mean relying on the same stuff you'd try to test | 11:24 |
| celebdor | that's what I meant to use | 11:24 |
| dmellado | dulek: that shouldn't be ran too often at all | 11:24 |
| dmellado | after all it'd be using stable api | 11:24 |
| dmellado | so I'm not sure it'd be even needed | 11:24 |
| dulek | dmellado: Stable API's are broken every week, don't remember os-vif, recent python-neutronclient, or whatever? | 11:25 |
| nmagnezi | dulek, that is a kind of question you should raise in a weekly meeting that we have (which happens tomorrow) | 11:25 |
| gmann | dulek: dmellado overall, octavia-tempest-plugin should provide stable interface and make it available in g-r. user does not need to add their jobs on dependent plugins side but can trust on dependent plugin for maintaining stable interface as stable. but using tag will not cause any failure on user plugin | 11:25 |
| dulek | gmann: Well, I find lack of job against "if it's not tested it's broken". We have had hard life until we've added our jobs on os-vif. | 11:26 |
| gmann | dulek: true which makes those dependent pckg does not maintain the stable interface in compatible way. but here we are using tag for each plugins. so as long as you (or g-r) update the tag it will show you that interface changed or not otherwise it will run good. | 11:28 |
| openstackgerrit | Merged openstack/kuryr-kubernetes master: Add missing key check while lbaas member add https://review.openstack.org/602017 | 11:28 |
| dulek | gmann: Oh, we're using releases. That's a bit better, we can lock on older version then. | 11:28 |
| gmann | dulek: i am not against of running cross CI that is much better. if octavia-tempest-plugins team is ok then it is fine | 11:28 |
| gmann | dulek: yes, that was main reason for me to introdice the release/tag for plugins | 11:29 |
| tosky | dulek: but please consider that locking releases can be problematic for downstream consumers; tempest plugins are branchless like tempest, should follow the same rules | 11:29 |
| dulek | gmann: Okay, that makes sense to me then, no need for a CI job, at least immediately. ;) | 11:29 |
| gmann | dulek: let me make 1. ocatvia-tempest-plugin provide the stable interface 2. add octavia-tempest-plugin in g-r and then you can have look how it looks like | 11:30 |
| dulek | tosky: Sure, but at least we can test and fix stuff before bumping a release. It's way worse if lib release breaks us for a week. | 11:30 |
| dulek | gmann: It works for me, thanks! | 11:31 |
| dulek | Okay, so I guess we kind of agreed on that? It wasn't too bad. :D | 11:31 |
| tosky | dulek: I'd say: keep it open, and constraint if something breaks; I'd still maintain that it's less likely to break (and constraint) that having to actively bump the constraint | 11:31 |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: devstack: use https to access the k8s cluster https://review.openstack.org/617552 | 11:33 |
| dmellado | +1 | 11:40 |
| dmellado | crysis settled xD | 11:40 |
| dmellado | now if we'd only have also gena on the channel... | 11:40 |
| dulek | dmellado: Sent him link to the discussion. | 11:44 |
| dmellado | dulek: thanks! | 11:44 |
| *** gkadam has quit IRC | 11:56 | |
| *** janki has quit IRC | 12:10 | |
| *** gianpietro has quit IRC | 12:13 | |
| *** gianpietro has joined #openstack-kuryr | 12:13 | |
| *** gkadam has joined #openstack-kuryr | 12:14 | |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: devstack: use https to access the k8s cluster https://review.openstack.org/617552 | 12:16 |
| *** gianpietro has quit IRC | 12:18 | |
| openstackgerrit | Daniel Mellado proposed openstack/kuryr-kubernetes master: Rename kuryr-status to kuryr-k8s-status https://review.openstack.org/618996 | 12:19 |
| *** gianpietro has joined #openstack-kuryr | 12:29 | |
| *** gianpietro has quit IRC | 12:33 | |
| *** gianpietro has joined #openstack-kuryr | 12:34 | |
| *** gianpietro has quit IRC | 12:35 | |
| *** gianpietro has joined #openstack-kuryr | 12:35 | |
| *** gianpietro has quit IRC | 12:39 | |
| openstackgerrit | Daniel Mellado proposed openstack/kuryr-kubernetes master: Rename kuryr-status to kuryr-k8s-status https://review.openstack.org/618996 | 12:46 |
| *** rh-jelabarre has joined #openstack-kuryr | 12:55 | |
| *** celebdor has quit IRC | 13:03 | |
| *** gianpietro has joined #openstack-kuryr | 13:17 | |
| *** celebdor has joined #openstack-kuryr | 13:36 | |
| *** celebdor has quit IRC | 13:40 | |
| *** celebdor has joined #openstack-kuryr | 13:40 | |
| *** gianpietro has quit IRC | 13:45 | |
