21:01:21 <strigazi> #startmeeting containers
21:01:22 <openstack> Meeting started Tue Mar 12 21:01:21 2019 UTC and is due to finish in 60 minutes.  The chair is strigazi. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:01:23 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:01:25 <openstack> The meeting name has been set to 'containers'
21:01:29 <strigazi> #topic Roll Call
21:01:41 <strigazi> o/
21:01:47 <imdigitaljim> o/
21:01:48 <alisanhaji> Hi
21:02:08 <jakeyip> o/
21:02:08 <strigazi> hello alisanhaji imdigitaljim
21:02:17 <strigazi> hi jakeyip
21:02:23 <jakeyip> hi strigazi
21:02:30 <ttsiouts> o/
21:02:53 <alisanhaji> I would like to talk about some evolutions to magnum, is their a slot for that during the meetings?
21:03:26 <strigazi> alisanhaji: sure, I'll ping you
21:03:36 <alisanhaji> strigazi: great thanks!
21:03:47 <strigazi> flwang: ?
21:04:27 <strigazi> #topic Stories/Tasks
21:05:08 <strigazi> 1. ttsiouts pushed a series of patches for nodegroupds
21:05:10 <strigazi> https://review.openstack.org/#/q/status:open+project:openstack/magnum+branch:master+topic:magnum_nodegroups
21:05:35 <strigazi> Please have a look, mostly the ones that don't say WIP
21:05:40 <strigazi> brtknr: ^^
21:06:31 <strigazi> I reviewed them in person but I'll leave comments in gerrit too. We need input from others too
21:06:53 <strigazi> Any questions comments?
21:07:00 <strigazi> about nodegroups
21:07:42 <ttsiouts> strigazi: I'll try to push some more things by the end of the week
21:07:48 <jakeyip> strigazi: I'm trying to find a story describing this... are you able to point me to one?
21:08:05 <ttsiouts> mostly regarding driver
21:08:13 <strigazi> jakeyip: https://github.com/openstack/magnum-specs/blob/master/specs/stein/magnum-nodegroups.rst
21:08:23 <jakeyip> thanks!
21:09:30 <strigazi> jakeyip: official source http://git.openstack.org/cgit/openstack/magnum-specs/tree/specs/stein/magnum-nodegroups.rst
21:09:46 <strigazi> for some reason I can't find it in http://specs.openstack.org/openstack/magnum-specs/
21:09:54 <strigazi> ttsiouts: thanks
21:10:22 <strigazi> ttsiouts: we might need to update some details in the spec? like min node-count?
21:10:40 <ttsiouts> strigazi: sure
21:10:48 <flwang> sorry, i was in a short meeting
21:10:49 <jakeyip> I see highly available magnum clusters in the text. our users wants to be able to create a cluster with workers in different AZ, does this work allow us to achieve that?
21:11:07 <strigazi> yes, that is the main use case
21:11:45 <jakeyip> that's great!
21:12:28 <strigazi> different AZ and gpu vs 'plain' vm are the most wanted for us too
21:13:17 <strigazi> any other question about NGs?
21:13:33 <jakeyip> thanks for the work ttsiouts, strigazi, much appreciated
21:14:00 <strigazi> ttsiouts++
21:14:07 <strigazi> next,
21:14:18 <strigazi> 2. Support <ClusterID>/actions/resize API https://review.openstack.org/#/c/638572/
21:14:50 <strigazi> I had a look and it works just fine. The only part we miss is the allow resize on update failed.
21:15:30 <strigazi> if you can have a look in the patch, it would be great.
21:15:42 <flwang> strigazi: does the resize for cluster in update_in_progress work for you ?
21:15:56 <strigazi> flwang:  right, that too :)
21:16:19 <flwang> strigazi: i will propose a new patch set, thanks
21:16:32 <strigazi> flwang: no, didn't work, he request goes in but doesn't do anything. (that was 24h hours ago)
21:16:53 <flwang> strigazi: ok, i will test it again to figure it out
21:17:03 <flwang> and i will leave a comment on the patch
21:17:34 <strigazi> flwang: you will also have a look in resize on UPDATE_FAILED?
