| *** tzumainn has quit IRC | 00:16 | |
| *** hongbin has joined #openstack-kuryr | 00:22 | |
| *** celebdor1 has quit IRC | 01:17 | |
| openstackgerrit | Peng Liu proposed openstack/kuryr-kubernetes master: Implement NPWG multi-vif driver https://review.openstack.org/578009 | 01:33 |
|---|---|---|
| openstackgerrit | Peng Liu proposed openstack/kuryr-kubernetes master: Implement NPWG multi-vif driver https://review.openstack.org/578009 | 01:34 |
| *** hongbin has quit IRC | 03:46 | |
| openstackgerrit | Peng Liu proposed openstack/kuryr-kubernetes master: Implement NPWG multi-vif driver https://review.openstack.org/578009 | 04:54 |
| *** janki has joined #openstack-kuryr | 04:56 | |
| *** kailun has joined #openstack-kuryr | 05:51 | |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Ensure OpenShift gate uses the namespace subnet/sg drivers https://review.openstack.org/580680 | 06:24 |
| *** phuoc has joined #openstack-kuryr | 06:40 | |
| *** phuoc_ has quit IRC | 06:40 | |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Ensure no api_root value for containerized deployments https://review.openstack.org/588153 | 07:03 |
| *** snapiri has joined #openstack-kuryr | 07:30 | |
| *** snapiri has quit IRC | 07:32 | |
| *** snapiri has joined #openstack-kuryr | 07:32 | |
| openstackgerrit | Michał Dulko proposed openstack/kuryr-kubernetes master: Log traceback on errors in Watcher https://review.openstack.org/587815 | 07:34 |
| *** pmannidi has joined #openstack-kuryr | 07:36 | |
| *** celebdor has joined #openstack-kuryr | 07:37 | |
| celebdor | ltomasbo: dulek: morning | 07:38 |
| ltomasbo | celebdor, good morning! | 07:38 |
| celebdor | ltomasbo: did you see I got the kuryr/demo from the tempest repo to be used instead of the one from dockerhub? | 07:39 |
| ltomasbo | I reviewed both ps! | 07:39 |
| ltomasbo | nice! | 07:39 |
| ltomasbo | celebdor, I created a bug and a fix for yesterday's issue: | 07:40 |
| ltomasbo | https://bugs.launchpad.net/kuryr-kubernetes/+bug/1785010 | 07:40 |
| openstack | Launchpad bug 1785010 in kuryr-kubernetes "kuryr-config ConfigMap malformed on containerized deployments" [Undecided,In progress] - Assigned to Luis Tomas Bolivar (ltomasbo) | 07:40 |
| ltomasbo | https://review.openstack.org/#/c/588153/1 | 07:40 |
| celebdor | cool | 07:40 |
| ltomasbo | I'm deploying it to check if it actually works! xD | 07:40 |
| celebdor | irenab: https://review.openstack.org/#/c/588153/1 | 07:41 |
| ltomasbo | celebdor, btw, I see you +W this one and remove it during my vacations: https://review.openstack.org/#/c/580678 | 07:42 |
| ltomasbo | I rebase it and add a couple of modifications in case you want to take a look | 07:42 |
| ltomasbo | then the service isolation is also waiting for you xD https://review.openstack.org/#/c/581421/ | 07:43 |
| ltomasbo | ohh, now the ConfigMap is properly configured, but kuryr-controller crashes... going to check why | 07:44 |
| dulek | ltomasbo: Oh my, how did you realized it's the api_root value? | 07:44 |
| ltomasbo | the value was empty on the generated file | 07:44 |
| *** kailun has quit IRC | 07:45 | |
| ltomasbo | I simply remove that line and regenerate the CM, and saw that the render was ok then | 07:45 |
| dulek | ltomasbo: Nice! | 07:45 |
| ltomasbo | it happened to my some time in the pass that I generated a bad configmap and it rendered like that | 07:45 |
| dulek | ltomasbo: Actually it was the value I needed to edit when I've noticed the issue. :D | 07:45 |
| ltomasbo | so I thought something had to be bad on that side | 07:45 |
| ltomasbo | xD | 07:46 |
| dulek | ltomasbo: Okay, though I'm not 100% sure your patch will not break anything. Need to check. | 07:46 |
| ltomasbo | dulek, yep, I'm not sure either... | 07:46 |
| ltomasbo | stack completed, and in theory that is not needed for the containerized, right? | 07:46 |
| *** leifmadsen_ has quit IRC | 07:47 | |
