openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: devstack: allow overriding keystone config https://review.openstack.org/487517 | 00:25 |
---|---|---|
openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: devstack: do not delete service ipallocation pool https://review.openstack.org/487518 | 00:25 |
*** kiennt has joined #openstack-kuryr | 00:42 | |
*** limao has joined #openstack-kuryr | 00:46 | |
*** caowei has joined #openstack-kuryr | 01:04 | |
openstackgerrit | Liping Mao proposed openstack/kuryr-libnetwork master: Fix Rally test case error. https://review.openstack.org/487476 | 01:14 |
*** yedongcan has joined #openstack-kuryr | 01:18 | |
*** yedongcan has quit IRC | 01:23 | |
*** yedongcan has joined #openstack-kuryr | 01:23 | |
*** yamamoto_ has quit IRC | 01:31 | |
*** yamamoto has joined #openstack-kuryr | 01:35 | |
*** tonanhngo has joined #openstack-kuryr | 01:48 | |
*** limao has quit IRC | 02:00 | |
*** limao_ has joined #openstack-kuryr | 02:00 | |
*** limao_ has quit IRC | 02:19 | |
*** limao has joined #openstack-kuryr | 02:20 | |
*** limao has quit IRC | 02:50 | |
*** limao has joined #openstack-kuryr | 02:51 | |
openstackgerrit | Jaivish Kothari(janonymous) proposed openstack/kuryr-kubernetes master: Systemd usage of kubernetes startup https://review.openstack.org/484754 | 03:00 |
*** rwallner has joined #openstack-kuryr | 03:16 | |
*** rwallner has quit IRC | 03:20 | |
*** aojea has joined #openstack-kuryr | 03:23 | |
*** limao has quit IRC | 03:26 | |
*** aojea has quit IRC | 03:28 | |
vikasc | irenab, hi | 03:36 |
vikasc | irenab, did you notice apuimedo's reply to your comments on network-addon patch? | 03:37 |
openstackgerrit | vikas choudhary proposed openstack/kuryr-kubernetes master: Add support to install Kuryr as a network addon https://review.openstack.org/466675 | 03:45 |
*** longfei_zhang has joined #openstack-kuryr | 04:06 | |
longfei_zhang | irenab, Kuryr support that set the fix clusterIP in service yaml file, i have tested this | 04:09 |
longfei_zhang | the previous error war due to the wrong clusterIP in yaml file | 04:10 |
openstackgerrit | Merged openstack/kuryr-kubernetes master: devstack: Move to devstack-plugin-container https://review.openstack.org/474238 | 04:11 |
*** limao has joined #openstack-kuryr | 04:30 | |
*** gouthamr has quit IRC | 04:38 | |
*** tonanhngo has quit IRC | 04:44 | |
*** limao has quit IRC | 04:47 | |
*** limao has joined #openstack-kuryr | 04:59 | |
*** caowei has quit IRC | 05:06 | |
*** hyunsun has joined #openstack-kuryr | 05:11 | |
*** aojea has joined #openstack-kuryr | 05:31 | |
*** kiennt has quit IRC | 05:44 | |
*** hongbin has joined #openstack-kuryr | 05:47 | |
*** janki has joined #openstack-kuryr | 05:50 | |
*** yuanying has joined #openstack-kuryr | 05:57 | |
*** caowei has joined #openstack-kuryr | 05:58 | |
*** yboaron_ has joined #openstack-kuryr | 06:01 | |
*** kiennt has joined #openstack-kuryr | 06:02 | |
*** kiennt has quit IRC | 06:16 | |
*** kiennt_ has joined #openstack-kuryr | 06:17 | |
*** kiennt has joined #openstack-kuryr | 06:17 | |
*** kiennt has quit IRC | 06:17 | |
*** kzaitsev_ws has joined #openstack-kuryr | 06:33 | |
*** tonanhngo has joined #openstack-kuryr | 06:34 | |
*** tonanhngo has quit IRC | 06:39 | |
*** kzaitsev_ws has quit IRC | 06:56 | |
*** hongbin has quit IRC | 07:04 | |
*** aojea_ has joined #openstack-kuryr | 07:08 | |
*** aojea has quit IRC | 07:12 | |
ltomasbo | hi apuimedo | 07:32 |
ltomasbo | I complete the installation without problems, and the information is right at the nested kuryr.conf | 07:33 |
ltomasbo | however, it is not working, and I'm not sure if it is due to the different project | 07:33 |
ltomasbo | http://paste.openstack.org/show/616666/ | 07:37 |
*** yboaron_ has quit IRC | 07:38 | |
ltomasbo | umm, weird: openstack port create --network k8s-pod-net --project k8s test | 07:39 |
ltomasbo | HttpException: Conflict (HTTP 409) (Request-ID: req-b97a79e6-b90d-44ea-88a3-d69c639536ac), No more IP addresses available on network 9dcaeaf9-2d54-439d-a45a-97e996a24b32. | 07:39 |
*** kural_ has joined #openstack-kuryr | 07:43 | |
*** kural has joined #openstack-kuryr | 07:43 | |
*** kural_ has quit IRC | 07:46 | |
*** kural has quit IRC | 07:46 | |
*** egonzalez has joined #openstack-kuryr | 07:48 | |
*** roeyc has joined #openstack-kuryr | 07:49 | |
*** garyloug has joined #openstack-kuryr | 08:04 | |
*** tonanhngo has joined #openstack-kuryr | 08:15 | |
*** tonanhngo has quit IRC | 08:19 | |
*** longfei_zhang has quit IRC | 08:21 | |
*** kural has joined #openstack-kuryr | 08:21 | |
*** yboaron_ has joined #openstack-kuryr | 08:33 | |
openstackgerrit | Daniel Mellado proposed openstack/kuryr-kubernetes master: DNM Infra test https://review.openstack.org/487762 | 08:49 |
*** hyunsun has quit IRC | 08:50 | |
*** kzaitsev_ws has joined #openstack-kuryr | 08:51 | |
kzaitsev_ws | hm, I wonder if we need a release for kuryr k8s for p-3 | 08:52 |
kzaitsev_ws | kuryr-k8s is cycle with intermediary, right? | 08:53 |
kzaitsev_ws | ok then we need to make a release by August 21st that will be the cycle release | 08:55 |
apuimedo | ltomasbo: ok. I know about this error ltomasbo | 08:55 |
apuimedo | ltomasbo: I'll upload a fix in a moment | 08:55 |
apuimedo | kzaitsev_ws: yes we do | 08:55 |