| openstackgerrit | Genadi Chereshnya proposed openstack/kuryr-tempest-plugin master: Moving delete_namespace to addCleanup in port_pool testing https://review.openstack.org/619000 | 13:46 |
| *** gianpietro has joined #openstack-kuryr | 14:05 | |
| *** yboaron_ has quit IRC | 14:05 | |
| *** yboaron_ has joined #openstack-kuryr | 14:06 | |
| openstackgerrit | Daniel Mellado proposed openstack/kuryr-tempest-plugin master: Add network policy security group rules tests https://review.openstack.org/603618 | 14:13 |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: devstack: use https to access the k8s cluster https://review.openstack.org/617552 | 14:13 |
| *** kiennt26 has joined #openstack-kuryr | 14:26 | |
| openstackgerrit | Daniel Mellado proposed openstack/kuryr-kubernetes master: Rename kuryr-status to kuryr-k8s-status https://review.openstack.org/618996 | 14:29 |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: devstack: use https to access the k8s cluster https://review.openstack.org/617552 | 14:45 |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Ensure pod relabeling is supported by the Network Policy https://review.openstack.org/618521 | 14:50 |
| *** kiennt26 has left #openstack-kuryr | 14:58 | |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: devstack: use https to access the k8s cluster https://review.openstack.org/617552 | 15:08 |
| celebdor | dulek: I think ^^ solves the issues | 15:09 |
| celebdor | at least it does on my env | 15:09 |
| dulek | celebdor: That's something. | 15:09 |
| celebdor | (although it's not a pristine environment) | 15:09 |
| celebdor | dulek: now I can go back to the misery of trying to make openshift 3.11, with all its changes, work on devstack | 15:10 |
| dulek | celebdor: What can I do… Wish you luck! | 15:11 |
| *** yboaron_ has quit IRC | 15:11 | |
| celebdor | xD | 15:11 |
| celebdor | dulek: I'll split it into smaller patches | 15:11 |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: devstack: parametrize openshift version fetching https://review.openstack.org/619018 | 15:18 |
| celebdor | dulek: of course, if you volunteer... I will not refuse you | 15:18 |
| celebdor | xD | 15:18 |
| dulek | celebdor: Ha, just got with catching-up routine to the Trello board, so will see how's my backlog. ;) | 15:18 |
| celebdor | ;-) | 15:19 |
| dmellado | trello notifications have been terrible, celebdor | 15:23 |
| dmellado | I'll make zuul send you emails for every kuryr* patch xD | 15:23 |
| celebdor | dmellado: don't make me write a bot so the emails come from aliases instead of trello if you happen to filter them | 15:24 |
| dmellado | lol, I'd do something even more evil and you know | 15:26 |
| dmellado | btw ltomasbo, ping | 15:26 |
| ltomasbo | dmellado, pong | 15:26 |
| ltomasbo | (in a meeting btw...) | 15:26 |
| dmellado | ltomasbo: I've just gotten a bummer regarding our octavia patch | 15:27 |
| dmellado | it doesn't look like it'll get accepted so we'd have to think about ugly workarounds | 15:27 |
| *** maysams has quit IRC | 16:02 | |
| openstackgerrit | Michał Dulko proposed openstack/kuryr-kubernetes master: Work out situation with KUBERNETES_NODE_NAME https://review.openstack.org/619032 | 16:26 |
| *** gkadam has quit IRC | 16:50 | |
| *** ccamposr__ has quit IRC | 16:51 | |
| *** premsankar has joined #openstack-kuryr | 16:51 | |
| *** garyloug has quit IRC | 17:43 | |
| *** celebdor1 has joined #openstack-kuryr | 18:46 | |
| *** pcaruana has quit IRC | 18:47 | |
| *** celebdor has quit IRC | 18:47 | |
| *** gianpietro has quit IRC | 19:08 | |
| *** gianpietro has joined #openstack-kuryr | 19:23 | |
| *** gianpietro has quit IRC | 19:32 | |
| *** gianpietro has joined #openstack-kuryr | 19:33 | |
| *** gianpietro has quit IRC | 19:37 | |
| *** gianpietro has joined #openstack-kuryr | 19:41 | |
| openstackgerrit | Maysa de Macedo Souza proposed openstack/kuryr-kubernetes master: Add quota readiness check to controller https://review.openstack.org/618655 | 20:46 |
| *** gianpietro has quit IRC | 21:25 | |
| *** rh-jelabarre has quit IRC | 21:27 | |
| *** rh-jelabarre has joined #openstack-kuryr | 21:28 | |
| *** gianpietro has joined #openstack-kuryr | 22:14 | |
| *** gianpietro has quit IRC | 23:32 | |
| openstackgerrit | Maysa de Macedo Souza proposed openstack/kuryr-kubernetes master: Add quota readiness check to controller https://review.openstack.org/618655 | 23:37 |
| openstackgerrit | Merged openstack/kuryr-tempest-plugin master: Add timeout to calls for ws_client.py functions https://review.openstack.org/618982 | 23:45 |
| *** rh-jelabarre has quit IRC | 23:52 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!