| *** limao has joined #openstack-kuryr | 00:42 | |
| *** yuanying has joined #openstack-kuryr | 01:03 | |
| *** yedongcan has joined #openstack-kuryr | 01:36 | |
| *** yuanying has quit IRC | 02:49 | |
| limao | ping mchiappero | 03:01 |
|---|---|---|
| *** vikasc has joined #openstack-kuryr | 03:02 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/fuxi: Updated from global requirements https://review.openstack.org/373745 | 03:38 |
| *** yedongcan has left #openstack-kuryr | 03:52 | |
| *** saneax-_-|AFK is now known as saneax | 04:23 | |
| *** yuanying has joined #openstack-kuryr | 04:58 | |
| *** ram_ has joined #openstack-kuryr | 05:31 | |
| openstackgerrit | Dongcan Ye proposed openstack/kuryr: Fix mock order https://review.openstack.org/409614 | 05:40 |
| *** irenab_ has joined #openstack-kuryr | 05:41 | |
| openstackgerrit | vikas choudhary proposed openstack/kuryr-libnetwork: Delay neutron extension checks to first neutron interaction https://review.openstack.org/403325 | 05:42 |
| *** reedip has quit IRC | 05:53 | |
| *** reedip has joined #openstack-kuryr | 06:03 | |
| *** yedongcan has joined #openstack-kuryr | 06:06 | |
| *** openstackgerrit_ has joined #openstack-kuryr | 06:23 | |
| *** openstackgerrit_ has quit IRC | 06:25 | |
| *** openstackgerrit_ has joined #openstack-kuryr | 06:26 | |
| *** openstackgerrit_ has quit IRC | 06:27 | |
| *** openstackgerrit_ has joined #openstack-kuryr | 06:34 | |
| *** openstackgerrit_ has quit IRC | 06:35 | |
| *** oanson has joined #openstack-kuryr | 06:43 | |
| *** openstackgerrit_ has joined #openstack-kuryr | 06:59 | |
| *** openstackgerrit_ has quit IRC | 07:00 | |
| *** dimak has joined #openstack-kuryr | 07:23 | |
| *** yamamoto has quit IRC | 07:26 | |
| openstackgerrit | Jaivish Kothari(janonymous) proposed openstack/kuryr-libnetwork: Unittest with mock https://review.openstack.org/405203 | 07:31 |
| janonymous | mchiappero: ^^ | 07:31 |
| ram_ | apuimedo: Hi. I stuck at basic devstack kuryr-kubernetes deployment. As I mentioned before I deployed devstack kuryr-kubernetes on Ubuntu and centOS. It was deployed. But docker service is not running. Can you please help me. (or) Can you please provide me any other alternative approach to get knowledge on Kuryr. | 07:37 |
| openstackgerrit | Dongcan Ye proposed openstack/kuryr: Passing port's project_id in port_bind https://review.openstack.org/409648 | 07:58 |
| ivc_ | irenab_ irenab ping | 08:06 |
| irenab_ | ivc_, good morning | 08:06 |
| irenab_ | ivc_, can you please check the devref patch, I embedded few diagrams | 08:07 |
| irenab_ | ivc_, also if you can share the Controller pipeline diagram as png file, so I can add it as well | 08:08 |
| ivc_ | irenab_ aye, i already did briefly | 08:09 |
| *** yamamoto has joined #openstack-kuryr | 08:09 | |
| irenab_ | ivc_, I wonder if services related stuff should go there as well. Mayeb better to add separate doc | 08:09 |
| *** yamamoto_ has joined #openstack-kuryr | 08:10 | |
| ivc_ | irenab_ actually i was thinking about some of the stuff that is expected to be reimplemented/refactored | 08:11 |
| ivc_ | irenab_ i mean if we expect some parts to change really soon, should we put it into devref? | 08:12 |
| irenab_ | ivc_, depends how soon we are going to reimplement, devref probably should reflect the current state + mention short term plans | 08:13 |
| *** yamamoto has quit IRC | 08:13 | |
| irenab_ | ivc_, I see the value of devref if it helps to understand the current approach without diving too deep into code | 08:14 |
| irenab_ | ivc_, what do you think? | 08:15 |
| ltomasbo | ping mchiappero | 08:17 |
| ivc_ | irenab_ https://dl.dropboxusercontent.com/u/12482084/ControllerPipeline.xmind | 08:17 |
| ivc_ | irenab_ https://dl.dropboxusercontent.com/u/12482084/pipeline.png | 08:18 |
| irenab_ | ivc_, thanks!, will push another patch with this diagram included | 08:18 |
| ivc_ | irenab_ ok. i'm thinking that we should "mention short term plans" in devref | 08:19 |