kzaitsev_ws | was just wrapping up murano release tasks. | 09:03 |
irenab | vikasc: I am going to verify https://review.openstack.org/#/c/466675/. Any advice on the steps? | 09:03 |
kzaitsev_ws | and thought to check | 09:03 |
openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: devstack: Use separate nets and project for kuryr https://review.openstack.org/486134 | 09:04 |
apuimedo | ltomasbo: here you have your fix ^^ | 09:04 |
vikasc | irenab, you will have to change configmap values for kuryr config, besides i think README instructions should be enough build the images | 09:05 |
apuimedo | vikasc: they are enough | 09:06 |
apuimedo | vikasc: but no need to build | 09:06 |
apuimedo | I already built and pushed them to dockerhub | 09:06 |
vikasc | yeah she can use the docker images | 09:06 |
apuimedo | so she can just do docker pull for them | 09:06 |
vikasc | yep | 09:06 |
vikasc | irenab, | 09:06 |
apuimedo | vikasc: dmellado: did you see about my idea to have a gate that builds the images and puts them to tarballs.openstack.org | 09:07 |
apuimedo | so other gates can use them? | 09:07 |
vikasc | yep, sounds good to me | 09:08 |
kzaitsev_ws | apuimedo: there was a lengthy discussion about why kolla shouldn't do exactly that | 09:08 |
kzaitsev_ws | on ML | 09:08 |
kzaitsev_ws | I don't remember what was the conclusion though | 09:08 |
apuimedo | kzaitsev_ws: but that's what they do right? | 09:09 |
apuimedo | they were my inspiration | 09:10 |
apuimedo | look, all I want is a darned infra docker registry | 09:10 |
apuimedo | but this is taking longer than Catalan independence already | 09:10 |
kzaitsev_ws | http://lists.openstack.org/pipermail/openstack-dev/2017-May/116677.html | 09:11 |
* kzaitsev_ws afk will be back in 15 mins | 09:12 | |
apuimedo | thanks kzaitsev_ws | 09:12 |
* apuimedo should read ml more often | 09:13 | |
*** longfei_zhang has joined #openstack-kuryr | 09:15 | |
*** kiennt_ is now known as kiennt | 09:15 | |
*** limao has quit IRC | 09:21 | |
*** limao has joined #openstack-kuryr | 09:22 | |
*** limao has quit IRC | 09:26 | |
longfei_zhang | Hi I have a question , can we set the pod_subnet and service_subnet to the same subnet? | 09:27 |
apuimedo | longfei_zhang: why do you want that? | 09:27 |
longfei_zhang | just curious | 09:28 |
longfei_zhang | if so then we will face conflict ip address? | 09:28 |
longfei_zhang | k8s pass the service cluster ip to kuryr , maybe this ip already used by pod , right? | 09:28 |
longfei_zhang | if they are using the same subnet | 09:29 |
apuimedo | longfei_zhang: I'm not surprised you are curious. It is an interesting proposition | 09:31 |
apuimedo | the conflict would come if and when Kubernetes would decide on a clusterIP that is already taken by the pods | 09:31 |
apuimedo | I don't think kubernetes IPAM looks at pod IPs to see which clusterIPs are free | 09:31 |
longfei_zhang | apuimedo, yes | 09:32 |
apuimedo | IIRC kubernetes assumes they are different allocation ranges | 09:32 |
longfei_zhang | ok | 09:32 |
apuimedo | ltomasbo: it passed ;-) https://review.openstack.org/#/c/486134/ | 09:32 |
dmellado | apuimedo: I didn't | 09:33 |
dmellado | do you have any email/blueprint about that? | 09:33 |
apuimedo | dmellado: nah... | 09:33 |
apuimedo | It was something I wrote yesterday on IRC | 09:33 |
longfei_zhang | apuimedo, currently the lb's performance is not good, so if they are in different subnets, they have to go thorugh the route | 09:34 |
dmellado | apuimedo: they're decommissioning my bouncer | 09:34 |
* dmellado cries | 09:34 | |
ltomasbo | apuimedo, going to test it now | 09:34 |
longfei_zhang | if they can use the same subnet ,then performance will be better | 09:34 |
ltomasbo | apuimedo, we should also remove all that stuff at the unstack (as irenab suggested) | 09:34 |
irenab | longfei_zhang: I am fraid this is not properly supported by kuryr | 09:37 |
apuimedo | dmellado: I can rent you an account in my home bouncer | 09:38 |
longfei_zhang | irenab, yes | 09:38 |
dmellado | apuimedo: does it include internal access? | 09:38 |
apuimedo | it costs one arbitrary patch a month | 09:38 |
apuimedo | dmellado: of course it does | 09:38 |
dmellado | apuimedo: I'll hook my znc onto yours | 09:39 |
dmellado | xD | 09:39 |
apuimedo | 99.5% availability | 09:39 |
apuimedo | ltomasbo: which stuff? | 09:40 |
apuimedo | irenab: longfei_zhang: I'm not even sure it could be supported | 09:40 |
apuimedo | longfei_zhang: I think the best bet would be to use a neutron plugin that doesn't go all the way to the router, like ovn or dragonflow | 09:40 |
apuimedo | irenab: longfei_zhang: do you have some idea how it could be done in the same subnet? | 09:41 |
apuimedo | mmmm | 09:41 |
apuimedo | moment, let me think | 09:41 |
apuimedo | ok, I haven't tried this | 09:41 |
irenab | from the admin perspective, I think it makes sense to have sparate subnets (or at least ranges ) for Pod sand Services | 09:41 |
apuimedo | but probably it would be possible to give kubernetes a subnet of the real subnet | 09:41 |
apuimedo | let's say that you have a /22 | 09:42 |
apuimedo | so you tell kubernetes API server that the service subnet is a /24 of that | 09:42 |
apuimedo | then you modify the pod subnet in neutron so that its allocation pools don't include that /24 | 09:43 |