| dulek | ltomasbo: Yes, yes. I'm just thinking that I was explicitly erasing it for a reason. :D | 07:47 |
| ltomasbo | ok, another option is to set it to really set it to "" | 07:48 |
| ltomasbo | api_root: "" | 07:48 |
| dulek | ltomasbo: Seems like it isn't set anywhere else. Let's see how gates reacts. | 07:48 |
| ltomasbo | instead of | 07:48 |
| ltomasbo | api_root: | 07:48 |
| dulek | ltomasbo: Sure. But let's check the gate. :) | 07:48 |
| ltomasbo | yep | 07:48 |
| ltomasbo | what I'm not sure is why we were not hitting this before, that patch was merged some time ago... | 07:49 |
| ltomasbo | right? | 07:49 |
| dulek | ltomasbo: Right. I think I know why I've set it. | 07:49 |
| ltomasbo | tell me! xD | 07:49 |
| dulek | ltomasbo: https://github.com/openstack/kuryr-kubernetes/blob/5421ce1ba53962cdfe157461b2e696f735498693/kuryr_kubernetes/config.py#L85-L87 | 07:49 |
| dulek | ltomasbo: This will default to localhost and actually will work for K8s. | 07:50 |
| dulek | ltomasbo: But e.g. OpenShift is deployed on different port. | 07:50 |
| *** leifmadsen has joined #openstack-kuryr | 07:50 | |
| ltomasbo | but is that used for containerized? | 07:50 |
| ltomasbo | or the value is taken from the env vars of the container? | 07:50 |
| dulek | ltomasbo: https://github.com/openstack/kuryr-kubernetes/blob/ce3305b9eff3eff736065c9b9a5921bfd593b375/kuryr_kubernetes/clients.py#L72-L81 | 07:53 |
| ltomasbo | yep, that is what I saw yesterday | 07:53 |
| ltomasbo | and I checked that the vars were actually on the containers | 07:53 |
| dulek | ltomasbo: Basically you can override it in any case. But to be taken from env vars you need to empty the api_root. | 07:54 |
| dulek | To be honest it shouldn't default to localhost:8080 in the first place, but I don't think it's sane to change the default now. | 07:54 |
| celebdor | dulek: why not? | 07:55 |
| ltomasbo | so, you think we should find the way to make the configmap.yaml to render with 'api_root: ""' instead, right? | 07:55 |
| dulek | ltomasbo: I think so. `iniset $KURYR_CONFIG kubernetes api_root '""' should do the job. | 07:55 |
| ltomasbo | let me try that! | 07:55 |
| dulek | celebdor: This is a setting that is pretty important so operators should be forced to set it explicitly. | 07:56 |
| dulek | celebdor: Or you're asking about changing the default? :D | 07:57 |
| irenab | celebdor, asap | 07:57 |
| celebdor | dulek: the default | 07:59 |
| dulek | celebdor: Well, it's shipped, people using upstream release may depend on that default. | 07:59 |
| ltomasbo | ok, celebdor, dulek: I'll wait for the gates, and then re-spin with dulek's suggestion | 08:01 |
| ltomasbo | to force it to '""' | 08:01 |
| celebdor | Ok. I would be surprised if the default value is useful to anyone, being that it is a non https port | 08:02 |
| dulek | celebdor: Yeah, me too. I'll check if it's possible to deprecate the default value somehow. | 08:05 |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Ensure no api_root value for containerized deployments https://review.openstack.org/588153 | 08:06 |
| *** pmannidi has quit IRC | 08:28 | |
| ltomasbo | dulek, celebdor: I tested this: https://review.openstack.org/#/c/588153 | 08:43 |
| ltomasbo | it fixes the issue, and sets api_root to "" | 08:43 |
| celebdor | ok, then | 08:43 |
| celebdor | let's take it in | 08:44 |
| celebdor | ltomasbo: you have merge rights on https://review.openstack.org/#/c/587746/ , right? | 08:54 |
| ltomasbo | yep! | 08:54 |
| celebdor | dulek: and you on https://review.openstack.org/#/c/588074/ , right? | 08:54 |
| celebdor | :P | 08:54 |
| celebdor | so come on | 08:55 |
| celebdor | :P | 08:55 |
| ltomasbo | xD | 08:55 |
| ltomasbo | done! | 08:55 |
| celebdor | thanks! | 08:55 |
| dulek | celebdor: https://review.openstack.org/#/c/585706 | 08:56 |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: devstack: deploy openshift registry on origin envs https://review.openstack.org/580201 | 08:56 |