| ivc_ | apuimedo irenab_ thats why i've added that comment after all :) maybe we can have a meeting this week to discuss our plans | 08:20 |
| *** yamamoto_ has quit IRC | 08:20 | |
| irenab_ | ivc_, +1 | 08:21 |
| irenab_ | ivc_, apuimedo : we can set meeting for tomorrow morning | 08:21 |
| ivc_ | irenab_ i'd prefer wed or thu | 08:22 |
| irenab_ | ivc_, Wed is good for me | 08:23 |
| irenab_ | ivc_, lets hear from apuimedo and set a meeting before you both disapear for holidays | 08:23 |
| ivc_ | aye :) | 08:23 |
| *** yamamoto has joined #openstack-kuryr | 08:24 | |
| *** pcaruana has joined #openstack-kuryr | 08:30 | |
| *** yamamoto has quit IRC | 08:34 | |
| apuimedo | ram_: which services are running? | 08:35 |
| apuimedo | irenab_: tomorrow 11utc? | 08:35 |
| *** pcaruana has quit IRC | 08:37 | |
| *** pcaruana has joined #openstack-kuryr | 08:38 | |
| apuimedo | mchiappero: you have both irenab_, limao and my +1 on your base patch. As soon as it passes UTs, it will go in | 08:46 |
| openstackgerrit | Berezovsky Irena proposed openstack/kuryr-kubernetes: kuryr kubernetes design reference document https://review.openstack.org/408078 | 08:46 |
| openstackgerrit | Merged openstack/kuryr: Fix mock order https://review.openstack.org/409614 | 08:46 |
| openstackgerrit | Dongcan Ye proposed openstack/kuryr: Passing port's project_id in port_bind https://review.openstack.org/409648 | 08:48 |
| apuimedo | irenab_: great! Thanks for the fixes! | 08:50 |
| irenab_ | apuimedo, will send an invite | 08:51 |
| apuimedo | thanks! | 08:52 |
| irenab_ | apuimedo, vikasc : I pushed yesterday this patch https://review.openstack.org/#/c/409516/, but going to abandon it since there is a proper fix suggested at the mailing list. | 08:53 |
| irenab_ | I will post the patch with fix as suggested | 08:53 |
| openstackgerrit | vikas choudhary proposed openstack/kuryr-libnetwork: Delay neutron extension checks to first neutron interaction https://review.openstack.org/403325 | 08:53 |
| apuimedo | irenab_: cool | 08:53 |
| vikasc | thanks irenab | 08:53 |
| ram_ | apuimedo: HI. Entire info : http://paste.openstack.org/show/592069/. Check onnce this. | 08:55 |
| *** janki has joined #openstack-kuryr | 09:01 | |
| *** roeyc has joined #openstack-kuryr | 09:05 | |
| ivc_ | irenab_ apuimedo https://review.openstack.org/#/c/376045/ - you can now try k8s with services on devstack :) | 09:13 |
| ivc_ | irenab_ apuimedo i'll keep this patch in its current state for a while and will be cherry-picking parts from it into smaller patches | 09:14 |
| irenab_ | ivc_, will try it today. | 09:15 |
| ivc_ | irenab_ apuimedo also there's no sg handling there yet. lbaasv2 handles them completely differently in haproxy-agent and octavia. but right now it works in devstack as everything is on the same sg | 09:17 |
| *** limao has quit IRC | 09:17 | |
| irenab_ | ivc_, so internal services should work. As for haproxy-agent vs Octavia, there is going to be different driver? | 09:18 |
| ivc_ | irenab_ only Pod services are supported now, yes (there's a filter that skips everything else, including 'kubernetes' service). it can work with external services too, but that would require sg | 09:20 |
| ivc_ | irenab_ as for octavia - yes. there's lbaas handler (that manages the logic and coupling of lbaas components) and lbaasv2 driver that implements neutron interactions | 09:22 |
| ivc_ | irenab_ so for octavia we only need to add the driver as they both have the same model | 09:22 |
| irenab_ | ivc_, thanks for confirmation. | 09:23 |
| ivc_ | irenab_ the driver only responsible for trivial operations (i.e. create/delete) and all the coupling is done in the handler | 09:23 |
| irenab_ | ivc_, so since the API is not different, Handler is generic and Driver is pecific | 09:24 |
| irenab_ | spcific | 09:24 |
| ivc_ | irenab_ eventually we might need to split the handler if octavia's model diverges from the lbaasv2 | 09:24 |
| irenab_ | ivc_, this will be probably v3 or something | 09:25 |
| irenab_ | ivc_, makes sense | 09:25 |
| irenab_ | ivc_, I tend to have separate devref for services support | 09:26 |
