*** limao has joined #openstack-kuryr | 00:26 | |
*** ivc has quit IRC | 00:31 | |
*** ivc has joined #openstack-kuryr | 00:36 | |
*** rwallner has joined #openstack-kuryr | 00:39 | |
*** kiennt has joined #openstack-kuryr | 00:47 | |
*** gouthamr has quit IRC | 01:05 | |
*** yedongcan has joined #openstack-kuryr | 01:21 | |
*** rwallner has quit IRC | 01:25 | |
*** hongbin has joined #openstack-kuryr | 01:59 | |
*** wangbo has joined #openstack-kuryr | 02:19 | |
*** hongbin_ has joined #openstack-kuryr | 02:37 | |
*** hongbin has quit IRC | 02:39 | |
hongbin_ | limao: hi liping | 02:44 |
---|---|---|
*** yedongcan has quit IRC | 02:48 | |
*** aojea has joined #openstack-kuryr | 02:53 | |
*** aojea has quit IRC | 02:58 | |
*** hongbin_ has quit IRC | 03:16 | |
*** limao has quit IRC | 03:16 | |
*** limao has joined #openstack-kuryr | 03:16 | |
*** limao has quit IRC | 03:19 | |
*** limao has joined #openstack-kuryr | 03:19 | |
*** hongbin has joined #openstack-kuryr | 03:32 | |
*** hongbin has quit IRC | 03:41 | |
*** limao has quit IRC | 03:59 | |
*** limao has joined #openstack-kuryr | 04:00 | |
*** limao has quit IRC | 04:04 | |
*** limao has joined #openstack-kuryr | 04:31 | |
*** limao has quit IRC | 04:36 | |
*** wangbo has quit IRC | 04:37 | |
*** limao has joined #openstack-kuryr | 04:51 | |
*** limao has quit IRC | 05:06 | |
*** limao has joined #openstack-kuryr | 05:07 | |
limao | hongbin_: sorry for the delay, I update in launchpad : https://bugs.launchpad.net/kuryr-libnetwork/+bug/1703698 | 05:24 |
openstack | Launchpad bug 1703698 in kuryr-libnetwork "Couldn't get docker plugin to work" [Undecided,New] | 05:24 |
*** aojea has joined #openstack-kuryr | 05:28 | |
*** wangbo has joined #openstack-kuryr | 05:28 | |
*** gouthamr has joined #openstack-kuryr | 05:34 | |
*** kiennt has quit IRC | 05:50 | |
*** aojea has quit IRC | 05:53 | |
*** kiennt has joined #openstack-kuryr | 05:58 | |
*** yedongcan has joined #openstack-kuryr | 06:00 | |
*** wangbo has quit IRC | 06:11 | |
*** wangbo has joined #openstack-kuryr | 06:13 | |
*** yboaron_ has joined #openstack-kuryr | 06:15 | |
*** limao has quit IRC | 06:28 | |
*** limao has joined #openstack-kuryr | 06:29 | |
*** limao has quit IRC | 06:33 | |
*** yedongcan has left #openstack-kuryr | 06:43 | |
*** limao has joined #openstack-kuryr | 06:46 | |
*** wangbo has quit IRC | 06:47 | |
*** kzaitsev_ws has joined #openstack-kuryr | 06:48 | |
*** wangbo has joined #openstack-kuryr | 06:52 | |
*** pcaruana has joined #openstack-kuryr | 06:52 | |
*** yedongcan has joined #openstack-kuryr | 07:09 | |
*** yedongcan has quit IRC | 07:10 | |
*** yedongcan has joined #openstack-kuryr | 07:11 | |
*** roeyc has joined #openstack-kuryr | 07:12 | |
*** janki has joined #openstack-kuryr | 07:16 | |
*** aojea has joined #openstack-kuryr | 07:17 | |
*** gouthamr has quit IRC | 07:22 | |
*** garyloug has joined #openstack-kuryr | 07:41 | |
*** garyloug has quit IRC | 07:42 | |
*** egonzalez has joined #openstack-kuryr | 07:47 | |
openstackgerrit | Kirill Zaitsev proposed openstack/kuryr-kubernetes master: Allow passing multiple VIFs to CNI https://review.openstack.org/471012 | 07:55 |
