13:59:36 <celebdor1> #startmeeting kuryr 13:59:37 <openstack> Meeting started Mon Jul 2 13:59:36 2018 UTC and is due to finish in 60 minutes. The chair is celebdor1. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:59:38 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:59:40 <openstack> The meeting name has been set to 'kuryr' 13:59:45 <celebdor1> Hello everybody 13:59:53 <celebdor1> and welcome to Kuryr's weekly meeting 13:59:58 <celebdor1> who's here today? 14:00:46 <ltomasbo_> o/ 14:00:54 <dulek_> o/ 14:02:05 <dmellado> o/ (After a *rushed* lunch) 14:02:20 <garyloug_> o/ 14:02:30 <celebdor1> dmellado: I didn't even have lunch 14:02:33 <celebdor1> nor breakfast 14:02:38 <celebdor1> anyway 14:02:44 <celebdor1> let's get the party started 14:02:47 <dmellado> celebdor1: really? have you become a cyborg in the end? 14:02:49 <celebdor1> #topic kuryr-kubernetes 14:02:59 <celebdor1> dmellado: I turn to fasting until Octavia is fast 14:03:39 <dmellado> celebdor1: have you ever head about Sokushinbutsu 14:03:41 <dmellado> ? 14:03:43 <celebdor1> #info we had a flask related regression. dmellado spotted it and fixed it 14:03:43 <dmellado> take care 14:03:55 <dulek_> celebdor1: Hey, it was me who fixed it! :P 14:04:05 <dmellado> with the help of dulek_ and yboaron 14:04:10 <dmellado> it was a terrible CI incident 14:04:11 <celebdor1> oh 14:04:12 <dmellado> xD 14:04:15 <celebdor1> I thought it was dmellado 14:04:16 <dmellado> a collaborative effort 14:04:20 <celebdor1> well, both start with d 14:04:22 <dulek_> celebdor1: ltomasbo_ was working on that as well! :) 14:04:27 <celebdor1> and it was 2:30 in the morning when I saw it 14:04:29 <celebdor1> xD 14:04:32 <dmellado> dulek_: ltomasbo_ too? xD 14:04:33 <celebdor1> sorry dulek_ 14:04:44 <dmellado> I thought all he did was patchwork 14:04:50 <dulek_> Talking about gate stability - this should help with LBaaS v2 occasional failures: https://review.openstack.org/#/c/579533/ 14:05:00 <dmellado> dulek_: yep 14:05:03 <dmellado> in any case, regarding that 14:05:05 <celebdor1> #info multinode is finally working 14:05:06 <dulek_> And we should be fairly stable at the moment. 14:05:14 <dulek_> So I expect something breaks again soon. :) 14:05:21 <dmellado> let's move those to queens/stable branch 14:05:22 <dulek_> So use this time wisely! 14:05:24 <dmellado> when we realease r 14:05:29 <celebdor1> duldulgood point 14:05:32 <dmellado> dulek_: you mean drinking beers? 14:05:35 <dmellado> duldul xD 14:05:38 <celebdor1> damned keyboard 14:05:51 <dulek_> dmellado: No need. stable/queens is still on Zuul v2. 14:06:01 <dulek_> dmellado: And flask version there is fixed. 14:06:02 <celebdor1> #info we were not fully configuring neutron-lbaasv2 in gates 14:06:06 <dulek_> So no need for backports IMO. 14:06:09 <celebdor1> thanks for the fix dulek 14:06:25 <dmellado> dulek_: let's keep it for now and check if they move to zuulv3 at some point 14:06:38 <dmellado> it'd be fairly simple to move that to that branch if needed 14:06:48 <celebdor1> #info snapiri split zuul into smaller files 14:06:56 <dmellado> thanks snapiri ! 