| ivc_ | yup :) | 09:26 |
| ivc_ | i'm ok with that | 09:26 |
| *** limao has joined #openstack-kuryr | 09:26 | |
| ivc_ | but then we need to rename ' | 09:26 |
| ivc_ | kuryr kubernetes design reference document' | 09:26 |
| ivc_ | or we can add services to the same devref later | 09:27 |
| ivc_ | apuimedo btw do you know whats wrong with our 'docs' gate? | 09:29 |
| irenab_ | ivc_, I know | 09:29 |
| ivc_ | irenab_ infra? | 09:29 |
| irenab_ | I had temporary fix https://review.openstack.org/#/c/409516/ | 09:29 |
| irenab_ | but will post soon the fix as suggested at the mailing list | 09:30 |
| irenab_ | like here: https://review.openstack.org/#/c/409630/1/test-requirements.txt | 09:30 |
| ivc_ | irenab_ hmm actually i find your ':remote:' fix much more appropriate | 09:31 |
| irenab_ | ivc_, according to the email, it just make build suceed but does not add budges | 09:32 |
| ivc_ | ok | 09:32 |
| apuimedo | sorry. I was in a meeting | 09:35 |
| apuimedo | ivc_: great! Will review | 09:36 |
| ram_ | apuimedo: Hi. If you are free. Please let me know. | 09:38 |
| *** garyloug has joined #openstack-kuryr | 09:38 | |
| ivc_ | apuimedo irenab_ yup, i think we can keep reviews in that patch and the split patches will only add ut/docs | 09:38 |
| apuimedo | ivc_: sounds good | 09:38 |
| apuimedo | irenab_: I'll send you new svg/png for the flow diagram | 09:38 |
| apuimedo | I made a couple of improvements | 09:39 |
| irenab_ | apuimedo, cool, thanks | 09:41 |
| *** yuanying has quit IRC | 09:43 | |
| *** yamamoto has joined #openstack-kuryr | 09:48 | |
| *** yamamoto has quit IRC | 09:49 | |
| openstackgerrit | Berezovsky Irena proposed openstack/kuryr-kubernetes: Pin docutils version https://review.openstack.org/409716 | 10:17 |
| irenab_ | apuimedo, vikasc ^^ | 10:17 |
| apuimedo | irenab_: let's see if the requirements gate likes it | 10:18 |
| vikasc | all looks green | 10:21 |
| *** yamamoto has joined #openstack-kuryr | 10:27 | |
| *** yamamoto has quit IRC | 10:30 | |
| * apuimedo merging | 10:30 | |
| *** yedongcan has left #openstack-kuryr | 10:31 | |
| vikasc | ram_, docker should be running in one of the devstack screens | 10:35 |
| openstackgerrit | Merged openstack/kuryr-kubernetes: Pin docutils version https://review.openstack.org/409716 | 10:35 |
| vikasc | ram_, if you do "screen -x", there should be a screen named docker | 10:35 |
| vikasc | ps ax | grep docke | 10:36 |
| openstackgerrit | Ilya Chukhnakov proposed openstack/kuryr-kubernetes: devstack: LBaaSv2 and K8s service subnet https://review.openstack.org/407244 | 10:36 |
| openstackgerrit | Ilya Chukhnakov proposed openstack/kuryr-kubernetes: [WIP] Service/Endpoints/LBaaS event processing https://review.openstack.org/376045 | 10:41 |
| *** yamamoto has joined #openstack-kuryr | 10:43 | |
| *** limao has quit IRC | 10:48 | |
| *** yamamoto has quit IRC | 10:51 | |
| apuimedo | irenab_: I just sent you the new version of the diagram | 10:58 |
| apuimedo | please, check it out and let me know | 10:58 |
| *** yamamoto has joined #openstack-kuryr | 11:03 | |
| *** garyloug has quit IRC | 11:03 | |
| *** garyloug has joined #openstack-kuryr | 11:06 | |
| *** yamamoto has quit IRC | 11:08 | |
| *** roeyc has quit IRC | 11:08 | |
| *** roeyc has joined #openstack-kuryr | 11:13 | |
| irenab_ | apuimedo, got it, will be at meeting till weekly meeting, will upload new version afterwards | 11:27 |
| apuimedo | irenab_: great | 11:30 |
| apuimedo | :-) | 11:30 |
| ram_ | vikasc: OK. Thank you. I will check that | 11:30 |
| openstackgerrit | Merged openstack/kuryr: Passing port's project_id in port_bind https://review.openstack.org/409648 | 11:40 |
| ram_ | vikasc/apuimedo: [stack@localhost devstack]$ ps -aef | grep docker stack 10955 10457 0 17:10 pts/0 00:00:00 grep --color=auto docker. docker not running | 11:41 |
| vikasc | screen -x, is there a screen named "docker" ? | 11:42 |
| *** lezbar__ has quit IRC | 11:45 | |
| *** lezbar has joined #openstack-kuryr | 11:46 | |
