14:00:16 #startmeeting kolla 14:00:16 Meeting started Wed Sep 14 14:00:16 2022 UTC and is due to finish in 60 minutes. The chair is mnasiadka. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:16 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:16 The meeting name has been set to 'kolla' 14:00:20 #topic rollcall 14:00:32 \o 14:00:46 (°Y.) 14:01:02 o/ 14:01:08 Rafal Lewandowski proposed openstack/kayobe master: updated write_bifrost_clouds_yaml logic https://review.opendev.org/c/openstack/kayobe/+/857594 14:02:03 o/ 14:03:11 #topic agenda 14:03:11 * CI status 14:03:11 * Release tasks 14:03:11 * Regular stable releases (first meeting in a month) 14:03:11 * Current cycle planning 14:03:13 * Additional agenda (from whiteboard) 14:03:13 * Open discussion 14:03:15 #topic CI status 14:03:26 whiteboard says green 14:03:45 it is 14:03:47 but I noticed k-a jobs that depend on kolla patch are failing on c9s and rl9 14:03:56 I have also run today all jobs on all supported branches 14:04:16 message:"[DNM] [CI] Run it all" status:open 14:04:22 or: 14:04:23 https://review.opendev.org/q/message:%2522%255BDNM%255D+%255BCI%255D+Run+it+all%2522+status:open 14:04:32 anyway CI green 14:04:36 #topic Release tasks 14:04:36 o/ 14:05:14 o/ 14:05:21 next week feature freeze 14:05:54 opensearch and rocky 9 related will probably need an exception 14:06:08 #topic Current cycle planning 14:06:11 speaking of the devil 14:06:28 I worked on opensearch, still in progress 14:06:35 missing opensearch-dashboards deployment 14:06:42 but opensearch with fluentd works properly 14:07:01 hrw: how are we with cs9 Kolla patch? ready to merge? 14:07:50 Bartosz Bezak proposed openstack/kayobe master: Move to CentOS Stream 9 / Rocky Linux 9 https://review.opendev.org/c/openstack/kayobe/+/855656 14:07:55 https://review.opendev.org/c/openstack/kolla/+/836664 14:08:04 mnasiadka: I would say that it is and then needs some improvements on top 14:08:19 ELK and Grafana should be fixed by OpenSearch 14:08:29 we use cron for logrotate but there is no cron in el9 14:08:45 and what's the replacement? 14:08:50 systemd timers 14:08:56 fantastic 14:08:57 which we lack in container 14:09:42 then we would need to run a container with systemd and use timers 14:09:47 seems like a followup to the original change 14:09:57 mnasiadka: or the host one 14:10:02 or use host one 14:10:09 yes, that's another option 14:10:23 install logrotate in host OS and point to kolla logs 14:10:41 Bartosz Bezak proposed openstack/kayobe master: Move to CentOS Stream 9 / Rocky Linux 9 https://review.opendev.org/c/openstack/kayobe/+/855656 14:10:46 and then we can deprecate logrotate image 14:11:00 ok then 14:11:24 hrw: well, we need logrotate somewhere 14:11:33 host systemd could run the container 14:11:55 I guess for now it would be easier to introduce systemd in that container? 14:12:03 it might be 14:12:28 yoctozepto: we keep logs on host OS disk so why host's logrotate can not handle it for us? 14:12:47 hrw: no idea 14:12:57 or look for some small alternatives 14:13:11 present in el9 or epel9? 14:13:31 nobody says it needs to be present there, if it's a small Go binary or similar, we can just curl it 14:13:44 cron go? 14:14:17 gocron, clockwork, others, anything :) 14:14:19 mnasiadka: "while true; do logrotate; sleep 1h;done"? 14:14:45 if we move to systemd this cycle - we can use systemd timer on logrotate container :) 14:15:09 CS9 images worked with systemd as init 14:15:23 yes, but if we need that, it should be an exception, not standard 14:15:24 hrw's loop might work well enough 14:15:51 or using host OS logrotate 14:16:15 ok, so - discussion is fine - is there anybody that wants to look into making that work? 14:16:16 each of them has it. and runs it somehow 14:17:15 yeah, I guess kolla might have overcontainerised some services 14:17:58 mnasiadka: assign it to me 14:18:25 #action hrw work on cron alternative 14:18:28 ok then 14:18:42 Rafal Lewandowski proposed openstack/kayobe master: updated