15:03:05 <v1k0d3n> #startmeeting openstack-helm 15:03:05 <openstack> Meeting started Tue May 16 15:03:05 2017 UTC and is due to finish in 60 minutes. The chair is v1k0d3n. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:03:06 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:03:09 <openstack> The meeting name has been set to 'openstack_helm' 15:03:14 <korzen> o/ 15:03:14 <alraddarla_> o/ 15:03:16 <jaugustine> \o/ 15:03:17 <dougbtv> o/ 15:03:17 <v1k0d3n> alraddarla_: ;) 15:03:21 <dulek> :) 15:03:24 <alraddarla_> :) 15:03:25 * alanmeadows waves. 15:03:28 <mburnett> o/ 15:03:38 <lrensing> o/ 15:03:41 <gagehugo> o/ 15:03:52 <jayahn> o/ 15:04:02 <korzen> #link https://etherpad.openstack.org/p/openstack-helm-meeting-2017-05-16 Agenda 15:04:11 <v1k0d3n> nice full house, light agenda. 15:04:19 <leifmadsen> o/ 15:04:31 <v1k0d3n> let's start with some of the outstanding issues we know about. 15:04:36 <mattmceuen> o/ 15:04:59 <v1k0d3n> portdirect: what is the current status on kubernetes v1.6.3 issues? any updates? 15:05:29 <mburnett> His computer is hung, he's still trying to join the meeting. 15:05:36 <srwilkers> o/ 15:06:00 * mburnett being portdirect 15:06:02 <v1k0d3n> let's move onto another one then for now. we'll see him join. 15:06:11 <mburnett> they seem to be mostly resolved 15:06:23 <mburnett> the reamining issues that I am seing are to do with cni 15:06:42 <mburnett> so for now i would recommend staying with 1.6.2 15:06:50 <v1k0d3n> mburnett (portdirect), is there any outstanding issues we need on our side or is there any PS you want the larger team to look at and prioritize? 15:07:10 <lrensing> +1, 1.6.2 is still what i'm going with 15:07:23 <mburnett> not yet, but during the day this may change 15:07:29 <bryan_att> #info Bryan Sullivan 15:07:46 <mburnett> I think we need to decide on a couple of CNI's to recommend using 15:07:56 <v1k0d3n> yup. the bug filed recommends that users stay on kube v1.6.2, so we have something to point them at. 15:08:18 <mburnett> as it appears that we flip flop between flannel/calico as being the most stable 15:08:22 <v1k0d3n> a couple of CNI's? you mean SDN's? 15:08:31 <mburnett> (I'm lumping flannel and canal together) 15:08:48 <mburnett> potato/potato :) but yeah 15:09:20 <alanmeadows> Ideally, it would be whatever CNIs the gate is going to validate 15:09:32 <mburnett> we could validate both 15:09:38 <v1k0d3n> yeah, i don't deploy flannel without policy anymore personally...it would probably be wise for us to categorize as "use whatever, we can recommend L2 or L3, canal+flannel/calico", and keep our docs in line with that. 15:09:48 <v1k0d3n> thoughts, agreement/disgreement....? 15:09:50 <mburnett> but for now 1.6.2 with calico is the path i'd recommend for least pain 15:10:22 <v1k0d3n> yeah +1 to alanmeadows comment. easier to catch issues and recommend that way. 15:10:49 <v1k0d3n> ok, so action item really is to update docs with relevant versions. 15:11:38 <v1k0d3n> next onto helm 2.4 vs 2.3. we do need to make some changes here. is anyone interested in working on resolving this? 15:12:21 <v1k0d3n> we could probably outline the issue a bit better here: https://bugs.launchpad.net/openstack-helm/+bug/1690863 15:12:22 <openstack> Launchpad bug 1690863 in openstack-helm "Helm 2.4.0 Issues with Openstack-Helm" [High,New] 15:12:49 * mburnett being me again :) 15:13:08 <mburnett> Did someone offer to work on that yesterday? 15:13:32 <v1k0d3n> mburnett: i didn't see anyone in IRC bring that up as an item they were working on. 15:13:59 <mburnett> i think chriss was gonna look at it (pete)? 15:14:34 <v1k0d3n> either is fine with me, as long as it gets assigned in launchpad. 15:14:54 <v1k0d3n> it's still unassigned. 15:16:07 <v1k0d3n> i can try to reach out to chris offline. 