| ram_ | vikasc/apuimedo: Yes. I got the point. We can see the status of docker-engine service like $sudo systemctl status docker-engine. $screen -x I restarted the docker-engine it restarted without errors. | 12:00 |
| openstackgerrit | Berezovsky Irena proposed openstack/kuryr-kubernetes: kuryr kubernetes design reference document https://review.openstack.org/408078 | 12:02 |
| *** garyloug has quit IRC | 12:03 | |
| *** yamamoto has joined #openstack-kuryr | 12:04 | |
| *** yamamoto has quit IRC | 12:05 | |
| ram_ | vikas/apuimedo: Imean We can't see the status of docker-engine service like $sudo systemctl status docker-engine correctly. | 12:15 |
| openstackgerrit | Berezovsky Irena proposed openstack/kuryr-kubernetes: kuryr kubernetes design reference document https://review.openstack.org/408078 | 12:16 |
| *** saneax is now known as saneax-_-|AFK | 12:33 | |
| *** limao has joined #openstack-kuryr | 12:34 | |
| *** limao has quit IRC | 12:40 | |
| *** limao has joined #openstack-kuryr | 12:47 | |
| apuimedo | ram_: but is it running on foreground in one of the screen tabs? | 13:00 |
| *** irenab_ has quit IRC | 13:01 | |
| ram_ | apuimedo: I restarted the docker-engine as mentioned aside http://paste.openstack.org/show/592109/. We can confirm like : I don't know exactly. even after restarting the docker-engine, docker commands are working. that means docker is running right. | 13:36 |
| ram_ | apuimedo: What happened exactly is: I am using devstack right. When I try to start the docker-engine service like : $sudo systemctl start docker-engine from CLI , I am facing issue: http://paste.openstack.org/show/592111/ | 13:40 |
| ram_ | apuimedo: Later I opened $screen -x and restarted docker-engine, then docker commands working again. | 13:41 |
| apuimedo | ram_: you shouldn't run docker-engine with systemctl | 13:41 |
| apuimedo | the devstack plugin run_docker function runs it for you inside a screen tab | 13:42 |
| apuimedo | irenab: ping | 13:47 |
| irenab | apuimedo: pong | 13:47 |
| apuimedo | the blueprint link that you request in https://review.openstack.org/#/c/407244/ is for some pre-existing blueprint, or to create a new blueprint and link to it? | 13:47 |
| *** s1061123 has quit IRC | 13:48 | |
| *** s1061123 has joined #openstack-kuryr | 13:48 | |
| apuimedo | irenab: cause there is https://blueprints.launchpad.net/kuryr/+spec/kuryr-k8s-integration only | 13:48 |
| irenab | apuimedo: I think the one we use for all k8s patches | 13:48 |
| irenab | apuimedo: maybe we didn’t … | 13:49 |
| irenab | let me check | 13:49 |
| irenab | apuimedo: like here | 13:49 |
| irenab | https://review.openstack.org/#/c/404038/ | 13:49 |
| apuimedo | ok, that's the one I was looking at | 13:50 |
| apuimedo | irenab: I'll add the partially-implements | 13:50 |
| apuimedo | and we can merge it | 13:50 |
| irenab | sure | 13:50 |
| apuimedo | irenab: posted | 13:51 |
| irenab | apuimedo: regarding the subnet created by devstack. Maybe worth to mention that its temporary till the tool is added? | 13:52 |
| apuimedo | irenab: where, in plugin.sh? | 13:53 |
| irenab | or commit msg | 13:53 |
| apuimedo | ok | 13:53 |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes: devstack: LBaaSv2 and K8s service subnet https://review.openstack.org/407244 | 13:55 |
| apuimedo | irenab: I just pushed it in plugin.sh | 13:55 |
| irenab | apuimedo: ok | 13:55 |
| apuimedo | I'm now running it to make sure it works :P | 13:56 |
| irenab | we can add it as ‘bug’ at launchpad, so anyone who wants to cotribute, can pick it. will do it after the sync on Wed | 13:56 |
| apuimedo | irenab: very well | 13:57 |
| apuimedo | irenab: although it kinda sounds more blueprintish to me | 13:57 |
| irenab | apuimedo: maybe, needs scope definition, so bp will be fine | 13:58 |
| apuimedo | agreed | 13:59 |
| *** garyloug has joined #openstack-kuryr | 14:01 | |
| apuimedo | vikasc: meeting? | 14:02 |
| *** oanson has quit IRC | 14:02 | |
| *** hongbin has joined #openstack-kuryr | 14:04 | |
| *** saneax-_-|AFK is now known as saneax | 14:13 | |