14:07:01 <dulek_> dmellado: Agreed, I'm pretty sure no one will want to move .zuul.yaml to V3 in stable/queens. ;) 14:07:04 <dmellado> \o/ 14:07:10 <celebdor1> #info lv jiawei fixed cgroups settings 14:07:44 <celebdor1> #info neutron's lbaasv2 implementation is officially deprecated 14:08:16 <dmellado> we'll keep those lbaas gates in master until next release 14:08:19 <celebdor1> #info yossi and ltomasbo_ have worked on stabilizing octavia's driver 14:08:21 <dmellado> then deprecate as well 14:08:30 <celebdor1> dmellado: until current release 14:08:37 <dmellado> celebdor1: that's what I meant 14:08:47 <dmellado> just different wording 14:08:48 <dmellado> xD 14:08:52 <ltomasbo_> celebdor1, we still have something to fix: https://review.openstack.org/#/c/579144 14:09:04 <ltomasbo_> dulek_, left a comment on: https://review.openstack.org/#/c/579144/5/kuryr_kubernetes/controller/drivers/lbaasv2.py 14:09:09 <celebdor1> #info dulek_ fixed the health binding address after flask behavior change 14:09:20 <dmellado> ltomasbo_: saw it 14:09:24 <ltomasbo_> dulek_, I'm not sure if that could affect other side effects 14:09:27 <celebdor1> ltomasbo_: I know. that's why I didn't say "fixed" 14:09:29 <celebdor1> xD 14:09:29 <dmellado> I'll give it a try on my env and review 14:09:42 <ltomasbo_> but I'm pretty much doing what it was already done when the id or provider was missing 14:10:01 <celebdor1> ok 14:10:10 <celebdor1> I think we covered most of what happened last week 14:10:12 <yboaron> ltomasbo_, will review it after the meeting 14:10:18 <celebdor1> What do we have on the plate for this week? 14:10:27 <dmellado> ltomasbo_: celebdor1 any info on the progress on the namespace SG thing? 14:10:34 <dulek_> ltomasbo_: Can't see it, you probably haven't submitted the comment. 14:10:34 <celebdor1> there is! 14:10:41 <celebdor1> ltomasbo_: has a prototype 14:10:43 <ltomasbo_> thanks yboaron! that I tested locally and it is working, but not sure if other cases are affected 14:10:44 <dmellado> I'll resume my NPS job and would like to take that into consideration 14:10:53 <dmellado> ltomasbo_: and you kept it for yourself 14:10:54 <dmellado> you greedy 14:10:57 <celebdor1> #info apuimedo sent a patch to enable openshift's DNS 14:11:03 <celebdor1> (on devstack 14:11:07 <ltomasbo_> celebdor1, dmellado: yes, I was about to test my PoC regarding that (but lost connection with the server...) 14:11:08 <celebdor1> the same should be done for kubernetes 14:11:18 <dmellado> celebdor1: I'm waiting on your tempest patch 14:11:21 <ltomasbo_> celebdor1, I found out some flakiness on devstack plugin though 14:11:23 <celebdor1> #info ltomasbo sent a PoC for openshift project isolation 14:11:32 <celebdor1> dmellado: I have it on the VM 14:11:34 <celebdor1> I'll push later 14:11:42 <celebdor1> how do I skip unless it is openshift? 14:11:52 <dmellado> you'd need to create a flag 14:11:54 <celebdor1> ltomasbo_: which flakiness? 14:11:58 <dmellado> just like we did last time 14:11:58 <celebdor1> dmellado: we don't have any?! 14:12:03 <dmellado> let me check if I can find it 14:12:05 <dmellado> we have a regex 14:12:07 <dmellado> on the gate 14:12:10 <dmellado> you can just define it there 14:12:13 <dmellado> or a flag 14:12:26 <celebdor1> #info peng liu is pushing forward with the standard multi vif effort 14:12:30 <ltomasbo_> security groups are awk-ed by 'default', so if any other sg has the string 'default' in it, it fails (as it happened to me on my PoC) 14:12:44 <ltomasbo_> so, I'll modify it as part of my patch to 'Default' 14:12:46 <dmellado> celebdor1: example 14:12:46 <celebdor1> ltomasbo_: that's funny 14:12:48 <dmellado> https://review.openstack.org/#/c/578062/ 14:12:58 <celebdor1> thanks dmellado 14:13:00 <dmellado> ltomasbo_: can't you try to use id? 14:13:25 <dmellado> I mean, I'd try to avoid relying on names if possible 14:13:34 <celebdor1> #info dulek_ is improving devstack with octavia timeouts