15:16:24 <v1k0d3n> next thing is dns issues, which a few people have mentioned. 15:16:35 <v1k0d3n> portdirect: is this somethiing you're actually looking into as well? 15:16:54 <mburnett> yeah - its related to the CNI issues 15:17:11 <mburnett> they are all part and parcel of the same thing really i think 15:17:27 <mburnett> so the same advice applies - 1.6.2 and calico for now :) 15:17:28 <v1k0d3n> all just part of v1.6.3? 15:17:50 <mburnett> it looks that way - but I'm hesitant to point the finger 15:17:51 <v1k0d3n> ok. i've seen other dns issues on 1.6.2 is why i ask. 15:18:03 <mburnett> as I've not seen the same stuff with other backends 15:18:17 <alanmeadows> Chris was going to look at the ceph chart 15:18:26 <alanmeadows> bringing that inline 15:18:34 <alanmeadows> not helm 2.4 to my knowledge 15:18:48 <alanmeadows> but we can verify offline 15:18:56 <v1k0d3n> that's true. i'll action to reach out. 15:19:03 <mburnett> just looks like calico and 1.6.3 for some reason dont like combing up together - if you poke them its seems to resole a lot - but via kubeadm its not so plain sailing 15:19:38 <v1k0d3n> if he's not working on helm 2.4 i will ping in our IRC openstack-helm channel for volunteers. 15:20:05 <mburnett> alanmeadows, i think as part of that he was gonna adress the secret gen, which is whats causing the issues with 2.4 now that it correctly fails :) 15:20:20 <v1k0d3n> mburnett: just fwiw...i am seeing dns related issues in general, even with other SDN's 15:20:24 <mburnett> but if not I'd be happy to port the stuff i did for upstream docker-ceph back here 15:21:16 <alanmeadows> mburnett: correct, it plays into 2.4 in that way 15:22:06 <alanmeadows> We need to stop spinning on k8s+calico configurations--riding releases 48 hours old is killing lots of time. We need to find one that works, that the community can easily select and are not blocked from installing (e.g. 1.5)--the only thing we should be rolling with tightly is helm in my opinion 15:22:19 <srwilkers> alanmeadows, agreed 15:23:07 <mburnett> +1 15:23:31 <jayahn> +1 15:23:37 <mburnett> the kubeadm-aio container is now pinned resonably tightly 15:23:46 <mburnett> and I'm pushing known good images to dockerhub 15:24:16 <alanmeadows> That may be 1.6.3, if we wait for calico to catchup to resolve the remaining issues being seen? 15:24:30 <mburnett> yup 15:25:00 <mburnett> my experienc is saying that it takes about 7 days for things to return to smooth waters after a k8s release 15:26:54 <alanmeadows> aio is nicely pinned at the moment, I agree, and can easily be adjusted to what we land on. To close on this topic I think once we find that stack, we should provide all assets in our documentation -- not just versions, today mostly we recommend versions, and tell you to (for example) head off to calico and find the right manifest as an example 15:27:26 <alanmeadows> This means direct links and cut and paste manifests for every aspect because our users will spend too much time spinning 15:28:06 <mburnett> that sounds good - I'll hit that today (pete) 15:28:12 <jayahn> we can probably help on that to add flannel option since we use flannel for k8s. 15:28:13 <mburnett> unless there are any other takers? 15:28:23 <mburnett> oh nice - cheers jayahn 15:28:41 <mburnett> we could get a gate using that as well really easily if it helps 15:29:11 <alanmeadows> and I think having two options for CNIs at least demonstrates other options exist, its just whatever is there needs to be validated 15:29:27 <v1k0d3n> jayahn: flannel + canal just so we have policy when needed. that work? 15:29:41 <v1k0d3n> L2 and L3...that's pretty awesome. 