| *** pmannidi has quit IRC | 14:18 | |
| *** pmannidi has joined #openstack-kuryr | 14:25 | |
| apuimedo | janonymous: go ahead | 14:44 |
| janonymous | apuimedo: hey, about the tls bp, it required to have https://github.com/docker/docker/blob/c4d45b6a29a91f2fb5d7a51ac36572f2a9b295c6/docs/extend/plugin_api.md#json-specification .json file | 14:46 |
| apuimedo | right | 14:46 |
| janonymous | which have directories, which could be configurable | 14:46 |
| apuimedo | sure | 14:47 |
| janonymous | so, should i write a tool to generate .json file from config? | 14:47 |
| janonymous | or put hardcoded json file with default directory paths, or a better approach? | 14:47 |
| apuimedo | janonymous: I think it's fine to just put a sample json file in contrib | 14:50 |
| apuimedo | and point to it in the readme | 14:50 |
| apuimedo | so packagers know to use it | 14:50 |
| janonymous | great!, apart from that how to test is another big issue.. | 14:51 |
| janonymous | because all the setup for tls needs to be created for docker... | 14:51 |
| apuimedo | would need parametrizing in devstack and the fullstack jobs | 14:53 |
| janonymous | apuimedo: Ohhkay,makes sense. | 14:55 |
| apuimedo | ;-) | 14:56 |
| janonymous | thanks! | 14:56 |
| *** saneax is now known as saneax-_-|AFK | 15:08 | |
| *** janki has quit IRC | 15:08 | |
| hongbin | apuimedo: ping | 15:09 |
| *** limao has quit IRC | 15:16 | |
| *** hongbin has quit IRC | 15:22 | |
| *** saneax-_-|AFK is now known as saneax | 15:32 | |
| *** dimak has quit IRC | 15:46 | |
| *** garyloug has quit IRC | 15:46 | |
| *** saneax is now known as saneax-_-|AFK | 16:01 | |
| *** saneax-_-|AFK is now known as saneax | 16:16 | |
| *** hongbin has joined #openstack-kuryr | 16:20 | |
| *** roeyc has quit IRC | 16:23 | |
| apuimedo | hongbin: pong | 16:32 |
| hongbin | apuimedo: hey | 16:32 |
| hongbin | apuimedo: want to get your opinion for one thing | 16:32 |
| hongbin | apuimedo: about collecting more use cases/contributions for fuxi, do you have any suggestion? i was thinking to send out ML to ask for contribution. what do you think about that? | 16:33 |
| apuimedo | hongbin: I think taking to ML is a good idea | 16:34 |
| apuimedo | an email in which you list some of the current use cases | 16:34 |
| apuimedo | and ask for more | 16:34 |
| hongbin | apuimedo: ack | 16:34 |
| apuimedo | ;-) | 16:34 |
| hongbin | apuimedo: besides that, do you have any other suggestion? | 16:35 |
| *** diogogmt has joined #openstack-kuryr | 16:35 | |
| hongbin | apuimedo: i will send out a ML and cc you . hope it help :) | 16:40 |
| apuimedo | mmm | 16:40 |
| apuimedo | hongbin: maybe check if rkt support makes sense | 16:40 |
| hongbin | apuimedo: i see. that could be a great feature | 16:41 |
| hongbin | apuimedo: maybe an extension of your proposal. supporting OCI containers might be a useful feature | 16:41 |
| hongbin | apuimedo: right now, vendors started to decouple docker via OCI/runc, it could be useful for fuxi to explore such use cases | 16:42 |
| hongbin | apuimedo: btw, will discuss that with the community to get more feedback | 16:43 |
| apuimedo | right | 16:44 |
| apuimedo | thanks hongbin! | 16:44 |
| hongbin | apuimedo: thank you | 16:44 |
| apuimedo | you're welcome | 16:46 |
| apuimedo | ivc_: ping | 16:53 |
| ivc_ | apuimedo bang! | 16:53 |
| apuimedo | :-D | 16:53 |
| ivc_ | reviewing https://review.openstack.org/#/c/408078/7/doc/source/devref/kuryr_kubernetes_design.rst :) | 16:53 |
| apuimedo | ivc_: can you check https://drive.google.com/file/d/0B19pY94Ttis4MC00VjFkNWNZZlFXZUhkbENYOXJneFlKNDJZ/view?usp=sharing | 16:54 |
| apuimedo | let me know if you have some suggestion for it | 16:54 |
| apuimedo | it's part of that patch | 16:54 |
| ivc_ | aye | 16:55 |
| *** pcaruana has quit IRC | 16:59 | |
| *** saneax is now known as saneax-_-|AFK | 17:06 | |
| apuimedo | ivc_: can you paste me your local.conf for services? | 17:07 |
| apuimedo | Neutron is failing to start saying lbaasv2 is missing for me | 17:07 |
| ivc_ | its in the comment | 17:08 |
