16:00:25 #startmeeting containers 16:00:26 Meeting started Tue Jan 30 16:00:25 2018 UTC and is due to finish in 60 minutes. The chair is strigazi. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:27 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:00:29 The meeting name has been set to 'containers' 16:00:31 #topic Roll Call 16:00:44 hi 16:00:49 hi 16:01:01 hello slunkad vabada 16:01:32 0/ 16:02:00 hi oikiki 16:02:35 hi strigazi 16:03:16 Meeting agenda: 16:03:19 #link https://wiki.openstack.org/wiki/Meetings/Containers#Agenda_for_2018-01-30_1600_UTC 16:03:29 #topic Announcements 16:03:41 Next week on Thursday we will branch, push for your changes 16:04:58 that's for the queens release, right? 16:05:14 vabada: yes 16:05:16 what does that mean? 16:05:30 So, what this mean for the three of you that join the meeting 16:05:36 s/mean/means 16:06:28 For slunkad, it would be nice to take the suse driver in tree, some working flavor of it, so people can have an opportunity to test 16:07:03 For oikiki, the two first tasks for the periodic tasks are needed to get merged 16:07:42 ah ok 16:07:59 oikiki: the first one, to make the send_metrics task is very important due to a bug in the kubernetes client, I'll explain later 16:08:06 strigazi: yeah I made some progress on the image too so hope we can manage in time 16:08:24 for vabada, try to push the patch to drop the ironic driver 16:08:53 slunkad: cool 16:09:45 let's move one, we can discuss this later again 16:09:53 I'm working on it. I'll ask some questions later after the meeting regarding the work done by Mathieu Velten 16:10:01 ok 16:10:07 #topic Review Action Items 16:10:14 strigazi to look for a new meeting time APAC friendly [incomplete] 16:12:06 I will try to make this meeting tha last one in this time 16:12:34 #action strigazi to send email for the new meeting time 16:12:43 #topic Blueprints/Bugs/Ideas 16:13:01 kubernetes python client and eventlet incompatibility 16:13:08 #link https://github.com/eventlet/eventlet/issues/147 16:13:35 After kubernetes client 4.0.0 there is the above incompatibility issue ^^ 16:14:34 Magnum has three periodic tasks, one to sync the magnum service, one to update the cluster status and one send cluster metrics 16:15:09 The send_metrics task uses the kubernetes client for kubernetes clusters and it crashes 16:15:46 The really big problem is that it also crashes the cluster status sync task 16:16:26 This why, we need the patch from oikiki, to be able to diasble the task until the problem is fixed. 16:16:51 i see. 16:16:52 I got that error today in devstack. Count on me for any testing that needs to be done 16:17:11 Best case scenario is to fix eventlet to play well with the multiproccessing module 16:17:19 vabada: you can test oikiki 's patch 16:17:27 oikiki: can you fetch the link? 16:17:36 sure one se 16:17:52 sure. Count on me oikiki 16:18:03 thanks vabada: https://review.openstack.org/#/c/529098/ 16:19:24 if any of you want's to dive in eventlet, is more that welcome :) 16:20:21 Finally, an option for us would be to use the api with python requests 16:20:40 or even the kubectl binary 16:21:24 or not use eventlet, this quite a change 16:21:28 ok next 16:22:13 Fix for multimaster clusters: 16:22:18 #link https://bugs.launchpad.net/magnum/+bug/1742420 16:22:20 Launchpad bug 1742420 in Magnum "Pod's can't communicate with kube-apiserver in multi master cluster" [Undecided,In progress] - Assigned to Spyros Trigazis (strigazi) 16:22:37 this line seem to fix the issue https://review.openstack.org/#/c/533593/12/magnum/drivers/common/templates/kubernetes/fragments/configure-kubernetes-master.sh@26 16:23:05 vabada: oikiki slunkad Could you test it? 16:23:44 I'll leave comments on the bug for how to test, 16:23:45 sure 16:24:11 basically, you need a 3-master cluster and check if coredns and the dashboard are working 16:24:27 that's great. I was thinking something like that 16:24:30 I'll do it 16:24:55 ok 16:25:10 Two final items from me 16:26:14 I will try to push a lot for a first itetation on cluster federation and cluster upgrades, so I will ask for help on testing 16:26:21 #link https://review.openstack.org/#/q/status:open+project:openstack/magnum+branch:master+topic:bp/federation-api 16:26:42 #link https://review.openstack.org/#/q/status:open+project:openstack/magnum+branch:master+topic:bp/cluster-upgrades 16:27:11 That was it from me, let's move to 16:27:14 #topic Open Discussion 16:27:40 oikiki: did you have time to look into the swarm-mdoe monitoring? 16:27:46 great, count on me for testing. If there are some guidelines, I'd be happy to help. 16:27:55 vabada: thanks 16:28:31 strigazi: i started on it but have been sick for the last few days... could we talk about it in a day or so. my progress stalled and i cant think so well today. 16:28:44 oikiki: sure 16:28:51 strigazi: i think im going to have to go back to bed after meeting :( 16:29:02 oikiki: :( 16:29:11 strigazi: i'll definitely help out on testing as well tho 16:29:31 strigazi: https://review.openstack.org/#/c/520063/ this is passing since I switched to using opensuse as the base image but I'm not entirely sure it is doing what it needs to 16:29:43 oikiki: get well soon! 16:29:59 slunkad: thank you! 16:30:06 get well oikiki 16:30:12 slunkad I saw it passing but it doesn't build anything 16:30:18 yeah 16:30:18 thanks vabada! 16:30:41 cuz locally I get errors with the build 16:32:05 slunkad: are there any examples on other suse images built with diskimage-builder? 16:32:26 maybe AJaeger could help 16:32:53 strigazi: I asked him but he doesn't really know 16:33:11 ok, let's take a look after 16:33:19 and I think there should be an example but I haven't found it yet 16:33:24 ok great 16:34:04 vabada: about ironic, any luck with devstack? 16:34:59 not really. Have tried different environments, and even though cloud-init finishes, the heat WAIT_CURL API calls fails due to authentication failed. 16:35:21 in the master node, so cluster creation fails due to timeout 16:35:42 vabada did you try with a custom template that only signals heat? 16:36:05 no, I could do that. You mean running it directly from heat? 16:36:14 we can try to narrow it like this, 16:36:15 yes 16:36:30 ok. good point. I'll do that 16:37:07 take kubemaster.yaml and remove almost everything, we need only to test wait_curl 16:37:37 ok, yes I did that once for another error, should be easy 16:37:41 thanks strigazi 16:37:49 @all any other business? 16:38:51 said twice 16:39:02 Thanks slunkad oikiki vabada 16:39:11 #endmeeting