apuimedo | and you tell kuryr-kubernetes that both the pod subnet and the service subnet are the same | 09:43 |
apuimedo | longfei_zhang: irenab: I think this could work | 09:43 |
openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: devstack: allow overriding keystone config https://review.openstack.org/487517 | 09:43 |
kzaitsev_ws | irenab: longfei_zhang: I believe k8s doesn't check anything but already allocated ips when choosing an ip for a new service | 09:44 |
vikasc | irenab, addressed your comment about adding reference to dockerhub images in README. PTAL | 09:44 |
kzaitsev_ws | ah | 09:44 |
kzaitsev_ws | yeah. probably using same subnet, different ranges might work | 09:44 |
irenab | kzaitsev_ws: correct | 09:44 |
apuimedo | longfei_zhang: It would be interesting if you can try this | 09:45 |
apuimedo | and report back | 09:45 |
apuimedo | maybe send a doc patch explaining this deployment variation | 09:45 |
longfei_zhang | apuimedo, ok I will try this | 09:48 |
vikasc | irenab, do you want me to move instructions from README to a doc, "containerized-installation.rst" in /source/installation? or you meant something else. | 09:48 |
apuimedo | thanks longfei_zhang. I'll eagerly await the result of the investigation | 09:49 |
irenab | vikasc: I think this is a good idea, it is about installation mainly | 09:49 |
apuimedo | vikasc: I think that's a good idea | 09:49 |
ltomasbo | apuimedo, the lbaas, subnet and networks created | 09:49 |
apuimedo | ltomasbo: good | 09:49 |
vikasc | ok, going to update | 09:49 |
apuimedo | thanks | 09:52 |
apuimedo | ltomasbo: so, can you create the ports now? | 09:52 |
irenab | vikasc: I think with devstack support, the kuryr.yml can be updated automatically | 09:53 |
apuimedo | irenab: yes. But do we want a devstack that build the images and then uses them? | 09:53 |
vikasc | irenab, yes | 09:53 |
vikasc | apuimedo, cant devstack have option to either build or pull from dockerhub? | 09:54 |
irenab | apuimedo: I think so, maybe it can use latest image based on some config in devstack | 09:54 |
irenab | vikasc: :-) | 09:54 |
irenab | eventually we will have a book on varios options in devstack | 09:55 |
vikasc | :D | 09:55 |
apuimedo | vikasc: sure it can | 09:55 |
apuimedo | vikasc: irenab: but please, not on this patch | 09:55 |
apuimedo | let's put it as a follow-up | 09:56 |
apuimedo | I don't want it any bigger | 09:56 |
apuimedo | irenab: I haven't forgotten about your worker node only multi node devstack | 09:57 |
apuimedo | I have it open and I'll get to it | 09:57 |
apuimedo | I hope soon I can stop doing devstack patches | 09:57 |
apuimedo | cause bash and I don't like each other | 09:57 |
irenab | apuimedo: I think dmellado already picked this one | 09:57 |
apuimedo | dmellado: you did? | 09:58 |
* dmellado gets into context | 09:58 | |
irenab | I did some ugly changes locally based on ltomasbo advice and got it run | 09:58 |
dmellado | oh, that one, yes | 09:58 |
dmellado | I still didn't get the time to work on that | 09:58 |
dmellado | so if you've a working patch before me don't hesitate to push it there | 09:58 |
* dmellado into endless meetings today... | 09:59 | |
*** yamamoto has quit IRC | 10:00 | |
ltomasbo | irenab, great! (and I agree they are ugly! At least the ones I did) | 10:00 |
irenab | ltomasbo: I totally ruined the one with kuryr-kubernetes enabled, but for worker node only it seems to work | 10:01 |
ltomasbo | :D | 10:02 |
ltomasbo | apuimedo, re-stacking | 10:02 |
ltomasbo | apuimedo, it was easy than remove the networks, that also were failing due to some ports bounds | 10:02 |
apuimedo | :-) | 10:05 |
irenab | apuimedo: vikasc : what should be set in kuryr.yml for the images from dockerhub? | 10:09 |
openstackgerrit | Merged openstack/kuryr-kubernetes master: Systemd usage of kubernetes startup https://review.openstack.org/484754 | 10:10 |
vikasc | irenab, in that case, imagePullPolicy: Never, will have to removed i guess | 10:10 |
apuimedo | irenab: vikasc: can I get https://review.openstack.org/#/c/486134/9 merged? | 10:11 |
irenab | apuimedo: this one is next after vikasc patch verifucation | 10:11 |
irenab | give 1 hour at most | 10:11 |
apuimedo | irenab: ok. it is exercised by devstack gate, so I think that for this one the gate can be trusted | 10:12 |
irenab | apuimedo: great, so will just do a quick overview | 10:13 |
vikasc | apuimedo, will take a look after updating updating the changes to move instaructions from readme | 10:13 |
apuimedo | thanks irenab | 10:19 |
apuimedo | vikasc: very well | 10:19 |
ltomasbo | apuimedo, I'm testing it right now | 10:21 |
openstackgerrit | Hyunsun Moon proposed openstack/kuryr-libnetwork master: Add new network ID pattern https://review.openstack.org/487802 | 10:21 |
ltomasbo | I'll give the +1 as soon as it finished for my nested deployment | 10:21 |
apuimedo | ltomasbo: good | 10:21 |
*** tonanhngo has joined #openstack-kuryr | 10:21 | |
ltomasbo | apuimedo, should be done in 5 minutes or so | 10:21 |
apuimedo | great, thanks | 10:21 |
*** tonanhngo has quit IRC | 10:26 | |
openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: devstack: allow overriding keystone config https://review.openstack.org/487517 | 10:27 |
