14:01:57 <dulek> #startmeeting Kuryr 14:01:58 <openstack> Meeting started Mon Jan 29 14:01:57 2018 UTC and is due to finish in 60 minutes. The chair is dulek. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:01:59 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:02:01 <openstack> The meeting name has been set to 'kuryr' 14:02:18 <dulek> Hi everyone! 14:02:31 <leyal> Hi 14:02:36 <maysamacedos> Hi 14:02:40 <yboaron_> Hi 14:03:03 <kaliya> Hi 14:03:10 <dulek> I've started the meeting as apuimedo attends a conference today. I think he's not the only one who's missing. 14:03:44 <dulek> Does anyone have anything to discuss regarding kuryr-lib or kuryr plugin? 14:03:53 <dulek> Or can we go straight to kuryr-kubernetes? 14:04:35 <dulek> #topic kuryr-kubernetes 14:05:22 <dulek> Okay! I can start by saying that I was swamped with some downstream work last week, so I don't have much to update on kuryr-kubernetes. 14:05:44 <dulek> Anyone wants to share some progress or point a patch that needs reviews? 14:06:01 <yboaron_> Yes 14:06:40 <yboaron_> I've sent two patches deals with Ingress controller/Router implementation , please review 14:06:40 <dulek> yboaron_: Microphone is yours. :) 14:07:05 <yboaron_> #link https://review.openstack.org/#/c/536387/ 14:07:17 <yboaron_> #https://review.openstack.org/#/c/538577/ 14:07:29 <dulek> #link https://review.openstack.org/#/c/538577/ 14:07:51 <yboaron_> that's it 14:07:58 <dulek> +800 and +1300. Big ones! 14:08:23 <dulek> I assume it's the Ingress one that has the common parts? 14:08:35 <yboaron_> yes 14:09:28 <dulek> yboaron_: Are you able to demonstrate new functionality in the gate? I mean - is it possible to create a Tempest test for Ingress? 14:09:52 <leyal> yboaron, is that possible to separate between the "Readiness for new PaaS" and the "Common part of Ingress Controller" ? 14:10:13 <leyal> just for make more easy to review 14:10:37 <yboaron_> dulek, I guess we can create new gate for both ingress controller and openshift route 14:11:14 <dulek> yboaron_: It would require whole new gate? 14:11:20 <dulek> yboaron_: Why? 14:11:33 <yboaron_> leyal, it's possible , but I don't see for what purpose 14:11:51 <dulek> yboaron_: I thought tests could work on the existing gates - either K8s or OpenShift. 14:12:32 <yboaron_> dulek, TBH , Im not familiar with gates issues , I'll catch gcheresh and see how/why/whuch gates we should create for that purpose 14:12:44 <leyal> yboaron, small patches are more easy to review .. 14:13:32 <dulek> yboaron_: Awesome. :) I think having it demonstrated in the gate would make core reviewers more confident with merging it. 14:13:33 <yboaron_> leyal, agree , but ~650 lines of 1330 are tests :-) 14:13:55 <dulek> leyal: On the other hand "common" changes would not be testable. 14:14:12 <yboaron_> dulek, OK , I'll follow your recommendation 14:14:31 <dulek> leyal: Personally I prefer to have code that's doing something instead of some dead codepaths merged initially. 14:14:41 <dulek> But that's just my point of view. :) 14:14:50 <yboaron_> dulek, +1 14:15:12 <leyal> yboaron,dulek, yep agree , thats the trade-off 14:16:25 <dulek> Okay, anyone else? 14:16:28 <maysamacedos> I also have one patch waiting for review that adds Kuryr Controller liveness, I would really appreciate if you could take a look 14:16:42 <maysamacedos> #link https://review.openstack.org/#/c/535548/ 14:16:53 <yboaron_> maysamacedos, I'll review ASAP 14:17:05 <dulek> maysamacedos: Yeaaah, I have it opened in a tab in the browser for like… a week? Sorry about that, I'll take a look. 14:17:08 <maysamacedos> thanks yboaron_ 14:17:20 <maysamacedos> thanks dulek 14:17:42 <dulek> #link https://review.openstack.org/#/c/533292/ 14:18:55 <dulek> So who else have something to discuss? 14:19:03 <maysamacedos> I also started taking a look on adding CNI health checks, and I saw that you, dulek, have worked on the daemon 14:19:43 <dulek> maysamacedos: That's right, though I was just finishing unfinished patch. :) 14:19:55 <dulek> maysamacedos: What's your idea of healthchecks on kuryr-daemon? 14:20:20 <maysamacedos> what do you think about starting the health manager as a separated process? just as the current server and watcher 14:20:21 <maysamacedos> ? 