15:00:13 #startmeeting kolla 15:00:13 Meeting started Wed Apr 14 15:00:13 2021 UTC and is due to finish in 60 minutes. The chair is yoctozepto. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:14 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:17 The meeting name has been set to 'kolla' 15:00:59 #topic Roll-call 15:01:02 \o/ 15:01:03 Michal Arbet proposed openstack/kolla-ansible master: Add mariadb arbitrator to mariadb role https://review.opendev.org/c/openstack/kolla-ansible/+/780811 15:01:04 Michal Arbet proposed openstack/kolla-ansible master: [CI] Test Mariadb-Arbitrator with shards in the nova cells scenario https://review.opendev.org/c/openstack/kolla-ansible/+/780970 15:01:26 o/ 15:01:30 o/ 15:03:39 it's not quantity but quality that counts they say 15:03:45 #topic Agenda 15:03:56 * Roll-call 15:03:56 * Announcements 15:03:56 ** mgoddard on holiday this week 15:03:56 ** Xena PTG next week! https://etherpad.opendev.org/p/kolla-xena-ptg 15:03:56 ** OpenStack User Survey 2020 is in! https://www.openstack.org/analytics 15:03:57 * Review action items from the last meeting 15:03:57 * CI status 15:03:58 * Review requests 15:03:58 * Wallaby release planning 15:03:59 ** Debian bullseye progress 15:03:59 * Xena PTG planning https://etherpad.opendev.org/p/kolla-xena-ptg 15:04:22 #topic Announcements 15:04:31 mgoddard on holiday this week 15:04:43 all wish him a good rest; he deserves it ;-) 15:04:59 Xena PTG next week! https://etherpad.opendev.org/p/kolla-xena-ptg 15:05:13 yeas! it's this time of the year again 15:05:40 grab your entry cards and join us on the ptg 15:05:58 and add discussion topics to the etherpad 15:06:07 it's this week where you end up exhausted like after running a marathon? ;) 15:06:21 mnasiadka: yeah, I already feel it in my bones, ouch 15:06:37 but it's worth the effort! :D 15:06:58 #info OpenStack User Survey 2020 is in! https://www.openstack.org/analytics 15:07:16 15% of deployments are powered by kolla 15:07:21 kolla stronk 15:07:56 17% of deployments on Stein? :) 15:08:16 and Queens 15:08:17 yay 15:08:21 yeas ;-) 15:08:30 many sad numbers there 15:09:05 2% of people using Kolla-Kubernetes 15:09:08 good one 15:09:29 more masakari than blazar priteau 15:09:37 * yoctozepto feels the pressure 15:09:50 yeah, kolla-kubernetes kills the system 15:09:53 oh, TripleO only used in 10% deployments and K-A in 15% 15:09:59 that's interesting 15:10:33 yeah, I wonder what those using RHOSP chose 15:10:43 some might be remembering the good ol' days of puppet 15:10:59 Which OpenStack Network (Neutron) drivers are you using? - this is my readability winner 15:11:25 well, and pure "ansible" is interesting as the deployment choice 15:12:12 yeah, re neutron I have no idea who looked at this and thought: yeas, it's perfect to display ;D 15:12:41 only table makes sense for that 15:12:46 and sorted by pop 15:12:48 but meh 15:13:25 well, anyhow 15:13:30 ubuntu used 40% 15:13:32 in storage ceph is the king 15:14:08 nothing new 15:14:08 wuchunya_: yeah but it's for 2020 and centos+rhel = 47 15:14:13 but hard to tell if they overlap or not 15:14:15 perhaps they do 15:14:28 nowadays ubuntu would probably be getting a boost 15:14:41 many simply decided to migrate to ubuntu 15:15:15 for lolz: more windows than suse 15:15:39 all right, enough dwelling on stats 15:15:48 #topic Review action items from the last meeting 15:16:23 mgoddard email openstack-discuss about quay.io credentials 15:16:23 he did not 15:16:23 #action mgoddard email openstack-discuss about quay.io credentials 15:16:26 yoctozepto deprecate enable_host_ntp 15:16:31 he did not either 15:16:35 #action yoctozepto deprecate enable_host_ntp 15:16:57 #topic CI status 15:17:13 I was not angry at CI so it must be pretty green 15:17:23 I like quay so far 15:17:32 it only failed once since 15:17:46 move to Bullseye needs mariadb work 15:17:54 ^ that has happened already 15:17:55 removing 15:18:20 yep, we're waiting for hrw to test on aarch 15:18:45 nobody cares about monasca-grafana I guess? 