write_bifrost_clouds_yaml logic https://review.opendev.org/c/openstack/kayobe/+/857594 14:18:42 so second thing around CS9 Kolla patch - in Zed PTG we decided to not publish cs9 images 14:19:03 and run cs9 jobs as non-voting or experimental to see what's ahead 14:19:23 are we changing that? 14:19:36 I would keep it that way - no publishing 14:19:57 ok, so the patch needs to be updated I guess - there are publish jobs defined 14:19:59 we merge them very late in a cycle so not tested as much yet as usual stuff was tested 14:20:21 well, I guess we want to publish rocky9 images instead - that was the plan 14:20:37 once we have rl9 merged working 14:20:37 maybe not initially 14:20:40 yeah 14:20:51 I'll push a patch based on cs9 one to add rl9 14:20:57 ok that is solved 14:21:03 bbezak: how's it going with rl9 for Kayobe? 14:21:43 lovely :). Biggest hurdle till now was dummy interfaces in network manager with network-interfaces role 14:22:00 but looks promising 14:22:14 so basically network-scripts missing in EL9 14:22:20 nice 14:22:38 yoctozepto: I noticed you started some reviews on systemd approach - do you think we can merge it this cycle? 14:23:00 mnasiadka: yes, it seems pretty ready; I would go with an exception for it 14:23:02 can review 14:23:14 however there is compatibility layer NetworkManager-initscripts-updown - https://www.redhat.com/en/blog/rhel-9-networking-say-goodbye-ifcfg-files-and-hello-keyfiles 14:23:16 I just can't commit much time this and the next week 14:23:20 ok, is there any other core that has cycles to review? 14:23:29 because we need a second one 14:23:38 in other words 14:23:50 any other core to review this by the end of September? 14:23:56 yoctozepto: sure, I would probably feel more comfortable to merge it on beginning of Antelope, but it's been hanging there for some time 14:24:05 same thinking here 14:24:25 maybe it's more motivating to merge at the last minute :D 14:25:05 ok, enough about that 14:25:22 Let's Encrypt is also eager to get some reviews, but maybe we need to postpone it as well 14:25:35 ok, let's go forward 14:25:38 #topic Additional agenda (from whiteboard) 14:25:57 (yoctozepto) systemd container control https://review.opendev.org/c/openstack/kolla-ansible/+/816724 14:26:02 already discussed 14:26:05 ok 14:26:10 please help review, seems reasonable 14:26:21 (mnasiadka) Monasca status 14:26:47 I think hrw mentioned we've had multiple issues with Monasca (not buildable on Jammy, now buildable, similar situation on EL9 I think) 14:26:47 monasca... napalm or thermite? 14:27:04 and then somebody would need to make it work with OpenSearch 14:27:12 we have monasca buildable on x86-64 in Zed 14:27:34 will it be buildable in next cycle? probably after half of cycle 14:27:38 I think we could deprecate this cycle, and if nobody steps up - remove next cycle together with removal of Elasticsearch 14:27:44 +3 14:28:11 +2 14:28:27 anybody wants to raise the patch and send a mail to openstack-ml or should I? 14:28:39 and monasca is yet-another-monitoring-as-a-service stuff 14:30:13 ok, so let me do that 14:30:29 #action mnasiadka deprecate Monasca and send mail to ML asking for contributors to step up 14:30:53 +1 (non core) :) 14:30:59 (yoctozepto) Topics for the PTG (I propose a discussion of Zens of Kolla) 14:31:25 #link https://etherpad.opendev.org/p/kolla-antelope-ptg 14:31:28 I created that 14:31:54 please add topics for the PTG 14:32:03 we also should book an operator hour it seems 14:32:22 should it be last hour of Tue session? 14:32:57 ok 14:33:07 ok, will arrange 14:33:14 #topic Open discussion 14:33:17 any open topics? 14:33:46 I sent patch to use shutil.copytree() better 14:33:59 have one to send which moves from os.path to pathlib 14:34:18 please do 14:34:19 nice, will review 14:34:36 Marcin Juszkiewicz proposed openstack/kolla master: Switch from os.path to pathlib https://review.opendev.org/c/openstack/kolla/+/857720 14:34:46 ok then, let's finish for today 14:34:49 thanks for coming 14:34:51 #endmeeting