14:20:44 <dulek> maysamacedos: Ah, that's a sure thing! 14:20:53 <maysamacedos> ok, great 14:21:07 <dulek> maysamacedos: So unlike kuryr-controller being based on oslo.service, kuryr-daemon uses cotyledon. 14:21:23 <dulek> Becasue oslo.service needs eventlet and eventlet breaks pyroute2. 14:22:07 <dulek> In that case you can just create another "Service" class and start and stop the healthcheck server there. 14:22:32 <dulek> I just wonder what's worth checking in case of daemon. 14:22:37 <maysamacedos> for now I was just thinking about adding checks in respect of k8s api.. and check the watcher threads when k8s is restarted 14:23:02 <leyal> dulek, eventlet also break ThreadPool -and it's causes some issues with k8s client 14:23:04 <dulek> Yeah, I was just writing that - daemon needs access to k8s API to work. 14:23:07 <celebdor> sorry I'm late. I'm on PTO 14:23:17 <maysamacedos> yes, after some time I realized that dulek 14:23:19 <dulek> celebdor: o/ 14:23:41 <celebdor> are we discussing getting rid of eventlet? 14:23:49 <dulek> maysamacedos: Awesome. Please ping me in case you'll need any help with understanding daemon code. 14:24:13 <maysamacedos> ok, thanks dulek 14:24:44 <dulek> celebdor: Not really, I was just mentioning that kuryr-daemon uses cotyledone, because eventlet breaks pyroute2. :) 14:25:26 <celebdor> ah, right 14:25:35 <dulek> celebdor: We've briefly discussed yboaron_ implementation patches and then maysamacedos efforts to create healthchecks. 14:25:42 <celebdor> cool 14:26:05 <celebdor> how about leyal's spec. Did you all review it? 14:26:59 * dulek suspects that's a "no"… 14:27:35 <yboaron_> I did (not the latest version) , LGTM 14:27:54 <yboaron_> I'll take a look at the latest vesion 14:27:57 <yboaron_> version 14:28:05 <celebdor> dulek: maysamacedos: did you review? 14:28:30 <dulek> celebdor: Nope, I'll go through it again today. 14:28:31 <leyal> celebdor, i found a small bug in the design , working on solution and i will update . maybe will set workflow -1 until than .. 14:28:43 <maysamacedos> No, not yet celebdor 14:29:13 <celebdor> very well 14:29:23 <celebdor> maysamacedos: I'll make a blueprint for the daemon liveness 14:29:39 <celebdor> dulek: You should take a look at that too 14:29:43 <celebdor> I'll shoot you an email 14:29:46 <maysamacedos> thanks celebdor 14:29:49 <dulek> celebdor: Okay! 14:30:26 <celebdor> irenab: did you see ttx's email about PTG schedule? 14:30:27 <yboaron_> celebdor, I need your +2 power for : #link https://review.openstack.org/#/c/529966/ 14:31:25 <maysamacedos> leyal, could you send the link of the spec in review 14:32:00 <dulek> celebdor: Anything changed regarding PTG? I cannot find the email. 14:32:13 <dulek> maysamacedos: https://review.openstack.org/#/c/519239 14:32:21 * dulek has it all opened… 14:32:32 <maysamacedos> thanks dulek 14:32:49 <leyal> maysamacedos, sure https://review.openstack.org/#/c/519239/ 14:33:57 <celebdor> dulek: well, irenab was asking for the kuryr track to be from tuesday 14:34:07 <celebdor> and ttx proposed to have tue-wed 14:34:18 <celebdor> and if we need more, to dynamically ask for some room on thu 14:35:01 <celebdor> what do you all think? 14:35:03 <dulek> Well, works for me. Thursday might also be useful to talk with other projects. 14:35:54 <maysamacedos> works for me as well 14:36:00 <leyal> celebdor, is remote participation will be enable ? 14:38:09 <dulek> leyal: I think we can try to do that, though it might be a bit hard without mic and audio. 14:40:24 <celebdor> will do as much as I can for remote 14:40:25 <leyal> dulek, cool 14:41:28 <maysamacedos> dulek: I just saw that you've linked the 'old' patch I had for Kuryr Controller liveness. The correct one is: 14:41:33 <maysamacedos> #link https://review.openstack.org/#/c/535548/ 14:41:44 <maysamacedos> I will abandon the other one, sorry 14:43:00 <dulek> :) 14:43:12 <dulek> Okay, so anything else to discuss? 14:43:25 <maysamacedos> not from me 14:43:50 <dulek> celebdor: I don't think irenab is online, so I guess you'll need to ask her about scheduling through email. 14:44:03 <dulek> celebdor: Anything else from you? 14:47:56 <dulek> Okay, I guess we can end the meeting then. 14:48:04 <dulek> Thanks for joining! 14:48:08 <dulek> #endmeeting