*** wangbo has quit IRC | 08:02 | |
*** wangbo has joined #openstack-kuryr | 08:05 | |
*** vikasc has quit IRC | 08:28 | |
ltomasbo | hi apuimedo | 08:30 |
apuimedo | ltomasbo: hi | 08:31 |
ltomasbo | why do you create_k8s_router_fake_service in https://review.openstack.org/#/c/467241/8/devstack/plugin.sh? | 08:32 |
ltomasbo | why do you need | 08:32 |
apuimedo | ltomasbo: this is defensive programming :-) | 08:33 |
apuimedo | Imagine you create a lot of services | 08:33 |
apuimedo | with each service you create, you increase the chance of K8s API deciding to use the IP/clusterIP of the router interface of the services network | 08:34 |
ltomasbo | but it gets and IP in a different subnet | 08:36 |
apuimedo | ltomasbo: how so? | 08:36 |
apuimedo | Let's say we have a deployment with three subnets | 08:36 |
ltomasbo | I see the k8s_service-subnet is 10.0.0.64/26 | 08:36 |
ltomasbo | and the svc/kuberentes gets the 10.0.0.65 | 08:37 |
ltomasbo | but the svc/kuryr-svc-router gets the 10.0.0.126 | 08:37 |
apuimedo | VM subnet 10.0.0.0/24, pod subnet 10.10.0.0/24 and services subnet 10.20.0.0/24 | 08:37 |
ltomasbo | ahh, 126, not 128 | 08:37 |
*** wangbo has quit IRC | 08:37 | |
apuimedo | right | 08:38 |
ltomasbo | not in devstack | 08:38 |
ltomasbo | subnet is 10.0.0.0/26 | 08:38 |
ltomasbo | and the service is 10.0.0.64/26 | 08:38 |
apuimedo | pod subnet is 10.0.0.0/26 which means 10.0.0.1-10.0.0.63 | 08:39 |
apuimedo | and service subnet is 10.0.0.64/26 which means 10.0.0.64-10.0.0.127 | 08:39 |
apuimedo | 127 is broadcast | 08:39 |
apuimedo | so the router goes on 126 | 08:40 |
apuimedo | which is what we tell to the fake service | 08:40 |
ltomasbo | yep, I missed 126 by 128 | 08:40 |
ltomasbo | ok, but the lbaas is complaining in the kuryr-kubernetes logs | 08:40 |
dmellado | about a missing endpoint, actually | 08:40 |
ltomasbo | yep | 08:40 |
dmellado | so you're gonna end up without fuet, apuimedo | 08:41 |
*** vikasc has joined #openstack-kuryr | 08:41 | |
*** wangbo has joined #openstack-kuryr | 08:44 | |
*** garyloug has joined #openstack-kuryr | 08:45 | |
apuimedo | I know | 08:46 |
apuimedo | that complaint is because we have a bug | 08:47 |
apuimedo | I thought I already reported it | 08:47 |
apuimedo | we don't support well endpointless services | 08:47 |
apuimedo | which are allowed in k8s | 08:47 |
ltomasbo | ok, so that should be a modification to the lbaas support, right? | 08:48 |
kzaitsev_ws | ltomasbo: right, I can even point you to exact place in the code that does that | 08:50 |
ltomasbo | :D | 08:52 |
ltomasbo | I'm taking a look... | 08:52 |
kzaitsev_ws | https://github.com/openstack/kuryr-kubernetes/blob/master/kuryr_kubernetes/controller/handlers/lbaas.py#L230 | 08:52 |
apuimedo | yup | 08:58 |
apuimedo | the assumption that services have endpoints is wrong | 08:58 |
apuimedo | kzaitsev_ws: ltomasbo: maybe if we just check if the service lacks a selector as a basis to skip it | 08:59 |
*** limao has quit IRC | 09:12 | |
ltomasbo | I quickly included this: | 09:12 |