and API IPs 14:13:58 <dulek_> Yup, I have an issue there, hopefully it was related to https://review.openstack.org/#/c/579533/ 14:14:09 <celebdor1> ok 14:14:14 <ltomasbo_> it is for deploying, and as it was relying on name, it was taking the wrong id when using my modification taht create a couple of extra SGs 14:14:49 <dmellado> ltomasbo_: huh, yeah, I've seen things like that in the past 14:14:51 <dmellado> blame devstack 14:15:02 <dmellado> I saw a tshirt on vancouver that it said 14:15:06 <dmellado> 'It fails on devstack' 14:15:08 <dmellado> xD 14:15:12 <ltomasbo_> cF 14:15:14 <ltomasbo_> xD 14:15:14 <dmellado> sadly I couldn't get to get a hold of it 14:15:16 <celebdor1> #info yboaron sent further documenation and tests for Openshift's routes 14:15:22 <celebdor1> folks... Review! 14:15:37 <dmellado> celebdor1: I will do that when it's documentation 14:15:44 <dmellado> you need a new keyboard! xD 14:16:20 <celebdor1> dmellado: I'm hungry 14:16:21 <dmellado> https://review.openstack.org/#/c/578806/ 14:16:23 <celebdor1> I eat letters too 14:16:26 <dmellado> FYI folks 14:16:40 <dmellado> as long as you don't do like gabert and eat my chocolates I'm fine 14:17:01 <celebdor1> dmellado: what's with the ovn gate? 14:17:10 <dmellado> celebdor1: it was broken due to their devstack plugin 14:17:21 <dmellado> we'd need to rebase the make-it-voting patch and that'd be it 14:17:33 <celebdor1> also, please review danil's sriov patches 14:17:37 <dulek_> dmellado: I see it's performing better now, so I'm all for it. 14:17:37 <dmellado> but I'd keep it as n-v until it stabilyzes 14:18:07 <dmellado> dulek_: well, we can always make it nv anytime 14:18:10 <dmellado> so we SHOULD be 14:18:14 <dmellado> done with infra for a while 14:18:23 <celebdor1> dmellado: merge https://review.openstack.org/#/c/576853/1 14:18:24 <dulek_> :) 14:18:25 <dmellado> (now is when dulek_ tells me we have three issues at the same time) 14:18:55 <dulek_> dmellado: Nah, not yet. But no worries, I'm on PTO for Thursday and Friday, so I expect I'll find something on Wednesday. 14:19:07 <dmellado> dulek_: lol 14:19:19 <dmellado> no worries, I'll just make a change so it breaks there 14:19:21 <dmellado> xD 14:19:34 <dmellado> in any case congrats everyone who was involved into this 14:19:39 <dmellado> we made it work and multinode already 14:19:43 <celebdor1> xD 14:19:43 <dmellado> hoorray \o/ 14:19:58 <celebdor1> openshift multinode is going to be a PITA 14:20:23 <celebdor1> anything else folks? 14:20:27 <celebdor1> I have nothing else for the meeting 14:20:36 <dmellado> https://toufayan-cdn.sirv.com/WP_SirvMediaLibrary/2014/12/napa_chicken_pita_salad.png 14:20:38 <dmellado> yeah 14:20:43 <dmellado> I've got another topic! 14:21:07 <dmellado> so, folks, next week's Mon I'd like to have a BJ 14:21:17 <dmellado> so we can further discuss topics for the Berlin summit 14:21:20 <dmellado> irenab: snapiri 14:21:30 <dmellado> I'll add to you and also some dell folks who are interested 14:21:47 <dmellado> danil might be too so I'll send the invite over the ML 14:22:11 <celebdor1> true 14:22:13 <celebdor1> :P 14:22:16 <celebdor1> ok, ok 14:22:22 <dmellado> and I guess that's it for now 14:22:27 <dmellado> thanks everyone for attending! 14:22:29 <AlexeyPerevalov> o/ hello 14:22:30 <celebdor1> #action dmellado to write abstracts 14:22:30 <dmellado> celebdor1: go for lunch! 