| celebdor | dulek: oh, about putting the bug link | 08:57 |
| dulek | celebdor: More like - which one to merge? itzikb was first. :) | 08:58 |
| celebdor | dulek: the problem is that one has not been tested in gates, has it? | 08:59 |
| celebdor | and it's in test-config | 08:59 |
| celebdor | it should be in extra | 08:59 |
| celebdor | otherwise it's a race with the docker configure step | 08:59 |
| celebdor | in the devstack plugin container | 08:59 |
| dulek | celebdor: test-config runs after extra. | 09:00 |
| *** pcaruana has joined #openstack-kuryr | 09:00 | |
| celebdor | dulek: you sure about that? | 09:01 |
| celebdor | well, if that's the case, that's bad | 09:02 |
| dulek | celebdor: https://docs.openstack.org/devstack/latest/configuration.html, but order is not mentioned, so lemme find it… | 09:02 |
| celebdor | dulek: sorry, confusion of today | 09:03 |
| celebdor | not of yesterday | 09:03 |
| celebdor | it must run after post-config | 09:03 |
| celebdor | so the immediate next is extra | 09:03 |
| dulek | celebdor: https://docs.openstack.org/devstack/latest/plugins.html#plugin-sh-contract | 09:03 |
| celebdor | that's why I chose it | 09:03 |
| celebdor | test-config is unnecessary late | 09:03 |
| ltomasbo | celebdor, https://review.openstack.org/#/c/587706/ | 09:04 |
| dulek | celebdor: Sure, extra will work. ;) Okay, I'll merge this. | 09:04 |
| celebdor | thanks | 09:05 |
| *** garyloug has joined #openstack-kuryr | 09:29 | |
| ltomasbo | dulek, celebdor: do you know if we have some kind of cache when reading info from config.CONF? | 09:29 |
| ltomasbo | I'm getting some weird 'race' when creating 2 pods on two different namespaces | 09:30 |
| ltomasbo | when reading this config.CONF.neutron_defaults.pod_security_groups | 09:30 |
| ltomasbo | the first one gets it right | 09:30 |
| ltomasbo | but the second one gets some additions (leading to the wrong sg being applied -- or even duplicates) | 09:31 |
| ltomasbo | Ohh, I think I know what it can be... | 09:37 |
| dulek | ltomasbo: We shouldn't have cache. | 09:45 |
| dulek | ltomasbo: http://logs.openstack.org/53/588153/2/experimental/kuryr-kubernetes-tempest-daemon-containerized-octavia-namespace/8c405be/job-output.txt.gz#_2018-08-02_09_09_03_186101 - seems like namespace tests time out? | 09:45 |
| celebdor | ltomasbo: what? | 09:45 |
| ltomasbo | dulek, yes, I'm investigating | 09:45 |
| ltomasbo | dulek, it does not timeout, but the kuryr-controller crashed | 09:46 |
| ltomasbo | probably due to the sg group duplicate I'm hitting | 09:46 |
| dulek | ltomasbo: Whoops. :D | 09:46 |
| ltomasbo | but there may be something else | 09:46 |
| celebdor | ltomasbo: I am fixing my buildah patch so it works from fedora as well, not just from centos7 | 09:46 |
| celebdor | reducing the amount of ops | 09:46 |
| celebdor | later I'll try on that bug you found on the pkg_resources | 09:47 |
| ltomasbo | as I'm trying to deploy with namespace and I'm getting something bad | 09:47 |
| ltomasbo | dulek, not sure why it fails to create the networks due to project_id being None | 09:47 |
| ltomasbo | so, investigating both issues on 2 devstacks... | 09:47 |
| dulek | ltomasbo: Uhm, no idea to be honest. | 09:48 |
| ltomasbo | dulek, also, it is normal that tempest test for namespace fail now (regardless of those issues) | 09:49 |
| ltomasbo | since we added the isolation and they need to be updated wtih | 09:49 |
| ltomasbo | https://review.openstack.org/#/c/580678 | 09:49 |
| ltomasbo | dulek, ^^ | 09:49 |
| dulek | Okay! | 09:50 |
| ltomasbo | dulek, that one already had your +2 and celebdor's one before | 09:50 |
| dulek | ltomasbo: It has it again. :) | 09:51 |
| ltomasbo | \o/ | 09:51 |
| ltomasbo | xD | 09:51 |
| *** itzikb has joined #openstack-kuryr | 09:53 | |
| celebdor | welcome itzikb! | 09:57 |
| itzikb | celebdor: Thanks! :-) | 09:57 |