21:17:56 <strigazi> we also need a client patch
21:18:28 <flwang> strigazi: sure, i will
21:18:37 <flwang> i will propose a patch for actions today
21:18:55 <flwang> or this week at least
21:19:42 <strigazi> ttsiouts: flwang, we will use resize for NGs? What do you think? I think it makes sense.
21:20:04 <ttsiouts> yeah.. I think it does...
21:20:34 <flwang> strigazi: that's on my to-do list, i will add a  todo comments in resize patch, i have already put a node group param there for placeholder
21:21:31 <flwang> we just need to make sure the API is what we want and we can support NG as long as it's landed
21:21:46 <ttsiouts> flwang: I will also try to review
21:21:57 <flwang> ttsiouts: it will be great, thanks
21:22:57 <strigazi> cool
21:23:23 <strigazi> 3. (WIP) Add cluster upgrade to the API https://review.openstack.org/#/c/514959/
21:24:03 <strigazi> I tested it with the new PS, looks good, I'm using it for the driver implementation, thanks flwang
21:24:17 <flwang> strigazi: cool
21:25:01 <flwang> team, please help review the upgrade api  https://review.openstack.org/#/c/514959/
21:25:18 <flwang> that's one of most important feature we'd like to get it in Stein
21:26:32 <strigazi> any comments/questions about it?
21:27:00 <flwang> the only tiny issue now is if user is using the default/built-in tag/version in heat template,  we can't get it to raise 400 error if user is doing downgrade accidentally
21:27:35 <flwang> but personally i think it's OK at this stage
21:27:41 <strigazi> +1
21:28:01 <flwang> because generally the template is proposed by cloud admin and they will be very careful for new template publishing
21:28:51 <jakeyip> (what do I, as a cloud admin, have to be careful about) ?
21:29:05 <strigazi> If we find a way to separate labels that are for tags and labels that are switches, to enable/disable features we are good
21:29:24 <strigazi> jakeyip: version compatibility
21:29:55 <strigazi> jakeyip: not make big jumps that will have big diff between cluster tempaltes
21:30:51 <flwang> strigazi: you just reminder me a feature, we should support enable an addon after cluster created
21:31:31 <flwang> strigazi: we should re define our label name convention
21:32:05 <flwang> using consistent pattern, which can help mitigate the issue you mentioned above
21:32:10 <jakeyip> I'm looking at the story at https://storyboard.openstack.org/#!/story/2002210 but I'm getting a "Not Found" for the gerrit topic links e.g. https://review.openstack.org/#q,topic:bp/cluster-upgrades,n,z is it just me?
21:32:12 <strigazi> not difficult to do, with the upgrade patch it shoudl be doable.
21:32:54 <strigazi> https://storyboard.openstack.org/#!/story/2002210
21:33:48 <strigazi> https://review.openstack.org/#/q/topic:cluster-upgrades+(status:open+OR+status:merged)
21:36:28 <strigazi> any more comments?
21:38:37 <flwang> let's move to next one
21:38:42 <jakeyip> I'm ok for now, still have lots of questions about the different edge cases but I'll wait for it to land first
21:38:50 <strigazi> flwang: do you have something?
21:39:32 <flwang> strigazi: a little bit.   1. I'm running for the Train cycle PTL
21:39:45 <strigazi> flwang++
21:40:56 <flwang> thank you for the support :D and i do need your support for the following release
21:41:10 <strigazi> no problem
21:41:14 <flwang> 2. Stein will be released soon https://releases.openstack.org/stein/schedule.html
21:41:44 <flwang> next week is RC1, we should be quick to get things done with good quality which is a challenge
21:41:54 <flwang> since we still have quite a lot on the table
21:42:25 <strigazi> ack
21:43:15 <flwang> 3. Fedora CoreOS
21:43:33 <flwang> instead of upgrading to Fedora Atomic 29, we probably should go for Fedora CoreOS 29
21:43:38 <flwang> strigazi: thoughts?
21:43:52 <flwang> or do both for a while
21:44:27 <flwang> strigazi: for Fedora CoreOS 29, do you think we need to change a lot if we copy the code from fedora-atomic driver?