15:30:21 <jayahn> we are simply using flannel, that said, having canal would be awesome. i agree. I will check with our network guys. 15:30:32 <v1k0d3n> ok, so i think that solves that at least. 15:30:41 <v1k0d3n> any other known issues that the team is fighting with? 15:31:06 <dulek> RBAC auth? 15:31:34 <dulek> I have a review posted that I think enables RBAC support: https://review.openstack.org/#/c/464630/ 15:31:41 <v1k0d3n> ah, yes...i saw that. 15:31:56 <v1k0d3n> can cores provide feedback? 15:32:15 <dulek> I would be grateful! :) 15:32:31 <mburnett> will do (pete) 15:32:53 <lrensing> will do 15:32:58 <srwilkers> yep 15:33:01 <v1k0d3n> same here dulek 15:33:05 <mburnett> sorry I've not got round to it - but I think we should probably have a etherpad re rbac before jumping in with code? 15:33:07 <v1k0d3n> thanks for that PS! 15:33:40 <v1k0d3n> i'm sorry mburnett can you explain a bit better? 15:33:42 <dulek> mburnett: There's a blueprint at least: https://blueprints.launchpad.net/openstack-helm/+spec/rbac-support 15:33:56 <dulek> (although not many details there…) 15:33:57 <mburnett> as we need to consider namespace / service seperation and isolation as part of this 15:34:00 <v1k0d3n> the blueprint and PS should be good discussion points. 15:34:13 <mburnett> nice - I'd missed that thanks dulek 15:34:18 <mburnett> I'll get on that asap 15:34:20 <v1k0d3n> optional namespace / service separation. 15:35:54 <v1k0d3n> so moving on, i'm opening the floor for people to call out important submitted PS. 15:36:33 <v1k0d3n> dulek started this already, we can expand. is there anything else that the team needs to lay eyes on as a priority? 15:36:45 <alanmeadows> re: https://blueprints.launchpad.net/openstack-helm/+spec/support-linux-bridge-on-neutron 15:36:53 <alanmeadows> is artur looking at this already? 15:37:01 <alanmeadows> or not yet - trying to keep from overlap 15:37:07 <dulek> v1k0d3n: Cinder config overrides are blocking at least one patch. 15:37:11 <jayahn> mburnett just confirming that we will merge "https://review.openstack.org/#/c/464104/" and work on "https://review.openstack.org/#/c/464669/1" together for long term? 15:37:28 <korzen> alanmeadows, yes I'm starting on it 15:37:51 <v1k0d3n> ok, let's break this up a little bit. 15:37:53 <mburnett> jayahn, I think that's what I prefer -- I'm not sure what others think (me) 15:37:56 <jayahn> for https://blueprints.launchpad.net/openstack-helm/+spec/support-linux-bridge-on-neutron, let me know whom to connect. 15:38:45 <mburnett> we should loop artur into that - but we can hash out the deatils in that launchpad (pete) 15:38:48 <v1k0d3n> jayahn: please get with korzen on this. 15:39:06 <jayahn> alanmeadows: v1k0d3n: we really want to collaborate on "creating pattern" for multi backend. 15:39:24 <mburnett> yeah this follows on from what we were discussing at the summit 15:39:24 <v1k0d3n> and we can talk openly in the openstack-helm channel about the pattern you're talking about jayahn 15:39:26 <v1k0d3n> i agree. 15:39:30 <lrensing> that'd be awesome jayahn :) 15:39:35 <jayahn> yeap. 15:39:45 <alanmeadows> jayahn/korzen: how about a simple etherpad 15:39:53 <alanmeadows> to capture the approach 15:40:04 <v1k0d3n> yes, etherpad. we can capture in the blueprint as well? 15:40:12 <jayahn> that would be good start, then, i can point etherpad to my guys working on this 15:40:15 <alanmeadows> and get alignment and of course all can participate in the review of the first submission, and then the door is open to others 15:40:17 <v1k0d3n> or would that be an entirely wrong place to have some directional points? 15:40:26 <srwilkers> yeah, we can link the etherpad on the blueprint in launchpad 15:40:27 <v1k0d3n> +1 to alanmeadows 15:40:51 <v1k0d3n> perfect. jayahn i will let you create and then paste in this meetings etherpad. 