| celebdor | itzikb: regarding https://review.openstack.org/#/c/585706/ . I was not aware of it (please add me as reviewer to your patches) and I made one yesterday that moves the build to extra, which is earlier | 09:58 |
| celebdor | we ended up merging that one | 09:58 |
| celebdor | sorry for the trouble :/ | 09:58 |
| celebdor | dulek: irenab: https://review.openstack.org/#/c/587873/ | 09:59 |
| openstackgerrit | Merged openstack/kuryr-tempest-plugin master: Make Port and http/https configurable for the test container https://review.openstack.org/587746 | 09:59 |
| celebdor | so that we don't use dockerhub's possibly stale kuryr/demo anymore | 09:59 |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Avoid namespace security group override https://review.openstack.org/588197 | 10:04 |
| openstackgerrit | Merged openstack/kuryr-tempest-plugin master: devstack: Move container build to stack extra https://review.openstack.org/588074 | 10:04 |
| ltomasbo | dulek, celebdor https://review.openstack.org/#/c/588197/ | 10:05 |
| ltomasbo | I merged this yesterday, and there is already a bug /o\ | 10:05 |
| ltomasbo | :( | 10:05 |
| dulek | ltomasbo: Hey, what's the difference in this code? | 10:06 |
| dulek | ltomasbo: Option sends in a list. It shouldn't matter really… | 10:06 |
| dulek | ltomasbo: Oh, you want to say that reading a config option and changing it in runtime changes it?! | 10:06 |
| ltomasbo | dulek, yep | 10:06 |
| ltomasbo | I was getting the default sec on the first time, lets say XXX | 10:07 |
| ltomasbo | and as it was adding 2 more to sg_list | 10:07 |
| ltomasbo | the next call to get_security_group was getting | 10:07 |
| ltomasbo | xxx,yyy,zzz as the neutron_defaults.pod_security_groups | 10:07 |
| dulek | ltomasbo: I wouldn't ever expect this. :D | 10:07 |
| ltomasbo | same here... | 10:07 |
| dulek | ltomasbo: Commented with an idea there. | 10:08 |
| ltomasbo | and it only happens if the calls are subsequent | 10:08 |
| ltomasbo | if I wait a bit, then it does not happen | 10:08 |
| ltomasbo | doing this | 10:08 |
| ltomasbo | kubectl -n test run --image kuryr/demo demo | 10:08 |
| ltomasbo | kubectl -n luis run --image kuryr/demo demo | 10:08 |
| ltomasbo | will fail without that modification | 10:08 |
| ltomasbo | but doing: | 10:08 |
| ltomasbo | kubectl -n test run --image kuryr/demo demo | 10:08 |
| ltomasbo | sleep 20 | 10:08 |
| ltomasbo | kubectl -n luis run --image kuryr/demo demo | 10:08 |
| ltomasbo | 10:08 | |
| ltomasbo | will not fail | 10:09 |
| ltomasbo | strange... | 10:09 |
| ltomasbo | let me add a note as you suggest | 10:09 |
| ltomasbo | ahh, ok, I'll do the copy... | 10:10 |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Avoid namespace security group override https://review.openstack.org/588197 | 10:12 |
| openstackgerrit | Merged openstack/kuryr-kubernetes master: Ensure no api_root value for containerized deployments https://review.openstack.org/588153 | 10:13 |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Add namespace isolation for services https://review.openstack.org/581421 | 10:13 |
| ltomasbo | celebdor, dulek: it is too late to stop this one: https://review.openstack.org/#/c/580678 | 10:21 |
| ltomasbo | genadi has a point on the cleanup | 10:21 |
| dulek | ltomasbo: It's not, -W should stop it from merging. | 10:22 |
| dulek | ltomasbo: Which I've did already. :) | 10:22 |
| ltomasbo | ok | 10:22 |
| ltomasbo | let me change 1 thing fast, and we can do it again! | 10:22 |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-tempest-plugin master: Namespace isolation tempest coverage https://review.openstack.org/580678 | 10:24 |
| ltomasbo | dulek, done ^ | 10:24 |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-tempest-plugin master: Namespace svc isolation tempest coverage https://review.openstack.org/587778 | 10:25 |
| openstackgerrit | Peng Liu proposed openstack/kuryr-kubernetes master: Implement NPWG multi-vif driver https://review.openstack.org/578009 | 10:27 |