15:19:19 seems so 15:19:35 are we deprecating in W or in X? 15:19:41 kayobe looks green too 15:19:54 mnasiadka: would ask dougsz 15:20:01 but I love to deprecate early 15:20:06 and drop early :-) 15:20:16 it's a monstrosity anyway 15:20:22 Will ask when I'll see him online 15:20:25 * yoctozepto feels bad recommending it 15:21:21 https://review.opendev.org/c/openstack/kolla/+/784901 15:21:27 we have this, seems there's consensus 15:21:47 mnasiadka: does your cephadm fix helps with that random issue mentioned on the whiteboard? 15:22:04 yoctozepto: well, I haven't seen it in there, but I'll do a couple of rechecks to be sure. 15:22:19 mnasiadka: thanks 15:22:25 let me check that monasca-grafana 15:23:30 +2 15:25:13 mnasiadka: you +2 too? 15:25:56 sure, I'll try to ask Doug if he needs help in K-A deprecation 15:26:03 #topic Review requests 15:26:07 mnasiadka: very well, thanks! 15:26:26 what do you want reviewed? 15:26:40 your favourite plant? 15:26:45 your favourite meal? 15:26:50 https://review.opendev.org/c/openstack/kolla-ansible/+/741340/ 15:26:53 I'm here to grant you your wish 15:27:12 ah yes 15:27:15 +966 15:27:15 -18 15:27:25 I also enjoy orchids 15:27:29 and now I know why it is still not merged :-) 15:27:42 orchids can be pretty 15:28:33 headphoneJames: well, my main issue is with supervisord (so multiple processes) in haproxy image... can we do anything to not do it? 15:29:40 I'm open to other approaches. 15:30:07 I'll try to look into, but that might happen in two weeks time... 15:30:08 The issue is that we need a process that monitors for new certificates, and causes an haproxy reload 15:30:52 or we need a newer haproxy, which is a crazy thing to do - because we would need additional repos... 15:30:58 I have a suspicion this is not going in the current release, so two weeks should be fine 15:31:05 morning 15:31:11 you might be right, headphoneJames, I am sorry 15:31:19 I really wished to get this in this cycle :/ 15:31:39 we could run a separate container with shared ipc namespace 15:31:42 s'all good 15:31:49 headphoneJames: thought about that? 15:32:22 I certainly have not. I would have to learn with that means ;) 15:32:38 headphoneJames: https://docs.docker.com/engine/reference/run/#ipc-settings---ipc - look at shareable 15:32:50 probably we could then restart haproxy from another container 15:33:13 we could send a signal 15:33:15 Is there a less automated version that could be merged in wallaby, with a manual command to update certificates and reload haproxy? 15:33:18 but what happens when you restart haproxy and if it joins the same ipc group, that is a mystery :) 15:33:20 but that would need share pid namespace, no? 15:33:42 yoctozepto: but it doesn't need to be host namespace, it seems in Docker you can create a ,,shareable'' one 15:33:43 priteau: yeah, my thoughts exactly 15:33:51 I remember we discussed some layering to this 15:34:05 mnasiadka: I wonder what the lifetime is ;-) 15:34:10 as you need that PID 1 15:34:26 it would be fugly to have either break the other 15:34:37 perhaps it's better to push newer haproxy for this 15:34:41 yoctozepto: that's why I mentioned a mystery, meaning it would need some degree of testing 15:34:47 headphoneJames: do you have an answer for priteau? 15:34:50 yoctozepto: both centos and ubuntu would need a repo 15:35:05 even current ubuntu? doh 15:35:07 yoctozepto: which in centos case there's a nfv repo - but for Ubuntu it's just somebody's PPA 15:35:34 we need 15:35:35 2.2? 