| ivc_ | for that patch | 17:08 |
| apuimedo | ok | 17:08 |
| apuimedo | I only looked at the sample | 17:08 |
| ivc_ | mm what sample? | 17:08 |
| ivc_ | http://paste.openstack.org/show/592053/ | 17:08 |
| apuimedo | ivc_: https://review.openstack.org/#/c/407244/7/devstack/local.conf.sample | 17:09 |
| ivc_ | yep its the same local.conf | 17:10 |
| apuimedo | mmm | 17:10 |
| ivc_ | the only change is i've added 'ref' to point at gerrit | 17:10 |
| apuimedo | ok, I'll investigate more | 17:10 |
| apuimedo | maybe I had some bad version of neutron (I had reclone=no) | 17:13 |
| ivc_ | apuimedo i've finally finished that review! https://review.openstack.org/#/c/408078/7 | 17:16 |
| apuimedo | 33 comments, not bad :-) | 17:18 |
| apuimedo | ivc_: what about the diagram? | 17:19 |
| ivc_ | yep, the diagram | 17:19 |
| ivc_ | right | 17:19 |
| ivc_ | i've checked it earlier in the morning and it looked good. i'll go through it one more now to double check | 17:20 |
| apuimedo | thanks | 17:22 |
| ivc_ | apuimedo i'm not sure what 'watch for pod running()' means or why its there | 17:22 |
| apuimedo | ivc_: that's to make it visible that the user pod start returns immediately and if they want to see when the pod status is 'started' they need to either poll or watch | 17:23 |
| ivc_ | ok | 17:24 |
| apuimedo | otherwise it seemed like end user pod creation blocked until the containers were running | 17:24 |
| apuimedo | meh. I still get http://paste.openstack.org/show/592131/ ivc_ | 17:25 |
| ivc_ | i mean its k8s contract, and has nothing to do with us, but i like the idea to show the transition to 'running' state | 17:26 |
| ivc_ | apuimedo don't tell me they've updated neutron and broke it xD | 17:26 |
| apuimedo | no idea | 17:26 |
| apuimedo | I'll be checking with newton now | 17:26 |
| ivc_ | it wont work | 17:27 |
| ivc_ | there are some dependency issues | 17:27 |
| apuimedo | ivc_: I searched like for 10 minutes where kubelet reports to k8s api 'started', but I only found some events for rkt impl. Not for docker | 17:27 |
| apuimedo | I assumed that it is async, that kubelet sends a request to k8s api, but can't find it | 17:28 |
| apuimedo | I'll look a bit later, to have it be more accurate | 17:28 |
| ivc_ | i'm gonna restack soon. i need to prepare a demo for tomorrow | 17:29 |
| ivc_ | there's a good chance neutron-lbaas can be broken as they transition to octavia | 17:30 |
| apuimedo | yeah | 17:31 |
| apuimedo | ivc_: btw, feel free to add to https://etherpad.openstack.org/p/kuryr-kubernetes-dec14-syncup | 17:40 |
| *** roeyc has joined #openstack-kuryr | 17:49 | |
| *** openstack has joined #openstack-kuryr | 17:58 | |
| *** roeyc has quit IRC | 17:58 | |
| ivc_ | apuimedo 'show port' polling is triggered by event that has VIF with active=False | 18:00 |
| ivc_ | apuimedo also maybe we can add a dash line from vif-plug to neutron to point out that it triggers port status=ACTIVE in neutron | 18:02 |
| *** roeyc has joined #openstack-kuryr | 18:04 | |
| ivc_ | apuimedo also i'm not sure what is 'create vif()' on CNI part. by 'create' do you mean 'deserialise from annotation'? | 18:04 |
| ivc_ | apuimedo added file coments https://review.openstack.org/#/c/408078/7/doc/images/pod_creation_flow.png,unified | 18:11 |
| openstackgerrit | Ilya Chukhnakov proposed openstack/kuryr-kubernetes: [WIP] Service/Endpoints/LBaaS event processing https://review.openstack.org/376045 | 18:14 |
| apuimedo | ivc_: good idea on the dashed line | 18:23 |
| apuimedo | ivc_: by create I meant the device | 18:23 |
| ivc_ | apuimedo then vif.plug happens before veth is created | 18:23 |
| ivc_ | apuimedo and yes, it's not confusing at all that we 'plug' before actually creating the device xD | 18:25 |
| *** ram_ has quit IRC | 18:26 | |
| *** vikasc has quit IRC | 18:28 | |
| *** reedip has quit IRC | 18:32 | |
| *** vikasc has joined #openstack-kuryr | 18:42 | |
| *** reedip has joined #openstack-kuryr | 18:44 | |
| *** roeyc has quit IRC | 18:55 | |
| *** tonanhngo has joined #openstack-kuryr | 19:01 | |