15:41:04 <alanmeadows> korzen can you create an etherpad, jot your approach as an outline and let Jayahn and team dogpile on? 15:41:16 <alanmeadows> Keeps from rewriting patches at the outset 15:41:29 <korzen> alanmeadows, sure 15:41:42 <jayahn> +1 15:41:42 <v1k0d3n> alanmeadows: you'll definitely want to provide a lot of feedback on these as well as portdirect 15:41:56 <v1k0d3n> ok, cool...that works. 15:43:15 <v1k0d3n> another big item i want to hit is the work that is being done with other tools and oslo. 15:43:19 <alanmeadows> v1k0d3n: the cinder topic was a good one... it would be great to complete that one, is it only waiting for reviews dulek? 15:43:26 <v1k0d3n> alanmeadows: are you aware of the current outstanding ps? 15:43:36 <alanmeadows> I got halfway through yesterday before getting distracted 15:43:48 <dulek> alanmeadows: I think so - I've addressed all of the comments so far. 15:43:48 <v1k0d3n> yeah i was going to hit cinder as well. 15:43:52 <v1k0d3n> glad you brought up. 15:44:09 <v1k0d3n> so the oslo thing is big, because once that is set...it's set,and will be very hard to change in the future. 15:44:25 <v1k0d3n> the tripelo team is very interested in having us at the table, and collaborating on this. 15:44:32 <v1k0d3n> openstack-ansible as well. 15:44:48 <v1k0d3n> do you happen to have that link in front of you so we can paste in etherpad? 15:44:59 <v1k0d3n> alanmeadows: ^^ 15:45:17 <alanmeadows> not sure which link 15:45:35 <v1k0d3n> ps submitted to oslo. one sec let me grab it. 15:46:07 <v1k0d3n> https://review.openstack.org/#/c/454897/ 15:46:21 <v1k0d3n> EmilienM: asked that we provide feedback, and has been really patient. 15:46:35 <v1k0d3n> portdirect: would be really great to have you looking at this as well. 15:46:35 <EmilienM> o/ 15:46:48 <v1k0d3n> hey man, bringing up the oslo stuff we talked about at the summimt. 15:46:51 <EmilienM> and also look https://etherpad.openstack.org/p/BOS-forum-future-of-configuration-management if you missed the session 15:46:59 <v1k0d3n> i really want our team to keep this close and provide feedback. 15:47:20 <EmilienM> v1k0d3n: thanks, really cool if you could provide feedback about the etherpad && spec 15:48:08 <v1k0d3n> absolutely. thanks for being patient with us EmilienM and keeping up on me about it. 15:48:15 <EmilienM> no problemo 15:48:53 <v1k0d3n> ok, so next thing is what you brought up alanmeadows about cinder backends. dulek you're up. 15:49:13 <v1k0d3n> want to talk a bit about it. i know you're looking for feedback. 15:49:23 <v1k0d3n> would be nice to have folks looking at this one as well... 15:49:30 <v1k0d3n> https://review.openstack.org/#/c/455682/ 15:50:01 <dulek> v1k0d3n: I think we've covered that it requires reviews. I guess discussion should continue on the review itself. 15:50:34 <v1k0d3n> yes, that works. 15:50:51 <v1k0d3n> ok. any others team 15:50:52 <v1k0d3n> ? 15:51:00 <jayahn> we will try to add feeback. 15:51:19 <v1k0d3n> thank jayahn that would be great 15:51:48 <mburnett> It would be good to get some feedback on the mariadb changes in https://review.openstack.org/#/c/464262/ 15:52:15 <v1k0d3n> and lastly...low hanging fruit review here as well: https://review.openstack.org/#/c/460896/ 15:52:28 <jayahn> okay, bring it on :) 15:52:38 <v1k0d3n> mburnett: noted and included in our review and bp seection. 15:52:43 <mburnett> Thanks 15:53:08 <v1k0d3n> anything else before i end the meeting? 15:53:50 <v1k0d3n> sorry there was more to cover than expected, but it was great to pass these by everyone. i'd like to keep up with these for the week and close them out as quickly as possible. 15:54:43 <v1k0d3n> alright ending the meeting team. 15:54:50 <v1k0d3n> thanks everyone for joining 15:54:53 <jayahn> cheers 15:55:03 <v1k0d3n> #endmeeting"