15:36:05 We can definitely push a version that doesn't do the automatic restart, and would require customers to restart ha proxy manually 15:36:44 Merged openstack/kolla master: centos: enable repos where needed https://review.opendev.org/c/openstack/kolla/+/785871 15:36:45 yoctozepto: ideally 2.2 15:36:53 However, the test case would need to then manually restart haproxy 15:37:01 headphoneJames: let's do it in another change then 15:37:05 perhaps we can merge that 15:37:09 mnasiadka: meh 15:37:11 headphoneJames: well, how do we notify them to restart haproxy? :) and how often do they need to do that? 15:37:36 They would have to check logs :) 15:37:38 yoctozepto: yeah, so centos nfv repo and ubuntu PPA https://launchpad.net/~vbernat/+archive/ubuntu/haproxy-2.2 - no other option with Focal 15:38:11 Keep in mind supervisord is only deployed and used with the let's encrypt option enabled 15:38:30 And works really well over several months of testing 15:38:54 I don't doubt it does work, but we don't run more than one process in any container image :) 15:39:26 Anyhow, my expectations are properly set for next release 15:40:21 I made a few updates to https://ethercalc.openstack.org/kolla-infra-service-matrix 15:40:28 I assumed we move to bullseye 15:41:05 headphoneJames: how long will it take you to simplify it as we discussed? 15:42:26 I can easily remove the supervisord part. 15:42:50 I will look into the shared ipc today 15:43:11 mnasiadka: don't look at the bifrost image then ;-) 15:43:29 priteau: bifrost is on the death path 15:43:41 Having customers manually restart haproxy is probably not the right approach 15:43:44 just no executioners around 15:43:59 headphoneJames: yes, it's not 15:44:22 but I have a feeling we are better off merging it in parts rather than postponing till the end of the world 15:45:29 Let me see how far I get today. I'm happy to discuss breaking it into smaller chunks 15:45:52 all right, ping us whenever 15:46:11 ooh, I have a last minute announcement 15:46:17 #topic Last minute announcements 15:46:31 #info OpenStack Wallaby released! yahoo! 15:46:44 yay! 15:47:04 priteau: bifrost is a monster 15:47:05 so when we branch? 15:47:12 that's the question for ptg 15:47:30 #topic Wallaby release planning 15:47:33 I think we can branch Kolla once we merge all hrw's patches ;) (UCA/Debian/RDO) 15:47:35 Debian bullseye progress 15:47:42 I think it's only pending a review 15:47:56 https://review.opendev.org/c/openstack/kolla/+/772479 15:48:05 I pasted it on the whiteboard already 15:48:08 anything to discuss? 15:49:30 well, since Debian is now on good path, I guess it's time to look into UCA/RDO changes to get them merged and images built 15:49:36 I will read meeting log :( 15:49:50 I'll try to look into those 15:50:18 UCA is upgrade V->W issue iirc 15:50:43 upgrade is always a release change issue :) 15:51:39 ah, looking for somebody brave enough to merge this: https://review.opendev.org/c/openstack/kolla-ansible/+/783669 15:53:17 * hrw drops - too many lags and reconnects today to be able to follow discussion 15:53:21 upc-- 15:53:35 yoctozepto: anything else left on agenda? because it seems I'm speaking with myself... 15:53:43 check the recorded logs hrw 15:54:02 mnasiadka: open discussion 15:54:13 ok then 15:54:15 #topic Xena PTG planning https://etherpad.opendev.org/p/kolla-xena-ptg 15:54:30 it was mentioned already 15:54:37 please add yourself and your discussion topics 15:55:46 #topic Open discussion 15:56:06 mnasiadka wanted to participate in open discussion 15:56:29 naah, I speak too much and mainly to myself, it doesn't sound like an open discussion ;) 15:56:38 and we will have too much time for discussions next week 15:57:46 mnasiadka: well, it's open-ended then! 15:57:53 but you are right 15:58:10 we are all tired simply by thinking how tired we will be after the ptg 15:58:16 that said 15:58:19 thank you for attending 15:58:28 #endmeeting