| ltomasbo | celebdor, can you please re-add +2/+W to https://review.openstack.org/#/c/580678 (I just needed to do a small fix) | 10:30 |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Avoid namespace security group overwrite https://review.openstack.org/588197 | 10:33 |
| celebdor | done | 10:34 |
| ltomasbo | thanks! | 10:38 |
| irenab | ltomasbo, ping | 10:44 |
| irenab | I am looking at the fix here https://review.openstack.org/#/c/588197/3/kuryr_kubernetes/controller/drivers/namespace_security_groups.py and not sure whatwasn't right with reading defauls sgs from the config | 10:45 |
| irenab | ltomasbo, never mind, its obvious .. | 10:46 |
| ltomasbo | irenab, I didn't expected either | 10:46 |
| irenab | python ... | 10:46 |
| ltomasbo | but when I created 2 pods (one after another | 10:46 |
| ltomasbo | the neutron_default.pod_securigy_gruoup got overwritten | 10:47 |
| irenab | it just appends to the list you read | 10:47 |
| irenab | points to the same one | 10:47 |
| ltomasbo | and the second pod was getting the same sg as the first one (with is wrong) | 10:47 |
| ltomasbo | yep, pretty much | 10:47 |
| ltomasbo | but if I wait between pods creation, it does not happen... | 10:48 |
| irenab | hmm, wierd | 10:48 |
| ltomasbo | yep | 10:48 |
| ltomasbo | as you said, python... xD | 10:48 |
| ltomasbo | and I think I'm getting something similar for the project id... | 10:50 |
| irenab | ltomasbo, maybe better to define my_sg and then extend it with the default? | 10:51 |
| ltomasbo | umm, that should get and empty list by default, right? | 10:51 |
| ltomasbo | let me check... | 10:51 |
| irenab | I meant the opposite, start with the sg from the spec and add the default ones to it | 10:52 |
| ltomasbo | sure, also I check that security group is a ListOpt, but it has no default | 10:53 |
| ltomasbo | irenab, you mean something more similar to what I did for ps1 https://review.openstack.org/#/c/588197/1/kuryr_kubernetes/controller/drivers/namespace_security_groups.py | 10:53 |
| ltomasbo | irenab, but moving L52 and 53 to the end | 10:54 |
| irenab | ltomasbo, yes, you can just use extend(CONF.sgs) instead of iterating and appending | 10:55 |
| ltomasbo | ok | 10:56 |
| irenab | I think as long as you does not do xxx=CONF.sgs, python should not trick the pointers | 10:56 |
| irenab | ltomasbo, let me if you want to respin or merge as it is now, since it does the work | 11:00 |
| ltomasbo | irenab, I'll send a modif in 30 secs! | 11:01 |
| irenab | sure | 11:01 |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Avoid namespace security group overwrite https://review.openstack.org/588197 | 11:02 |
| ltomasbo | irenab, ^^ | 11:02 |
| irenab | posted comment on the loop | 11:04 |
| ltomasbo | irenab, I cannot use extend there | 11:07 |
| ltomasbo | as I need to do the str() cast | 11:08 |
| irenab | isn't is another list? | 11:08 |
| ltomasbo | is there a trick to do the str() at the same time? | 11:08 |
| irenab | got it, fine then | 11:08 |
| ltomasbo | is dmellado working today? | 11:08 |
| irenab | I hate this once line python magic that does 10 things | 11:08 |
| ltomasbo | xD | 11:08 |
| ltomasbo | I'm on bug hunting day today, yet another one.... | 11:10 |
| celebdor | ltomasbo: which? | 11:10 |
| celebdor | ltomasbo: dmellado is on PTO today | 11:11 |
| celebdor | traveling to Galiza | 11:11 |
| ltomasbo | ok, I found that one of his patch sets deleted one line, and that was making get_project to return None, instead of project id | 11:11 |
| ltomasbo | I'll fill the bug and send a fix right now | 11:11 |
| ltomasbo | https://review.openstack.org/#/c/580982/11/kuryr_kubernetes/controller/drivers/default_project.py | 11:11 |
| dmellado | ltomasbo: just landed | 11:13 |
| dmellado | feel free sending a patch | 11:13 |
| ltomasbo | yep, I just filled the bug | 11:14 |