*** limao has joined #openstack-kuryr | 09:12 | |
ltomasbo | 230 def _should_ignore(self, endpoints, lbaas_spec): | 09:12 |
ltomasbo | 231 return not(lbaas_spec and | 09:12 |
ltomasbo | 232 self._has_pods(endpoints) and | 09:12 |
ltomasbo | ~ 233 self._is_lbaas_spec_in_sync(endpoints, lbaas_spec) and | 09:12 |
ltomasbo | + 234 endpoints['metadata']['name'] != 'kuryr-svc-router') | 09:12 |
dmellado | oh, kzaitsev_ws so ltomasbo was cheating when he showed me xD | 09:12 |
*** limao has quit IRC | 09:14 | |
*** limao has joined #openstack-kuryr | 09:14 | |
*** huats has joined #openstack-kuryr | 09:15 | |
apuimedo | dmellado: when he showed you what? | 09:17 |
dmellado | apuimedo: kzaitsev_ws's link | 09:17 |
apuimedo | ltomasbo: that's too specific | 09:17 |
dmellado | don't think bad xD | 09:17 |
ltomasbo | just testing if it fix the problem | 09:18 |
dmellado | apuimedo: btw, will tell you afterwards but I was thinking about having some kind of code walk-through sessions for new contributors | 09:18 |
dmellado | recorded and so on | 09:18 |
*** vikasc has quit IRC | 09:18 | |
dmellado | and I was wondering if you had something like that already | 09:18 |
dmellado | there's an 'upcoming' guy who I'd really like to attend | 09:18 |
dmellado | I'd do the same for the tempest plugin | 09:18 |
apuimedo | dmellado: we don't | 09:19 |
apuimedo | but it is a nice idea | 09:19 |
kzaitsev_ws | I managed to have lbaas for podless services some time ago | 09:19 |
apuimedo | kzaitsev_ws: you don't need lbaas for podless services, do you? | 09:21 |
kzaitsev_ws | I can dig out the patch, but it should be rather easy to replicate that anywat | 09:21 |
dmellado | kzaitsev_ws: would you be up also for that session? I'll be having a to-be-mentored guy joining soon xD | 09:21 |
kzaitsev_ws | apuimedo: well I wanted kubernetes service to be handled correctly =) | 09:22 |
kzaitsev_ws | but in the end it turned out I did not need that =) | 09:23 |
kzaitsev_ws | dmellado: which session, where? =) | 09:23 |
apuimedo | need to move to a place with better internet | 09:23 |
apuimedo | dmellado: I'll join the meeting with you asap | 09:23 |
openstackgerrit | Kirill Zaitsev proposed openstack/kuryr-kubernetes master: Add SR-IOV capabilities to VIF handler https://review.openstack.org/462455 | 09:24 |
dmellado | kzaitsev_ws: we were thinking about hosting some sessions for code walk-thru for new contributors | 09:24 |
*** limao has quit IRC | 09:24 | |
dmellado | for both kuryr-kubernetes and the testing part | 09:24 |
dmellado | apuimedo: ack, heading to the meeting room | 09:25 |
kzaitsev_ws | dmellado: when and where? I think I can contribute if I don't have concurrent meetings (which I might but should not have) | 09:26 |
dmellado | kzaitsev_ws: TBD | 09:26 |
kzaitsev_ws | ah =) ok | 09:26 |
dmellado | we'll discuss it and sort out dates :D | 09:26 |
kzaitsev_ws | I actually wanted to draw my own diagram of the controll-flow of kuryr-k8s. Similar to the ones I use to draw on the whiteboards when explaining how things work | 09:27 |
kzaitsev_ws | will get to that sometime | 09:27 |
ltomasbo | apuimedo, not a good solution, and not working either... | 09:28 |