21:44:45 <strigazi> we should stick to one
21:45:09 <strigazi> now that the heat agent does a lot of things, it should be doable
21:45:23 <strigazi> we just need to move some things to ignition
21:45:40 <flwang> cool
21:45:47 <strigazi> not sure with a timeline of one week, maybe two or three
21:45:51 <jakeyip> I thought they are only merging in Fedora CoreOS 30? is there a fedora coreos 29 image already?
21:45:58 <strigazi> 29 is bera
21:46:00 <strigazi> 29 is beta
21:46:34 <strigazi> https://ci.centos.org/artifacts/fedora-coreos/prod/builds/latest/
21:47:37 <jakeyip> I see. that's a good image for testing :)
21:47:51 <flwang> that's all from me
21:48:08 <strigazi> yeap, if you want to give feedback this is the place or #fedora-coreos
21:48:24 <strigazi> https://github.com/coreos/fedora-coreos-tracker/
21:48:42 <jakeyip> anyone testing that out yet?
21:49:09 <strigazi> I booted a couple of vms, but not much, didn't have time
21:49:19 <jakeyip> we have users asking for coreos support, current answer is wait for fedora coreos 30
21:49:36 <strigazi> a few months left
21:50:15 <strigazi> since we are running out of time
21:50:25 <strigazi> alisanhaji: you wanted to bring smth up?
21:51:01 <alisanhaji> yes
21:51:23 <alisanhaji> I am interested in having Kuryr-kubernetes supported with Magnum
21:52:09 <alisanhaji> as a first step to gettting the clusters deployed by magnum communicating with the OpenStack that created them
21:52:32 <strigazi> that is working already
21:52:36 <alisanhaji> the goal is to have containers and VMs communicating in the same networks
21:53:15 <alisanhaji> is kuryr a network driver already? I only see calico and flannel
21:53:26 <strigazi> alisanhaji: does kuryr require trunk ports?
21:53:44 <strigazi> no it is only calico and flannel at the moment
21:54:16 <alisanhaji> yes when using VMs, but you can also run kuryr without trunk ports
21:54:28 <alisanhaji> but it requires a neutron-agent in the k8s node
21:54:31 <strigazi> alisanhaji: do you want to propose a patch?
21:55:02 <alisanhaji> I was wondering if I needed to submit a blueprint or RFE in launchpad
21:55:25 <strigazi> alisanhaji: https://storyboard.openstack.org/#!/project/openstack/magnum
21:55:37 <flwang> alisanhaji: storyboard
21:55:51 <strigazi> if the patch look like this https://github.com/openstack/magnum/blob/master/magnum/drivers/common/templates/kubernetes/fragments/calico-service.sh it should be easy
21:56:52 <alisanhaji> Thanks, I will see how kuryr can be deployed like this
21:57:01 <alisanhaji> are trunk ports a problem with magnum?
21:57:29 <strigazi> it might be complicated, it would better if it can work without
21:58:12 <flwang> using trunk ports may need more change for code and heat template, which is complicated and we'd like to avoid
21:58:20 <alisanhaji> I see, in this case a neutron-agent should be installed to the k8s nodes, and talk to neutron-server
21:58:28 <alisanhaji> ah ok
21:58:41 <strigazi> alisanhaji: agent in the nodes works better for magnum
21:59:22 <strigazi> anything else? time is up?
21:59:25 <alisanhaji> Ok, thanks! And what about having OVN as a network driver, I am thinking about integrating it too to Magnum
22:00:10 <flwang> alisanhaji: feel free to propose a story and spec
22:00:18 <flwang> we can start to discuss from there
22:00:35 <alisanhaji> flwang great:
22:00:40 <alisanhaji> !
22:00:54 <strigazi> alisanhaji: go for it, the network drivers are pretty well encapsulated
22:01:12 <strigazi> CNI is great
22:01:27 <alisanhaji> yes it is, thanks strigazi and flwang
22:02:05 <strigazi> thanks for joining the meeting everyone!
22:02:12 <jakeyip> thanks strigazi, flwang!
22:02:18 <strigazi> see you around
22:02:20 <strigazi> jakeyip: cheers
22:02:36 <flwang> strigazi: thank you
22:02:46 <strigazi> #endmeeting