| celebdor | dmellado: heh... You botched it | 11:14 |
| ltomasbo | sending the patch in 1 min | 11:14 |
| celebdor | xD | 11:14 |
| dmellado | yeah, just a typo | 11:14 |
| dmellado | your fault for not noticing it | 11:14 |
| dmellado | botchers | 11:14 |
| dmellado | ltomasbo++ | 11:14 |
| dmellado | celebdor-- | 11:14 |
| dmellado | xD | 11:15 |
| ltomasbo | xD | 11:15 |
| ltomasbo | well, this was breaking my tempest namespace gate | 11:15 |
| ltomasbo | that is how I notice it | 11:15 |
| ltomasbo | xD | 11:15 |
| dmellado | xD | 11:15 |
| dmellado | in any case just a typo | 11:15 |
| dmellado | thanks for noticing it xD | 11:15 |
| dmellado | my bad, too much lack of sleep as of late | 11:16 |
| dmellado | AlexeyPerevalov: you pinged me before, please notice the planned meeting on mon | 11:16 |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Ensure project id is returned for DefaultNamespaceProjectDriver https://review.openstack.org/588214 | 11:16 |
| ltomasbo | dmellado, celebdor ^^ | 11:17 |
| *** rh-jelabarre has joined #openstack-kuryr | 11:18 | |
| dmellado | ltomasbo: btw, I don't know if you noticed because of the PTO and so | 11:18 |
| dmellado | http://lists.openstack.org/pipermail/openstack-dev/2018-July/132659.html | 11:19 |
| ltomasbo | dmellado, is that todya? | 11:19 |
| dmellado | nope, mon | 11:19 |
| ltomasbo | ahh, ok | 11:19 |
| dmellado | today I'm on pto and logging off in 2 min xD | 11:19 |
| dmellado | I forgot to turn off my phone irc, tbh xD | 11:19 |
| ltomasbo | ok ok | 11:19 |
| ltomasbo | I'll add it to my calendar | 11:19 |
| ltomasbo | thanks for the reminder! | 11:20 |
| dmellado | yw! | 11:20 |
| ltomasbo | dulek, irenab: another easy to review patch https://review.openstack.org/#/c/588214/ | 11:20 |
| ltomasbo | and dulek check this again please: https://review.openstack.org/#/c/588197/ | 11:20 |
| ltomasbo | dulek, that is catch by my namespace tempest tests actually... | 11:22 |
| ltomasbo | but you are right, I'll add it | 11:23 |
| dulek | ltomasbo: If it's not too much trouble. | 11:23 |
| ltomasbo | it is the same function as before | 11:23 |
| ltomasbo | so, it is basically copy and paste | 11:23 |
| ltomasbo | should I add the ones for the other driver? the networkpolicy too? | 11:23 |
| dulek | ltomasbo: I know that we have some big gaps in UT, so if you'll need to setup a whole new testcase, then I understand it's troublesome. | 11:24 |
| dulek | ltomasbo: I guess so. | 11:24 |
| ltomasbo | ok, give me 5 min! | 11:24 |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Ensure project id is returned for DefaultNamespaceProjectDriver https://review.openstack.org/588214 | 11:30 |
| ltomasbo | dulek, ^^ done! | 11:30 |
| ltomasbo | and break for lunch... | 11:31 |
| dulek | ltomasbo: It took you less than 5 minutes. xD | 11:39 |
| dulek | dmellado: Hey, what the heck: http://logs.openstack.org/73/587873/3/check/kuryr-kubernetes-tempest-multinode-octavia/0d18504/compute1/logs/local_conf.txt ? | 11:48 |
| dulek | dmellado: We're not running kubelet on the compute node in multinode? | 11:49 |
| celebdor | dulek: no | 11:52 |
| celebdor | that node, IIRC, is only for running octavia amps | 11:52 |
| dulek | celebdor: Was that worth all the work? I though we'll be testing multiple kubelets and pod-to-pod-on-other-node traffic there. | 11:53 |
| celebdor | dulek: heh | 11:53 |
| celebdor | not my planning | 11:53 |
| celebdor | that's dmellado's | 11:54 |
| celebdor | I expect that he'll add extra nodes | 11:54 |
| celebdor | for using them as k8s nodes | 11:54 |
| dulek | celebdor: I doubt there's a nodeset that has 3 VMs. Maybe he wanted to do nested there? | 11:54 |
| celebdor | IIRC he said that adding more nodes was not an issue | 11:55 |
| openstackgerrit | Michał Dulko proposed openstack/kuryr-kubernetes master: Enable kubelet on multinode's gate subnode https://review.openstack.org/588222 | 11:57 |