*** yamamoto has joined #openstack-kuryr | 10:31 | |
openstackgerrit | vikas choudhary proposed openstack/kuryr-kubernetes master: Add support to install Kuryr as a network addon https://review.openstack.org/466675 | 10:32 |
openstackgerrit | vikas choudhary proposed openstack/kuryr-kubernetes master: Add support to install Kuryr as a network addon https://review.openstack.org/466675 | 10:34 |
vikasc | irenab, moved instructions from readme to containerized.rst ^ | 10:36 |
irenab | vikasc: checking | 10:36 |
vikasc | thanks | 10:36 |
irenab | vikasc: with devstack, do I need to stop kuryr-kubernetes and kubelet before or its ok after the kuryr.yml deployment? | 10:41 |
vikasc | irenab, you will need to add subnet id etc in kuryr.yml, which will available only after devstack has run. | 10:42 |
vikasc | irenab, or may be i misread your question | 10:43 |
apuimedo | irenab: before | 10:44 |
vikasc | irenab, if thats the case , can you please reword your query | 10:44 |
irenab | apuimedo: thanks, since I forgot and now need to restack to verify … | 10:44 |
*** rwallner has joined #openstack-kuryr | 10:44 | |
apuimedo | why? | 10:45 |
*** rwallner has quit IRC | 10:45 | |
apuimedo | irenab: ^^ | 10:45 |
irenab | pods just stuck in the pending | 10:45 |
*** rwallner has joined #openstack-kuryr | 10:45 | |
irenab | do not know why, let me do it properly | 10:45 |
apuimedo | ok | 10:45 |
irenab | 1. devstack 2. set kuryr.yml with config 3. stop kubelet and kuryr-k8s services 4. kubectl apply -f kuryr,yml. correct? | 10:46 |
vikasc | correct | 10:46 |
vikasc | irenab, or just dont enable kuryr-kubernetes plugin in devstack localrc | 10:47 |
irenab | vikasc: how would I get the config for the yml then? | 10:47 |
vikasc | irenab, i just set default subnet id and demo project ids in configmap | 10:48 |
irenab | vikasc: hm. If the flow I suggested won’t work, will do your way | 10:49 |
apuimedo | irenab: if you don't disable them, it may be necessary to disable kuryr cni (so they don't get stuck on pending) | 10:49 |
apuimedo | is that right, vikasc ? | 10:49 |
apuimedo | or do we have cni code that skips host networking pods? | 10:49 |
irenab | apuimedo: vikasc : open for any verification flow you suggest as long as its clear | 10:50 |
vikasc | apuimedo, hostnetworking pod, i.e cni daemonset pod, dont need cni plugin | 10:52 |
apuimedo | vikasc: is the controller the same? | 10:53 |
vikasc | irenab, which pod was in pending state, controller or cni or the application pods? | 10:53 |
vikasc | apuimedo, yes | 10:53 |
vikasc | apuimedo, controller container also host networking | 10:53 |
irenab | apuimedo: vikasc : now at step 3 | 10:54 |
irenab | currently both are pending | 10:55 |
vikasc | are you able launch a sample pod? | 10:56 |
vikasc | just to see if NoSchedule taint is not there | 10:56 |
irenab | I stopped kuryr and kubelet, so won’t work | 10:56 |
irenab | brb | 10:56 |
vikasc | kubelet you should not stop | 10:56 |
vikasc | irenab, | 10:56 |
vikasc | irenab, i am assuming that you have installed k8s using kubeadm | 10:57 |
apuimedo | vikasc: no, she said devstack | 10:58 |
apuimedo | so hyperkube | 10:58 |
vikasc | is she is stopping kubelet how pods will run | 10:59 |
vikasc | apuimedo, is it expected that this patch will work on a non-kubeadm installed deployment also | 11:00 |
apuimedo | vikasc: yeah. kubelet should not be stopped | 11:00 |
apuimedo | only kuryr things should be stopped | 11:01 |
vikasc | apuimedo, let me check network-addon functionality docs to verify it is supported only with kubeadm or with all schemes | 11:01 |
apuimedo | vikasc: any recent kubernetes deployment should be game | 11:01 |
apuimedo | and by game I meant that we'd accept patches to increase compatibility | 11:01 |
vikasc | apuimedo, current patch should work on devstack installed k8s as well. | 11:04 |
apuimedo | yes, I think so | 11:04 |
vikasc | irenab, can you please just stop kuryr-k8s and not kubelet | 11:05 |
vikasc | irenab, also please delete devstack installed cni binary from /opt/cni/bin | 11:06 |
vikasc | apuimedo, ^ make sense? | 11:06 |
vikasc | and then move on to step 4. | 11:06 |
irenab | back | 11:08 |
irenab | vikasc: apuimedo : seems it just cannot get the image | 11:09 |
apuimedo | irenab: do a "docker pull" | 11:10 |
apuimedo | and then tag them with docket tag | 11:10 |
apuimedo | as kuryr/cni:latest and kuryr/controller:latest | 11:10 |
irenab | not the tag in the dockerhub? | 11:11 |
irenab | I modified the yml with it | 11:11 |
apuimedo | irenab: paste the yml | 11:11 |
apuimedo | pleas | 11:11 |
apuimedo | *please | 11:11 |
irenab | a sec | 11:12 |
vikasc | irenab, can you please try with imagePullPolicy: IfNotPresent | 11:13 |
irenab | https://pastebin.com/sixBYmaF | 11:15 |
irenab | trying with the pull | 11:16 |
vikasc | irenab, "imagePullPolicy: IfNotPresent" or "imagePullPolicy: Always" should also work. | 11:19 |
irenab | vikasc: gonna try it now | 11:19 |
vikasc | irenab, thanks and sorry for wrong information to delete imagePullPolicy earlier. | 11:20 |
irenab | vikasc: please add the details about cni removal and kuryr service stop somewhere | 11:22 |
apuimedo | kzaitsev_ws: ltomasbo: do you see the mistake in https://review.openstack.org/#/c/487517/4 ? | 11:23 |
apuimedo | I'm a bit puzzled | 11:23 |