14:22:32 <dmellado> xD 14:22:40 <celebdor1> #chair dmellado 14:22:41 <openstack> Current chairs: celebdor1 dmellado 14:22:51 <celebdor1> dmellado: unfortunately now I have to go pick up the kids 14:22:58 <celebdor1> and then join the meeting 14:22:58 <dmellado> celebdor1: ttyl in the meeting then 14:23:00 <dmellado> enjoy! 14:23:04 <celebdor1> I don't know when I'll be able to eat 14:23:12 <celebdor1> AlexeyPerevalov: hi 14:23:19 <AlexeyPerevalov> I still have to say something regarding vhostuser on baremetal installation. 14:23:26 <dmellado> AlexeyPerevalov: go for it 14:23:34 <AlexeyPerevalov> I already implemented PoC patche, now it's on internal code review. But I faced with issue on Kubernetes side. 14:23:47 <AlexeyPerevalov> It doesn't obtain ip from CNI output, it gets it from container's eth0 ( 14:23:54 <AlexeyPerevalov> so I fixed it, 14:24:06 <AlexeyPerevalov> there https://github.com/AlexeyPerevalov/kubernetes/commit/ba863551448b8f4fc92953097a589359dbf1a58c 14:24:17 <AlexeyPerevalov> need to discuss with Kubernetes community. 14:24:49 <AlexeyPerevalov> So I'll update spec, as soon it will be clear with kubernetes behaviour in case when we don't have eth0 inside container. 14:25:03 <dmellado> AlexeyPerevalov: but do you mean using kuryr as CNI with kubernetes? 14:25:20 <dulek_> AlexeyPerevalov: Hey, that's surprising that K8s doesn't follow the CNI spec. 14:25:24 <ltomasbo> btw, yboaron: seems a different modification in octavia was needed: https://review.openstack.org/#/c/579559/ 14:25:28 <dmellado> totally 14:25:32 <dmellado> #chair dulek_ ltomasbo 14:25:33 <openstack> Current chairs: celebdor1 dmellado dulek_ ltomasbo 14:25:42 <dmellado> folks, I'll need to drop for a moment too 14:25:52 <dmellado> pls close the meeting when all is done 14:26:36 <AlexeyPerevalov> dulek_: for me too ) kubernetes unittests rely on eth0 as well 14:27:35 <AlexeyPerevalov> dmellado: kuryr-kubernetes, but not kuryr ) 14:27:50 <dulek_> AlexeyPerevalov: Anything else to share? Do you have an action item for us? 14:28:15 <AlexeyPerevalov> dmelado: why you think about kuryr? 14:28:45 <AlexeyPerevalov> dulek_: yes, btw your CNI output is little bit out dated, kuryr-kubernetes tells 0.3.0 version, but format is in previous version (0.2.0), ;) 14:29:28 <AlexeyPerevalov> so I'm going to fix it too, now I'm working it around. 14:29:33 <dulek_> AlexeyPerevalov: Oh, this is totally a bug. Can you file it? bugs. https://bugs.launchpad.net/kuryr-kubernetes 14:29:39 <dulek_> AlexeyPerevalov: Awesome, thanks! 14:30:10 <AlexeyPerevalov> dulek_: ok, I'll file it. 14:31:09 <AlexeyPerevalov> that's all from my side. 14:32:06 <dulek_> Anyone else has something to share before we close the meeting? 14:32:11 <pliu> With regarding to the Multi-VIF driver, currently 2 approaches are proposed, however neither of it got enough comments. 14:32:52 <pliu> https://review.openstack.org/575629 and https://review.openstack.org/#/c/577070 14:34:36 <dulek_> Yeah, I need to walk through them as well. I guess dmellado and celebdor aka apuimedo should too. 14:35:14 <dulek_> Patches are now sitting in the limbo of my opened browser tabs. Hopefully that'll help them get my review. :) 14:36:05 <pliu> :) I am thinking that people can take a look at both patch and give comments. and help to find a way forward. 14:36:30 <dulek_> Sure! 14:36:58 <dulek_> So, anything else, anyone? :) 14:37:48 <pliu> BTW, https://review.openstack.org/#/c/578009/ is a reference implementation of https://review.openstack.org/575629 14:38:18 <dulek_> pliu: Okay, noted! 14:40:36 <dulek_> Okay, I guess this is the moment to close this. Thanks everyone! 14:41:01 <dulek_> #endmeeting