| openstackgerrit | Michał Dulko proposed openstack/kuryr-kubernetes master: WIP: Add HA gate https://review.openstack.org/588223 | 11:57 |
| dulek | Let's see how this'll work. :) | 11:57 |
| dulek | Ah, I'm missing one thing there probably… | 11:57 |
| openstackgerrit | Michał Dulko proposed openstack/kuryr-kubernetes master: Enable kubelet on multinode's gate subnode https://review.openstack.org/588222 | 11:59 |
| openstackgerrit | Michał Dulko proposed openstack/kuryr-kubernetes master: WIP: Add HA gate https://review.openstack.org/588223 | 11:59 |
| celebdor | dulek: I was thinking if it would be worth it to make it so that the containerized devstack gates | 12:05 |
| celebdor | had the kuryr code on a mounted volume | 12:05 |
| celebdor | so that it would not be necessary to rebuild the image for changes | 12:06 |
| dulek | celebdor: Aw, forgot to turn on containerized. xD | 12:06 |
| dulek | celebdor: I assume we're talking about development here - no need for that in gates. | 12:09 |
| celebdor | dulek: righ | 12:10 |
| openstackgerrit | Michał Dulko proposed openstack/kuryr-kubernetes master: WIP: Add HA gate https://review.openstack.org/588223 | 12:10 |
| celebdor | *right | 12:10 |
| dulek | celebdor: We've thought about something like this in Kolla context a while ago. It should work if we'll install kuryr on containers with `pip -e <dir-where-we'll-mount-the-code>`. | 12:10 |
| dulek | I mean in the Dockerfile. | 12:10 |
| celebdor | dulek: and rely on the volume replacing it getting the egg-link to work | 12:12 |
| dulek | celebdor: Well, I guess its a nice idea for Friday evening writeup. ;) | 12:13 |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Add namespace isolation for services https://review.openstack.org/581421 | 12:13 |
| celebdor | then you just make the changes on /opt/stack/kuryr-kubernetes | 12:14 |
| celebdor | kill the container | 12:14 |
| celebdor | and when it spawns again | 12:14 |
| celebdor | it will use the new code | 12:14 |
| dulek | celebdor: One minor thing I'd like to fix is kuryr-daemon container killing timeout. | 12:15 |
| dulek | celebdor: It doesn't stop with SIGTERM and K8s is killing them, I think. | 12:16 |
| dulek | celebdor: Anyway `kubectl delete kuryr-cni` takes ~60 seconds to happen. | 12:16 |
| celebdor | dulek: do you know why? | 12:17 |
| dulek | celebdor: I've analyzed this a while ago, but thought your changes to watcher.py will fix it. | 12:20 |
| dulek | celebdor: So probably got it wrong. :D | 12:21 |
| openstackgerrit | Merged openstack/kuryr-kubernetes master: kuryr-tempest-plugin devstack plugin to build kuryr/demo https://review.openstack.org/587873 | 12:27 |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Avoid namespace security group overwrite https://review.openstack.org/588197 | 12:32 |
| ltomasbo | dulek, ^^ I fixed the dummy mistake when changing the sg list | 12:32 |
| openstackgerrit | Merged openstack/kuryr-tempest-plugin master: Namespace isolation tempest coverage https://review.openstack.org/580678 | 12:45 |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: Add buildah support https://review.openstack.org/571707 | 12:48 |
| openstackgerrit | Peng Liu proposed openstack/kuryr-kubernetes master: Implement NPWG multi-vif driver https://review.openstack.org/578009 | 13:10 |
| celebdor | shadower: my phone literally overheated to hell | 13:13 |
| celebdor | I have to see how I fix it | 13:13 |
| celebdor | sorry | 13:13 |
| celebdor | shadower: the focus should be: | 13:14 |
| *** rh-jelabarre has quit IRC | 13:14 | |
| celebdor | Keeping env for OpenShift QE | 13:14 |
| celebdor | and increase the coverage in the places that we know that our stack (openstack) can have trouble | 13:15 |
| celebdor | both in storage and in networking | 13:15 |
| celebdor | Oh, and have the openstack/kuryr-kubernetes gates have ocp tests as third party ci | 13:19 |
| *** janki has quit IRC | 13:24 | |