| *** tonanhngo_ has joined #openstack-kuryr | 19:03 | |
| *** tonanhngo has quit IRC | 19:06 | |
| *** vikasc has quit IRC | 19:06 | |
| irenab | ivc_: ping | 19:16 |
| ivc_ | irenab pong | 19:16 |
| irenab | I had few minutes to play with the services patch | 19:16 |
| ivc_ | irenab did it ./stack.sh successfully? | 19:17 |
| irenab | I see that per each pod there is a separate pool | 19:17 |
| irenab | yes, but I ran it around 4 pm | 19:17 |
| irenab | I used the same local.conf as you did | 19:17 |
| ivc_ | apuimedo was having some issues with neutron/lbaas (unrelated to kuryr) | 19:17 |
| irenab | I didn’t check the connectivity yet, but deployment wise looks ok | 19:18 |
| irenab | so I was not sure why each pod in service gets its own pool | 19:18 |
| irenab | what is the model mapping you have in mind? | 19:19 |
| ivc_ | its not exactly what you think :) | 19:19 |
| ivc_ | only one of those pools is the working one | 19:19 |
| irenab | all polls have the same name | 19:19 |
| irenab | I see ... | 19:19 |
| irenab | so its so issue with lbass provider? | 19:20 |
| ivc_ | https://review.openstack.org/#/c/376045/11/kuryr_kubernetes/controller/drivers/lbaasv2.py@200 | 19:20 |
| ivc_ | i thought it would be enough | 19:20 |
| ivc_ | and it was working ok, but i too have the same issue now (tho i've only got 2 pools for 3 pods) | 19:21 |
| ivc_ | in fact | 19:21 |
| ivc_ | https://review.openstack.org/#/c/376045/11/kuryr_kubernetes/controller/drivers/lbaasv2.py@285 | 19:21 |
| ivc_ | it is there to prevent it | 19:22 |
| irenab | I tried with 2 pods, so maybe its always 2 pools, regardless number of pods | 19:22 |
| ivc_ | its random | 19:22 |
| ivc_ | i need to remove that coment from https://review.openstack.org/#/c/376045/11/kuryr_kubernetes/controller/drivers/lbaasv2.py@200 | 19:22 |
| *** vikasc has joined #openstack-kuryr | 19:23 | |
| ivc_ | you can uncomment that line and restart kuryr | 19:23 |
| irenab | let me check if its commented | 19:23 |
| ivc_ | it is | 19:23 |
| ivc_ | i forgot to uncomment it while i was playing with bogus_cleanup@200 | 19:24 |
| irenab | it is not commented | 19:24 |
| ivc_ | hmm | 19:25 |
| ivc_ | u sure? | 19:25 |
| ivc_ | it certainly is commented in gerrit | 19:25 |
| ivc_ | https://review.openstack.org/#/c/376045/11/kuryr_kubernetes/controller/drivers/lbaasv2.py@285 | 19:26 |
| ivc_ | # self._wait_for_provisioning(loadbalancer, remaining) | 19:26 |
| irenab | I thought you are refering to line 200 | 19:26 |
| irenab | uncommented this one, let me recheck | 19:27 |
| openstackgerrit | Ilya Chukhnakov proposed openstack/kuryr-kubernetes: [WIP] Service/Endpoints/LBaaS event processing https://review.openstack.org/376045 | 19:27 |
| irenab | ivc_: now the issue is fixed | 19:31 |
| ivc_ | irenab yep | 19:32 |
| ivc_ | irenab the problem is lbaasv2. you create_lbaas_pool while loadbalancer's provisioning_state!=ACTIVE, and it fails with an exception | 19:32 |
| ivc_ | irenab but it still creates the pool | 19:33 |
| irenab | ivc_: bug or feature? | 19:33 |
| ivc_ | irenab dunno | 19:33 |
| ivc_ | irenab they require loadbalancer to be ACTIVE before you update anything | 19:34 |
| irenab | ivc_: how do you check connectivity? | 19:34 |
| ivc_ | irenab curl | 19:34 |
| ivc_ | u run nginx image? | 19:35 |
| irenab | yep | 19:35 |
| ivc_ | do 'hostname > /usr/.../html/test' and curl http://serviceip/test | 19:35 |
| ivc_ | to check for round robin | 19:36 |
| irenab | from devstack host? | 19:36 |
| ivc_ | for /html/test you need to go into each pod's shell | 19:37 |
| ivc_ | i was going to write a script to do it (or build an image) | 19:37 |
| ivc_ | but i've just screwed up my env by running 10 nova instances | 19:38 |
| irenab | you already started with nested case? | 19:38 |
| ivc_ | nop | 19:38 |
| ivc_ | just 'for fun' to see how long it would take compared with our pods | 19:39 |
| ivc_ | but i forgot they also require disk space and openstack does not seem to care to check if i have enough of it to schedule vms | 19:40 |