apuimedo | (probably as a result of staying with devstack until 3am) | 11:23 |
apuimedo | :P | 11:23 |
kzaitsev_ws | apuimedo: looking at it right now ) | 11:23 |
apuimedo | I just don't see that trailing backslash | 11:24 |
apuimedo | fscking hell | 11:24 |
vikasc | irenab, sure. you got it working? | 11:25 |
apuimedo | line 194 of devstack/lib/kuryr_kubernetes | 11:25 |
irenab | vikasc: not yet | 11:26 |
vikasc | irenab, still cni and controller pods in pending state? | 11:27 |
irenab | no, some error in yml to json | 11:27 |
openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: devstack: allow overriding keystone config https://review.openstack.org/487517 | 11:28 |
kzaitsev_ws | apuimedo: found? =) | 11:29 |
apuimedo | kzaitsev_ws: nah... | 11:30 |
apuimedo | those are cosmetic things I improved | 11:30 |
kzaitsev_ws | ok then I think I found it =) | 11:30 |
apuimedo | ok, let's try | 11:31 |
kzaitsev_ws | $ openstack --help | grep project-id | 11:31 |
kzaitsev_ws | [--os-project-id <auth-project-id>] | 11:31 |
apuimedo | kzaitsev_ws: right | 11:32 |
apuimedo | the error message seems to point to something else though | 11:32 |
apuimedo | I wonder if it's just poor error handling | 11:32 |
openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: devstack: allow overriding keystone config https://review.openstack.org/487517 | 11:32 |
vikasc | irenab, i am trying again locally to verify any kuryr.yml related error | 11:33 |
irenab | ok | 11:33 |
kzaitsev_ws | http://logs.openstack.org/17/487517/4/check/gate-install-dsvm-default-kuryr-kubernetes/aa21237/logs/devstacklog.txt.gz#_2017-07-27_10_42_22_244 | 11:34 |
kzaitsev_ws | although yes | 11:34 |
kzaitsev_ws | I'm perplexed by how it' | 11:34 |
kzaitsev_ws | s showing the whole line instead of just the | 11:35 |
kzaitsev_ws | actual error. (having trouble printing straight) | 11:35 |
kzaitsev_ws | s/printing/typing/ | 11:35 |
vikasc | irenab, it ran fine without any yml to json related error | 11:35 |
irenab | I think I got rid of it now | 11:36 |
irenab | pods are running | 11:36 |
irenab | now going to try to deploy some app and see that it works | 11:36 |
irenab | vikasc: what do you usually use to verify? | 11:37 |
apuimedo | :-) | 11:38 |
openstackgerrit | Merged openstack/kuryr-kubernetes master: devstack: Use separate nets and project for kuryr https://review.openstack.org/486134 | 11:39 |
irenab | vikasc: another question. any reason to use yml extension and not yaml? | 11:39 |
vikasc | irenab, no specific reason | 11:40 |
irenab | in k8s they use yaml all over in the examples | 11:40 |
vikasc | irenab, i am naive and dont understand difference between the two | 11:40 |
*** yboaron_ has quit IRC | 11:41 | |
irenab | vikasc: seems it has almost religion involved :-) https://stackoverflow.com/questions/22268952/what-is-the-difference-between-yaml-vs-yml-extension | 11:41 |
vikasc | :D | 11:41 |
vikasc | i was always using both interchangebly | 11:42 |
vikasc | irenab, | 11:43 |
vikasc | kuryr-kubernetes]# cat ~/demo1.yml | 11:43 |
vikasc | apiVersion: v1 | 11:43 |
vikasc | kind: Pod | 11:43 |
vikasc | metadata: | 11:43 |
vikasc | name: my.pod1 | 11:43 |
vikasc | spec: | 11:43 |
vikasc | containers: | 11:43 |
vikasc | - name: demo1 | 11:43 |
vikasc | image: nginx:latest | 11:43 |
kzaitsev_ws | seems like jpeg vs jpg... | 11:44 |
irenab | vikasc: I tryed the guestbook-all-in-one k8s example, all pods are in Container create | 11:44 |
apuimedo | kzaitsev_ws: it is yml vs jpeg | 11:44 |
vikasc | i just run this test pod to see if it gets ip fromneutron | 11:44 |
apuimedo | xD | 11:44 |
kzaitsev_ws | the only problem with nginx is that it doesn't have ip/ifconfig nothing, so I use 2-container pods with busybox =) | 11:44 |
irenab | network: failed to find plugin \"kuryr-cni\" in path [/opt/stack/cni/bin /opt/kuryr-cni/bin]" | 11:46 |
irenab | vikasc: this is what I get in the pod events | 11:46 |
irenab | any idea what I did wrong? | 11:47 |
vikasc | irenab, if cni daemonset pod is running fine, it should have copied kuryr-cni at /opt/cni/bin | 11:47 |
irenab | with devstack we put it in the /opt/stack/cni/bin | 11:48 |
irenab | vikasc: I see it in the path you posted | 11:48 |
*** apuimedo_ has joined #openstack-kuryr | 11:49 | |
vikasc | irenab, if you see in the error, kubelet is not searching in correct path | 11:49 |
irenab | #!/bin/bash | 11:49 |
irenab | export PBR_VERSION='0.1.1.dev26' | 11:49 |
vikasc | irenab, cni directory path is passed to kubelet | 11:49 |
*** apuimedo has quit IRC | 11:49 | |
*** apuimedo_ is now known as apuimedo | 11:49 | |
vikasc | irenab, you can do a "ps ax | grep kubelet" to see what path is being passed to kubelet for cni location | 11:50 |
kzaitsev_ws | apuimedo: still failing, but we caught the right error there http://logs.openstack.org/17/487517/6/check/gate-install-dsvm-default-kuryr-kubernetes/fb19cbf/logs/devstacklog.txt.gz#_2017-07-27_11_40_22_527 | 11:51 |
vikasc | irenab, two solutions, either pass /opt/cni/bin to kubelet or copy contents of /opt/cni/bin/ to /opt/kuryr-cni/bin | 11:51 |
irenab | --cni-bin-dir=/opt/stack/cni/bin --cni-conf-dir=/opt/stack/cni/conf | 11:51 |
vikasc | yeah | 11:52 |
vikasc | thats the inconsistency between paths | 11:52 |
irenab | vikasc: I just restarted the kubelet service after the kuryr.yml was done | 11:52 |
