16:23:13 #startmeeting openstack_ansible_meeting 16:23:14 Meeting started Tue Jul 7 16:23:13 2020 UTC and is due to finish in 60 minutes. The chair is noonedeadpunk. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:23:15 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:23:17 The meeting name has been set to 'openstack_ansible_meeting' 16:23:25 #topic office hours 16:23:27 hi everyone! 16:24:37 so regarding centos8 - only these patches are blockers: https://review.opendev.org/#/c/739026/ and https://review.opendev.org/#/c/735926/ 16:24:38 * jrosser waves 16:25:07 are you happy it's all working as we didnt make any NV jobs for it 16:25:27 spatel also placed patches for most of the services (didn't check if for all of them) which is really cdool! 16:25:40 yes that is awesome :) 16:30:15 not sure if really all, as spatel's patches don't depend on centos integrated repo 16:30:21 so they're not tested at the moment 16:30:41 SO id' say to merge them once we merge integrated repo 16:31:23 also we didn't do releases for ages, so I'd say let's merge https://review.opendev.org/#/c/738357/ and https://review.opendev.org/#/c/738689/ 16:31:32 and really leave centos8 for the next rc 16:31:46 even if we make it next week 16:32:31 regarding train I guess I bumped all backported keystone stuff 16:39:49 so I think we need also to do tha same for Ubuntu focal where required 16:39:52 excellent - that will reduce our number of patches overhead 16:40:16 but tbh we are quite on top of that so i don't think delaying train release should be blocked on the keystone stuff 16:40:17 btw centos patches are good to see if focal is passing as well :p 16:40:37 i was looking at the graphs a bit too http://grafana.openstack.org/d/4-pSHcImz/openstack-ansible?orgId=1&from=now-7d&to=now 16:40:54 yeah, I think we're ready to merge train, so let's go aahead 16:41:04 i can almost convince myself they are going down toward the right hand end, but its hard to tell 16:41:41 so failure rate has grown up 16:42:59 but totally not two times... 16:43:05 oh, btw, regarding jobs 16:43:15 I think we should somehow decrease number of jobs 16:43:36 and eventually we will need to clean up experimental pipeline (and clean it) 16:44:02 so maybe we should leave just bionic metal for master, and move rest to the experimental 16:44:05 as soon as we merge centos-8 (and release U) we can remove the centos-7 jobs 16:44:33 and slim down bionic yes 16:44:44 and maybe consider experimental not as "experimental" but as smth that should work and we don't check each time? 16:45:09 maybe there's more appropriate pipeline for that... 16:45:35 we put lots of cpu cycles into periodic jobs, but don't look at them 16:45:55 if it was a monorepo it would be easier, but as there are soooooo many jobs its tough 16:45:56 tbh me too... 16:46:17 i cannot think of an easy way to make a grafana for this 16:46:30 so we can drop them I guess. Btw, once you said about periodics, we don't update docs on the regular basis. 16:46:54 I can some distant memories aboult talk with infra folks regarding it and it was not so easy to fix 16:47:43 so we like get release tag (ie 20.0.0) left for the rest of the release cycle, so our deployment guide is not really relevant... 16:48:11 the easy way to change doc and post the way to get latest tag for the release. 16:49:29 oh, btw, glance registry finally got removed from glance code 16:49:45 and we have https://review.opendev.org/#/c/739656/ to cover that.. 16:51:05 but periodics might be pretty useful for stable branches... if only we looked at them... 16:52:44 btw I guess I know how to draw periodics on grafana if only Grafyaml support variables... 16:53:27 eventually target can be regexp I guess. so this should be an issue 16:54:01 but without access to data so uncomfortable to drow just in yaml... 16:59:59 Is there an OSA configurable compute load balancer? I found "watcher" but I also found an OS page that says it's no longer maintained. I want a service that automatically balances the compute nodes based on resource contraints 17:07:47 I guess you need MetricsWeigher from nova scheduler.... 17:07:58 https://docs.openstack.org/nova/latest/user/filter-scheduler.html#weights 17:12:26 noonedeadpunk: There's nothing that will automatically live-migrate instances? 17:14:16 oh, you mean that 17:15:18 noonedeadpunk this is my neutron.conf http://paste.openstack.org/show/795637/ and this is yours http://paste.openstack.org/show/795638/ 17:15:25 I dont find anything 17:15:53 #endmeeting