dmellado | apuimedo: chapucero | 09:32 |
dmellado | xD | 09:32 |
kzaitsev_ws | ltomasbo: I found my services_without_pods commit. its dirty and hacky +) | 09:33 |
kzaitsev_ws | can share it if you want to though | 09:33 |
kzaitsev_ws | btw, folks. Can I kill LOG.debug("%s handler(s) available", len(handlers)) pleeeeeeease? | 09:34 |
*** vikasc has joined #openstack-kuryr | 09:35 | |
kzaitsev_ws | I can spend some time adding better debugging output. but will only do that if we agree to kill `1 handler(s) available` log avalanche. =) | 09:35 |
dmellado | lol, that's fine for me xD | 09:36 |
dmellado | scooby doo, err apuimedo where are you? | 09:38 |
openstackgerrit | Kirill Zaitsev proposed openstack/kuryr-kubernetes master: Add SR-IOV binding driver to CNI https://review.openstack.org/462456 | 09:40 |
kzaitsev_ws | hm | 10:06 |
kzaitsev_ws | does anyone have a master installation and can check smth for me? | 10:06 |
*** kiennt has quit IRC | 10:09 | |
apuimedo | kzaitsev_ws: ltomasbo prolly has | 10:15 |
apuimedo | I could have one in a little hilw | 10:15 |
apuimedo | *while | 10:15 |
apuimedo | kzaitsev_ws: yeah, you can kill it | 10:15 |
irenab | ltomasbo, dmellado, apuimedo did someone ran nested setup lately, have an issue that pod cannot be created with some wierd error, that host side interface is not created | 10:15 |
apuimedo | irenab: you mean that pyroute2 fails to create the vlan device? | 10:16 |
irenab | yes | 10:16 |
kzaitsev_ws | apuimedo: I see that during pod deletion the CNI is called up to 3 times. it's by design idempotent and doesn't crash, but that's weird. I wanted to confirm if it's I or if everyone get's the same behaviour | 10:17 |
ltomasbo | kzaitsev_ws, I have a setup, let me know what to try | 10:18 |
apuimedo | kzaitsev_ws: I did not notice that | 10:19 |
apuimedo | but that's bad | 10:19 |
kzaitsev_ws | ltomasbo: one thing is I can probably mishandle the moment when to let the handler go | 10:19 |
kzaitsev_ws | in the CNI side | 10:20 |
apuimedo | kzaitsev_ws: did you check if it didn't just retry due to some resource not ready or somethign? | 10:20 |
apuimedo | is it really the kubelet doing multiple calls? | 10:20 |
apuimedo | or just the handler going more than once | 10:20 |
kzaitsev_ws | apuimedo: that's what I'm trying to understand. actually for resource deletion (even in unmodified code) the interface will get deleted on the first event | 10:24 |
kzaitsev_ws | https://github.com/openstack/kuryr-kubernetes/blob/master/kuryr_kubernetes/cni/handlers.py#L40 and https://github.com/openstack/kuryr-kubernetes/blob/master/kuryr_kubernetes/cni/handlers.py#L87 | 10:24 |
apuimedo | kzaitsev_ws: if it find the annotation | 10:26 |
apuimedo | s/find/finds/ | 10:26 |
apuimedo | kzaitsev_ws: we should be able to tear down a pod without going to the K8s API | 10:26 |
apuimedo | I had a proposal for storing the vif information of a pod in /var/run/kuryr | 10:27 |
apuimedo | to avoid such things and speed it up | 10:27 |
apuimedo | but didn't get down to implementing it | 10:27 |