vikasc | systemctl restart kubelet? | 11:53 |
vikasc | irenab, | 11:53 |
irenab | devstack@kubelet.service | 11:53 |
vikasc | ok | 11:53 |
*** kural has quit IRC | 11:54 | |
vikasc | you can change cni dir path from within kuryr devstack plugin, i think | 11:54 |
kzaitsev_ws | apuimedo: I think we sould add a user role to the newly created project out there | 11:54 |
openstackgerrit | Hyunsun Moon proposed openstack/kuryr-libnetwork master: Add new network ID pattern https://review.openstack.org/487802 | 11:57 |
vikasc | irenab, worked? | 12:05 |
*** caowei has quit IRC | 12:05 | |
irenab | vikasc: no | 12:06 |
vikasc | irenab, what error now | 12:06 |
irenab | can you try on your side with the devstack ? | 12:06 |
irenab | network: netplugin failed but error parsing its diagnostic message \"\": unexpected end of JSON input | 12:07 |
irenab | vikasc: how to see the logs from the kuryr-controller? | 12:09 |
vikasc | irenab, i use devstack just to install neutron and keystone and install kubernetes using kubeadm. This patch does not claim to be supporting kuryr devstack plugin. I can try though. | 12:10 |
vikasc | irenab, kubectl logs <controller-pod-name> | 12:10 |
*** kiennt has quit IRC | 12:11 | |
irenab | vikasc: I think maybe we need more complete installation guide, or just list of prerequisite | 12:11 |
apuimedo | vikasc: is the cni conf specifying use_stderr? | 12:12 |
vikasc | apuimedo, it must. | 12:13 |
vikasc | apuimedo, cp /opt/kuryr-kubernetes/etc/cni/net.d/* "$CNI_CONFIG_DIR_PATH" | 12:13 |
vikasc | apuimedo, are you suggesting to make this change | 12:13 |
apuimedo | vikasc: as a follow-up | 12:13 |
apuimedo | kzaitsev_ws: you were right | 12:14 |
apuimedo | that was the problem | 12:14 |
vikasc | apuimedo, sure | 12:14 |
*** yamamoto has quit IRC | 12:14 | |
apuimedo | funnily enough, now it tells me that there's some auth issue | 12:14 |
apuimedo | http://logs.openstack.org/17/487517/6/check/gate-install-dsvm-default-kuryr-kubernetes/fb19cbf/logs/devstacklog.txt.gz | 12:14 |
apuimedo | :/ | 12:14 |
*** atoth has joined #openstack-kuryr | 12:15 | |
vikasc | irenab, use_stderr = true, must be there in cni config map | 12:15 |
irenab | vikasc: it is there | 12:16 |
apuimedo | irenab: I answered your question on vikas' patch | 12:17 |
vikasc | irenab, if i mention in the containerized.rst that currently kuryr devstack plugin is not supported, and steps on how to install k8s, devstack and kuryr, will that be fine | 12:18 |
apuimedo | vikasc: fine to me | 12:19 |
irenab | vikasc: I would prefer to have it working for devs as well | 12:21 |
irenab | vikasc: if I understand correctly what you are saying, it should be fine | 12:22 |
apuimedo | irenab: is it fine for you if I add it as a follow-up patch now and we verify the follow-up patch? | 12:22 |
apuimedo | and then merge both together? | 12:22 |
irenab | apuimedo: I am just not sure how to verify the current patch | 12:22 |
irenab | how did you do it? | 12:23 |
vikasc | 1. run devstack without kuryr plugin | 12:23 |
vikasc | 2. install k8s using kubeadm | 12:23 |
vikasc | 3. update kuryr.yml configmaps | 12:24 |
vikasc | 4. kubectl apply -f kuryr.yml | 12:24 |
vikasc | irenab, ^ | 12:24 |
irenab | vikasc: do you have instructions for step 2? | 12:25 |
vikasc | irenab, sure | 12:25 |
vikasc | irenab, this is the simplest way to install k8s, wait a moment | 12:25 |
irenab | apuimedo: I recall we planned to have kubeadm replace hyperkube in devstack. Is it still omn the map? | 12:26 |
vikasc | irenab, https://kubernetes.io/docs/setup/independent/create-cluster-kubeadm/ | 12:26 |
vikasc | irenab, to summarize, install kubeadm, https://kubernetes.io/docs/setup/independent/install-kubeadm/ | 12:27 |
vikasc | and then kubeadm init | 12:27 |
irenab | vikasc: for step 1; do you have local.conf? | 12:29 |
vikasc | yes | 12:29 |
vikasc | irenab, wait, will share paste | 12:30 |
apuimedo | irenab: it is on the roadmap for p3 | 12:31 |
apuimedo | but it is pending on this patch from vikas, actually :P | 12:31 |
irenab | :-) | 12:32 |
apuimedo | irenab: also on other things | 12:33 |
apuimedo | it is not well documented how to point it to an existing etcd, for example | 12:33 |
apuimedo | I researched it a bit | 12:33 |
apuimedo | but I had to prioritize the octavia stuff first | 12:33 |
apuimedo | since what we currently have, works | 12:33 |
irenab | apuimedo: I would prefer to have dev andproducton envs as similar as possible | 12:34 |
apuimedo | irenab: I agree | 12:34 |
apuimedo | it's just harder to justify the time spent | 12:34 |
apuimedo | when replacing something that works | 12:34 |
vikasc | irenab, localrc http://paste.openstack.org/show/616714/ | 12:35 |
*** kural has joined #openstack-kuryr | 12:35 | |
vikasc | irenab, please note that at the end of file, kubelet and kuryr related lines are commented | 12:36 |
*** apuimedo has quit IRC | 12:42 | |
*** apuimedo has joined #openstack-kuryr | 12:42 | |
irenab | apuimedo: yea | 12:44 |
*** alraddarla has joined #openstack-kuryr | 12:46 | |
irenab | vikasc: sorry, it takes me time to verify, hope to complete it today | 12:53 |
vikasc | irenab, no problem. I will hold on and see if get more suggestions from you before updating the patch | 12:54 |