| irenab | do not say nova is faster | 19:40 |
| ivc_ | :) | 19:41 |
| ivc_ | it only takes ~5 seconds for 1 pod | 19:42 |
| ivc_ | no way nova can beat that | 19:42 |
| ivc_ | tho things get worse as you add --replicas :) | 19:42 |
| *** shashank_hegde has joined #openstack-kuryr | 19:50 | |
| irenab | ivc_: seems to work fine. I verifiedthrough deploying https://github.com/kubernetes/kubernetes/blob/master/examples/guestbook/frontend-deployment.yaml | 20:03 |
| irenab | and it runs curl to nginx service ip | 20:04 |
| ivc_ | cool :) | 20:05 |
| irenab | namefor the load balancers is a bit wierd | 20:06 |
| irenab | Endpoints:default/frontend | 20:06 |
| irenab | will add comment to the patch | 20:06 |
| ivc_ | what part of it is wierd? | 20:10 |
| irenab | Endpoints | 20:10 |
| ivc_ | its object_kind:object_namespace/object_name | 20:10 |
| irenab | I think service is more appropriate, but I understand the automation you did | 20:11 |
| irenab | maybe we need object_kid -> representation_string mapping | 20:12 |
| ivc_ | yeah i understand about 'Service' and it was that way initially | 20:13 |
| ivc_ | but my thinking was since the loadbalancer's lifecycle is bound to endpoints, it would be more appropriate | 20:14 |
| irenab | endpoints is sort of implementation details | 20:14 |
| ivc_ | we could add name to LBaaSServiceSpec | 20:15 |
| ivc_ | so it will be generated for service | 20:15 |
| irenab | so it will set the name too? | 20:17 |
| ivc_ | yes. LBaaSServiceSpec is created by service watcher, so it knows everything about service | 20:18 |
| irenab | yes, seems proper place to put it | 20:18 |
| ivc_ | and endpoints handler will use LBaaSServiceSpec.name for LB name | 20:18 |
| ivc_ | LBaaSServiceSpec is present on both service and endpoints | 20:19 |
| irenab | Great work, could run only into nits :-) | 20:20 |
| ivc_ | code you mean? :) | 20:20 |
| irenab | heading to sleep now | 20:20 |
| irenab | yes, the services support | 20:20 |
| ivc_ | thnx :) | 20:20 |
| ivc_ | nn :) | 20:21 |
| *** oanson has joined #openstack-kuryr | 20:56 | |
| openstackgerrit | Hongbin Lu proposed openstack/fuxi: Add the remote tag to the badge image https://review.openstack.org/409939 | 21:36 |
| openstackgerrit | Hongbin Lu proposed openstack/fuxi: Updated from global requirements https://review.openstack.org/373745 | 21:38 |
| *** jgriffith is now known as jgriffith_AutoAw | 21:47 | |
| *** oanson has quit IRC | 21:50 | |
| *** jgriffith_AutoAw is now known as jgriffith | 21:52 | |
| openstackgerrit | Hongbin Lu proposed openstack/fuxi: Updated from global requirements https://review.openstack.org/373745 | 22:14 |
| *** yamamoto has joined #openstack-kuryr | 22:22 | |
| *** jgriffith is now known as jgriffith_AutoAw | 22:23 | |
| hongbin | hi, if any core is here, need your help to approve a fix to the docs gate: https://review.openstack.org/#/c/409939/ | 22:45 |
| hongbin | apuimedo: irenab ^^ | 22:45 |
| apuimedo | hongbin: checking | 22:45 |
| hongbin | thx | 22:46 |
| openstackgerrit | Merged openstack/fuxi: Add the remote tag to the badge image https://review.openstack.org/409939 | 22:50 |
| *** jgriffith_AutoAw is now known as jgriffith | 22:57 | |
| *** pmannidi has quit IRC | 23:07 | |
| *** pmannidi has joined #openstack-kuryr | 23:07 | |
| openstackgerrit | Hongbin Lu proposed openstack/fuxi: Enable use of constraints for all tox based jobs. https://review.openstack.org/409968 | 23:09 |
| *** yuanying has joined #openstack-kuryr | 23:13 | |
| *** diogogmt has quit IRC | 23:18 | |
| *** yuanying has quit IRC | 23:20 | |
| *** yuanying has joined #openstack-kuryr | 23:25 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/fuxi: Updated from global requirements https://review.openstack.org/373745 | 23:28 |
| *** reedip has quit IRC | 23:39 | |
| *** dougbtv has quit IRC | 23:51 | |
| *** reedip has joined #openstack-kuryr | 23:53 | |
| *** dougbtv has joined #openstack-kuryr | 23:54 | |
| *** dougbtv has quit IRC | 23:54 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