*** yamamoto has quit IRC | 10:31 | |
*** wangbo has quit IRC | 10:39 | |
kzaitsev_ws | apuimedo: yeah. so I suppose, that I get the same annotation several times | 10:46 |
apuimedo | right, same event several times | 10:50 |
*** atoth has joined #openstack-kuryr | 10:55 | |
*** wangbo has joined #openstack-kuryr | 10:58 | |
*** aojea has quit IRC | 11:04 | |
*** aojea has joined #openstack-kuryr | 11:06 | |
*** aojea has quit IRC | 11:09 | |
*** yamamoto has joined #openstack-kuryr | 11:13 | |
*** yamamoto has quit IRC | 11:18 | |
*** yamamoto has joined #openstack-kuryr | 11:21 | |
*** yedongcan has left #openstack-kuryr | 11:23 | |
*** yamamoto has quit IRC | 11:26 | |
*** yamamoto has joined #openstack-kuryr | 11:36 | |
*** yamamoto has quit IRC | 11:40 | |
*** yamamoto has joined #openstack-kuryr | 11:50 | |
*** yamamoto has quit IRC | 11:54 | |
*** yamamoto has joined #openstack-kuryr | 12:17 | |
*** yboaron_ has quit IRC | 12:22 | |
*** mattmceuen has quit IRC | 12:58 | |
*** aojea has joined #openstack-kuryr | 13:24 | |
*** wangbo has quit IRC | 13:24 | |
*** aojea has quit IRC | 13:25 | |
*** aojea has joined #openstack-kuryr | 13:25 | |
*** deep-book-gk_ has joined #openstack-kuryr | 13:27 | |
*** deep-book-gk_ has left #openstack-kuryr | 13:28 | |
*** mattmceuen has joined #openstack-kuryr | 13:33 | |
*** roeyc has quit IRC | 13:56 | |
*** hongbin has joined #openstack-kuryr | 14:06 | |
openstackgerrit | Kirill Zaitsev proposed openstack/kuryr-kubernetes master: Allow passing multiple VIFs to CNI https://review.openstack.org/471012 | 14:08 |
openstackgerrit | Kirill Zaitsev proposed openstack/kuryr-kubernetes master: [DRAFT] Allow requesting additional VIFs via annotation https://review.openstack.org/482544 | 14:08 |
openstackgerrit | Kirill Zaitsev proposed openstack/kuryr-kubernetes master: Add SR-IOV binding driver to CNI https://review.openstack.org/462456 | 14:08 |
openstackgerrit | Kirill Zaitsev proposed openstack/kuryr-kubernetes master: Add SR-IOV capabilities to VIF handler https://review.openstack.org/462455 | 14:08 |
apuimedo | kzaitsev_ws: rebase? | 14:15 |
openstackgerrit | OpenStack Proposal Bot proposed openstack/kuryr-tempest-plugin master: Updated from global requirements https://review.openstack.org/483411 | 14:15 |
kzaitsev_ws | yeah, sorry =) | 14:18 |
kzaitsev_ws | I found a small bug in the very 1st commit. or rather it was always there | 14:18 |
kzaitsev_ws | probably, but the very 1st in a chain is a good place to fix. when asking for `subnet.gateway` of the ovo if there is no gateway it would raise a NotImplemented exception | 14:19 |
kzaitsev_ws | if there is no gateway here https://github.com/openstack/kuryr-kubernetes/blob/master/kuryr_kubernetes/os_vif_util.py#L72 it's never set and ovo raises an exception here https://github.com/openstack/oslo.versionedobjects/blob/master/oslo_versionedobjects/base.py#L67 | 14:24 |
*** gouthamr has joined #openstack-kuryr | 14:37 | |
kzaitsev_ws | and yeah, only found it 'cause I added 2 subnets to the pod and one did not have a gateway =) | 14:37 |
*** janki has quit IRC | 15:17 | |
*** kzaitsev_ws has quit IRC | 15:24 | |
openstackgerrit | Jaivish Kothari(janonymous) proposed openstack/kuryr-kubernetes master: [DRAFT] CNI Daemon template https://review.openstack.org/480028 | 15:35 |