apuimedo | irenab: keep pinging if you encounter issues | 12:59 |
irenab | apuimedo: sure | 13:02 |
apuimedo | irenab: I'll make the follow-up patch that adds support for it in devstack | 13:02 |
irenab | apuimedo: great, thanks | 13:02 |
irenab | not having kuryr services running as k8s elements is really bad | 13:03 |
apuimedo | irenab: hence why I'm so eager to merge this | 13:05 |
apuimedo | xD | 13:05 |
irenab | apuimedo: my first prio for the rest of theday | 13:05 |
apuimedo | irenab: thanks a lot | 13:07 |
apuimedo | your verification is improving the quality | 13:08 |
irenab | I hope so | 13:10 |
*** yedongcan has left #openstack-kuryr | 13:10 | |
*** yamamoto has joined #openstack-kuryr | 13:14 | |
*** yamamoto has quit IRC | 13:21 | |
*** gouthamr has joined #openstack-kuryr | 13:31 | |
irenab | vikasc: what about docker? | 13:34 |
*** kural has quit IRC | 13:36 | |
*** kural has joined #openstack-kuryr | 13:36 | |
apuimedo | irenab: what about it? | 13:38 |
irenab | apuimedo: it was commented in the local.conf that vikasc posted. I wonder if he installed it manually since it seems not be part of the kubeadm install | 13:39 |
irenab | kubeadm init complains about not having it | 13:40 |
apuimedo | irenab: you know what. I start the patch | 13:44 |
apuimedo | for devstack integration | 13:44 |
apuimedo | I'll rebase vikasc patch on master as well | 13:44 |
*** yboaron_ has joined #openstack-kuryr | 13:45 | |
irenab | apuimedo: I am lamost done deploying the env. but I think we need some sort of the policy to have proper info for testing if the patch does not include testing | 13:45 |
apuimedo | irenab: I agree | 13:46 |
* irenab starting to master devstack a little | 13:46 | |
apuimedo | but yeah, install docker <=1.12.1 | 13:46 |
apuimedo | irenab: if you say such things I'll assign you the devstack patches. I'd be happy not to have to do them XD | 13:46 |
irenab | apuimedo: heh, mostly on user side | 13:47 |
apuimedo | pity | 13:47 |
apuimedo | I'm so sick of devstack | 13:47 |
apuimedo | I wish I could replace it with ansible | 13:47 |
irenab | i can imagine | 13:47 |
*** roeyc has quit IRC | 13:49 | |
*** ashishb_ has joined #openstack-kuryr | 14:01 | |
*** ashishb has quit IRC | 14:04 | |
*** janki has quit IRC | 14:14 | |
apuimedo | dmellado: do you know how to refer to the devstack directory from plugin.sh ? | 14:29 |
apuimedo | (as in, where the local.conf lives) | 14:29 |
dmellado | I'd say let's write ansiblestack | 14:29 |
dmellado | and drop devstack xD | 14:29 |
dmellado | I do *recall* that there were some var | 14:29 |
dmellado | let me check | 14:29 |
dmellado | oh, you mean the devstack dir where it was cloned | 14:31 |
dmellado | I was thinking about /opt/stack | 14:32 |
dmellado | which is DEST=${DEST:-/opt/stack} | 14:32 |
dmellado | apuimedo: from the stack.sh run | 14:33 |
dmellado | that dir is being kept as | 14:33 |
dmellado | 69 # Keep track of the DevStack directory | 14:34 |
dmellado | 70 TOP_DIR=$(cd $(dirname "$0") && pwd) | 14:34 |
dmellado | so depending on what you're trying to modify you should be able to get that var | 14:34 |
apuimedo | dmellado: I meant $TOP_DIR | 14:35 |
apuimedo | just found it now | 14:35 |
dmellado | apuimedo: please don't break devstack more than it already is xD | 14:35 |
apuimedo | dmellado: you've seen nothing | 14:36 |
apuimedo | It's going to make uruk hai look pretty | 14:36 |
openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Spec for adding OpenDaylight support in kuryr-k8s https://review.openstack.org/487885 | 14:56 |
*** kzaitsev_ws has quit IRC | 14:56 | |
openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Spec for adding OpenDaylight support in kuryr-k8s https://review.openstack.org/487885 | 15:01 |
openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Update documentation about nested deployment https://review.openstack.org/487906 | 15:26 |
apuimedo | irenab: ltomasbo: dmellado: Octavia is working! | 15:32 |
dmellado | apuimedo: \o/ | 15:33 |
apuimedo | http://paste.openstack.org/show/616753/ | 15:33 |
dmellado | pls send any needed patch and I'll update the gates when you're done ;) | 15:33 |
apuimedo | dmellado: it's already merged ;-) | 15:33 |
apuimedo | only the local.conf is missing | 15:33 |
dmellado | awesome, so we can modify that on the gates :P | 15:33 |
apuimedo | dmellado: I'd rather you create additional devstack and tempest gates for octavia | 15:34 |
dmellado | oh, so much work! xD | 15:34 |
apuimedo | until we deprecate and drop plain old neutron lbaasv2 | 15:34 |
dmellado | I'll create a new -octavia gate | 15:34 |
apuimedo | dmellado: come on, you can do it before going on holiday | 15:34 |
*** yboaron_ has quit IRC | 15:34 | |
dmellado | I deff can xD | 15:35 |
dmellado | but I'll need to delay my bp about moving everything to devstackgaterc and reorganize | 15:35 |
dmellado | xD | 15:35 |
ltomasbo | apuimedo, great! | 15:35 |
apuimedo | dmellado: http://paste.openstack.org/show/616754/ | 15:35 |
apuimedo | dmellado: is that a lot of work? | 15:35 |
dmellado | nah, but I'll need to sort a few things out | 15:36 |
ltomasbo | apuimedo, so, you need to enable q-lbaas-v2 and a few octavia services and pluggins (plus barbican) | 15:36 |
ltomasbo | and that is pretty much it? | 15:37 |