| ltomasbo | celebdor, wrong channel? | 13:29 |
| ltomasbo | xD | 13:29 |
| *** garyloug has quit IRC | 13:34 | |
| celebdor | ltomasbo: It also applies here :-) | 13:35 |
| * celebdor lunch | 13:35 | |
| ltomasbo | :D | 13:35 |
| openstackgerrit | Merged openstack/kuryr-kubernetes master: Ensure project id is returned for DefaultNamespaceProjectDriver https://review.openstack.org/588214 | 13:39 |
| shadower | celebdor: yea I agree | 13:40 |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Add namespace isolation for services https://review.openstack.org/581421 | 13:46 |
| *** tzumainn has joined #openstack-kuryr | 13:51 | |
| openstackgerrit | Peng Liu proposed openstack/kuryr-kubernetes master: Implement NPWG multi-vif driver https://review.openstack.org/578009 | 13:52 |
| *** shachar has joined #openstack-kuryr | 13:53 | |
| *** shachar has quit IRC | 13:54 | |
| *** janki has joined #openstack-kuryr | 14:07 | |
| *** garyloug has joined #openstack-kuryr | 14:09 | |
| garyloug | Hi all, I‘m finding that ports are not being deleted when I delete my pods. Looks like the release_vif function is not being called in the controller driver. I’ve seen this on nested-macvlan driver and the in review nested-dpdk. Anyone seen similar or have any ideas what could be the issue in my env? | 14:10 |
| *** celebdor has quit IRC | 14:12 | |
| *** celebdor has joined #openstack-kuryr | 14:15 | |
| *** hongbin_ has joined #openstack-kuryr | 14:16 | |
| *** celebdor1 has joined #openstack-kuryr | 14:33 | |
| *** celebdor has quit IRC | 14:36 | |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: devstack: deploy openshift registry on origin envs https://review.openstack.org/580201 | 14:45 |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: drop deprecated oadm https://review.openstack.org/588303 | 14:49 |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: drop deprecated oadm https://review.openstack.org/588303 | 14:50 |
| openstackgerrit | Genadi Chereshnya proposed openstack/kuryr-tempest-plugin master: Testing VM to loadBalancer service connectivity https://review.openstack.org/571968 | 14:51 |
| *** celebdor1 is now known as celebdor | 14:52 | |
| *** rh-jelabarre has joined #openstack-kuryr | 14:54 | |
| *** itzikb has quit IRC | 15:36 | |
| openstackgerrit | Michał Dulko proposed openstack/kuryr-kubernetes master: Enable kubelet on multinode's gate subnode https://review.openstack.org/588222 | 15:49 |
| openstackgerrit | Michał Dulko proposed openstack/kuryr-kubernetes master: WIP: Add HA gate https://review.openstack.org/588223 | 15:49 |
| *** janki has quit IRC | 15:58 | |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: devstack: deploy openshift registry on origin envs https://review.openstack.org/580201 | 16:06 |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: drop deprecated oadm https://review.openstack.org/588303 | 16:08 |
| *** spotz has quit IRC | 16:18 | |
| *** janki has joined #openstack-kuryr | 16:31 | |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Avoid namespace security group overwrite https://review.openstack.org/588197 | 16:46 |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Ensure OpenShift gate uses the namespace subnet/sg drivers https://review.openstack.org/580680 | 16:46 |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Add namespace isolation for services https://review.openstack.org/581421 | 16:51 |
| *** janki has quit IRC | 16:52 | |
| ltomasbo | celebdor, I did some rebase, can you +W this: https://review.openstack.org/#/c/588197/ | 16:54 |
| *** garyloug has quit IRC | 17:04 | |
| openstackgerrit | Merged openstack/kuryr-kubernetes master: Avoid namespace security group overwrite https://review.openstack.org/588197 | 19:36 |
| *** rh-jelabarre has quit IRC | 20:35 | |
| *** hongbin_ has quit IRC | 22:39 | |
| *** tzumainn has quit IRC | 22:56 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!