*** pcaruana has quit IRC | 15:38 | |
janonymous | apuimedo irenab kzaitsev_pi dmellado ltomasbo: i have uploaded initial draft version for review ^^ , please check. i was facing some issue with response body from daemon which i will check tommorrow, maybe something related to content-type..review comments would be helpful :) | 15:39 |
*** aojea has quit IRC | 15:55 | |
*** aojea has joined #openstack-kuryr | 16:02 | |
*** aojea has quit IRC | 16:07 | |
*** aojea has joined #openstack-kuryr | 16:12 | |
*** aojea has quit IRC | 16:16 | |
*** saneax has quit IRC | 16:18 | |
*** aojea has joined #openstack-kuryr | 16:21 | |
*** aojea has quit IRC | 16:25 | |
*** aojea has joined #openstack-kuryr | 16:30 | |
*** aojea has quit IRC | 16:36 | |
*** aojea has joined #openstack-kuryr | 16:39 | |
*** gouthamr has quit IRC | 16:41 | |
*** gouthamr has joined #openstack-kuryr | 16:41 | |
*** aojea has quit IRC | 16:44 | |
*** aojea has joined #openstack-kuryr | 16:48 | |
*** egonzalez has quit IRC | 16:50 | |
*** aojea has quit IRC | 16:52 | |
*** aojea has joined #openstack-kuryr | 16:57 | |
*** aojea has quit IRC | 17:02 | |
*** aojea has joined #openstack-kuryr | 17:06 | |
*** aojea has quit IRC | 17:10 | |
*** aojea has joined #openstack-kuryr | 17:15 | |
*** aojea has quit IRC | 17:20 | |
*** aojea has joined #openstack-kuryr | 17:24 | |
*** aojea has quit IRC | 17:29 | |
*** tonanhngo has joined #openstack-kuryr | 17:36 | |
*** roeyc has joined #openstack-kuryr | 18:06 | |
*** roeyc has quit IRC | 18:10 | |
*** garyloug has quit IRC | 18:12 | |
*** pcaruana has joined #openstack-kuryr | 18:38 | |
openstackgerrit | OpenStack Proposal Bot proposed openstack/fuxi master: Updated from global requirements https://review.openstack.org/483522 | 19:29 |
*** yboaron_ has joined #openstack-kuryr | 19:33 | |
*** pcaruana has quit IRC | 19:52 | |
*** rwallner has joined #openstack-kuryr | 19:59 | |
*** rwallner has quit IRC | 20:03 | |
*** aojea has joined #openstack-kuryr | 20:35 | |
*** aojea has quit IRC | 20:41 | |
*** rwallner has joined #openstack-kuryr | 21:07 | |
*** limao has joined #openstack-kuryr | 21:26 | |
*** rwallner has quit IRC | 21:31 | |
*** limao_ has joined #openstack-kuryr | 21:33 | |
*** limao has quit IRC | 21:33 | |
*** gouthamr has quit IRC | 21:47 | |
*** gouthamr has joined #openstack-kuryr | 21:49 | |
*** saneax has joined #openstack-kuryr | 22:01 | |
*** mattmceuen has quit IRC | 22:05 | |
*** limao_ has quit IRC | 22:08 | |
*** limao has joined #openstack-kuryr | 22:08 | |
*** gouthamr has quit IRC | 22:10 | |
*** rwallner has joined #openstack-kuryr | 22:28 | |
*** rwallner has quit IRC | 22:32 | |
*** aojea has joined #openstack-kuryr | 23:28 | |
*** limao has quit IRC | 23:29 | |
*** aojea has quit IRC | 23:33 | |
*** hongbin has quit IRC | 23:34 | |
*** aojea has joined #openstack-kuryr | 23:37 | |
*** aojea has quit IRC | 23:41 | |
*** aojea has joined #openstack-kuryr | 23:46 | |
*** aojea has quit IRC | 23:51 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!