dmellado | apuimedo: are you enabling barbican for any reason? | 15:37 |
apuimedo | ltomasbo: and the placement API | 15:38 |
apuimedo | dmellado: it is used for the amphorae | 15:38 |
ltomasbo | apuimedo, placement API is always needed for VMs, isn't it? | 15:38 |
apuimedo | ltomasbo: I don't know :P | 15:41 |
apuimedo | xD | 15:41 |
ltomasbo | yep, I think they split the placement out from the scheduler a few months ago | 15:42 |
ltomasbo | anyway, great news! and nice to know that we only needed to change the local.conf! that is awesome! | 15:42 |
*** hongbin has joined #openstack-kuryr | 15:48 | |
apuimedo | ltomasbo: well, some devstack changes as well ;-) | 15:49 |
ltomasbo | :D | 15:49 |
*** egonzalez has quit IRC | 16:02 | |
dmellado | apuimedo: https://review.openstack.org/487928 | 16:08 |
dmellado | start preparing fuets | 16:08 |
apuimedo | dmellado: it failed :P | 16:13 |
dmellado | dammity | 16:14 |
dmellado | typo | 16:14 |
dmellado | xD | 16:14 |
apuimedo | :-) | 16:14 |
*** roeyc has joined #openstack-kuryr | 16:15 | |
apuimedo | meh | 16:15 |
apuimedo | I just noticed that I create an interface for the kubelet to be able to do probes | 16:15 |
apuimedo | but now it's in a different subnet and without a route to it | 16:15 |
* apuimedo facepalms | 16:16 | |
*** roeyc has quit IRC | 16:16 | |
dmellado | apuimedo: as long as you don't create a makefile I'm fine | 16:17 |
apuimedo | xD | 16:17 |
apuimedo | dmellado: you know that golang uses makefiles, right? | 16:17 |
dmellado | that's a different story | 16:18 |
dmellado | you know which ones I'm referring to | 16:18 |
dmellado | xD | 16:18 |
dmellado | don't play dumb xD | 16:18 |
*** kural has quit IRC | 16:18 | |
openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: devstack: create host interface to the pod subnet https://review.openstack.org/487935 | 16:24 |
apuimedo | dmellado: please, review ^^ | 16:24 |
apuimedo | irenab: if you are still here too | 16:25 |
apuimedo | :-) | 16:25 |
dmellado | apuimedo: https://review.openstack.org/487928 let's exchange reviews xD | 16:25 |
apuimedo | dmellado: why experimental man! | 16:28 |
apuimedo | then it only runs if manually triggered | 16:28 |
dmellado | because I want to test this refactor | 16:28 |
apuimedo | put it in check as non voting | 16:28 |
dmellado | and it's totally cheap to move from experimental | 16:28 |
dmellado | xD | 16:28 |
apuimedo | ok, ok | 16:28 |
apuimedo | can you do one for gate-install? | 16:28 |
apuimedo | hey, wait a sec | 16:28 |
apuimedo | you didn't put the postconfig | 16:29 |
dmellado | oh, was that needed? | 16:29 |
apuimedo | dmellado: well, in rdocloud otherwise the creation of the first amphora will fail due to time out | 16:29 |
dmellado | let's check this out here first | 16:29 |
dmellado | it might need to be tweaked to the infra | 16:29 |
apuimedo | I don't know how fast the infra nodes are | 16:30 |
apuimedo | ok | 16:30 |
*** limao has joined #openstack-kuryr | 16:32 | |
*** limao has quit IRC | 16:36 | |
*** yboaron_ has joined #openstack-kuryr | 16:53 | |
*** c00281451_ has quit IRC | 17:01 | |
*** c00281451_ has joined #openstack-kuryr | 17:02 | |
*** garyloug_ has joined #openstack-kuryr | 17:47 | |
*** garyloug has quit IRC | 17:47 | |
*** vikasc has quit IRC | 18:19 | |
*** vikasc has joined #openstack-kuryr | 18:32 | |
openstackgerrit | Merged openstack/kuryr-kubernetes master: Update documentation about nested deployment https://review.openstack.org/487906 | 18:50 |
*** garyloug_ has quit IRC | 19:42 | |
*** rwallner has quit IRC | 20:21 | |
*** rwallner has joined #openstack-kuryr | 20:24 | |
*** rwallner has quit IRC | 20:28 | |
*** edisonxiang has quit IRC | 20:54 | |
*** edisonxiang has joined #openstack-kuryr | 20:54 | |
*** alraddarla has quit IRC | 21:12 | |
*** yamamoto_ has joined #openstack-kuryr | 21:13 | |
*** yamamoto_ has quit IRC | 21:20 | |
*** aojea_ has quit IRC | 21:40 | |
*** aojea has joined #openstack-kuryr | 21:41 | |
*** aojea has quit IRC | 21:45 | |
*** aojea has joined #openstack-kuryr | 21:54 | |
*** pmannidi has joined #openstack-kuryr | 21:57 | |
*** aojea has quit IRC | 21:59 | |
*** yboaron_ has quit IRC | 22:04 | |
*** yamamoto has joined #openstack-kuryr | 22:04 | |
*** rwallner has joined #openstack-kuryr | 22:07 | |
*** gouthamr_ has joined #openstack-kuryr | 22:10 | |
*** gouthamr has quit IRC | 22:10 | |
*** rwallner has quit IRC | 22:12 | |
*** aojea has joined #openstack-kuryr | 22:14 | |
*** aojea has quit IRC | 22:19 | |
*** rwallner has joined #openstack-kuryr | 22:20 | |
*** rwallner_ has joined #openstack-kuryr | 22:21 | |
*** rwallner has quit IRC | 22:24 | |
*** gouthamr_ has quit IRC | 22:37 | |
*** openstack has joined #openstack-kuryr | 22:42 | |
*** mchiappero has quit IRC | 22:48 | |
*** rwallner_ has quit IRC | 22:51 | |
*** gouthamr has joined #openstack-kuryr | 22:53 | |
*** hongbin has quit IRC | 23:03 | |
*** edisonxiang_ has joined #openstack-kuryr | 23:38 | |
*** edisonxiang has quit IRC | 23:38 | |
apuimedo | bleh | 23:55 |
apuimedo | Octavia works, but we have some mess to sort out with security groups | 23:56 |
apuimedo | also, it seems it is needlessly taking ports in the pod subnet | 23:56 |
apuimedo | anyway. Tomorrow's another day | 23:56 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!