15:00:19 <noonedeadpunk> #startmeeting openstack_ansible_meeting
15:00:19 <opendevmeet> Meeting started Tue Jul 16 15:00:19 2024 UTC and is due to finish in 60 minutes.  The chair is noonedeadpunk. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:19 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:19 <opendevmeet> The meeting name has been set to 'openstack_ansible_meeting'
15:00:26 <noonedeadpunk> #topic rollcall
15:00:28 <noonedeadpunk> o/
15:01:59 <jrosser> o/ hello
15:02:10 <damiandabrowski> hi!
15:07:39 <opendevreview> Dmitriy Rabotyagov proposed openstack/openstack-ansible-repo_server master: Replace Nginx with Apache2  https://review.opendev.org/c/openstack/openstack-ansible-repo_server/+/924162
15:07:41 <noonedeadpunk> #topic office hours
15:07:56 <noonedeadpunk> so there're quite some things going on I believe
15:08:21 <noonedeadpunk> a lot of work was put into moving playbooks into plugins repo
15:08:28 <noonedeadpunk> and seems that POC is working
15:08:49 <noonedeadpunk> #link https://review.opendev.org/c/openstack/openstack-ansible-plugins/+/924156
15:09:28 <noonedeadpunk> then, we should be able to move playbooks into collection... Hopefully, that's a good move :D
15:09:51 <noonedeadpunk> another thing, I worked on yesterday, was replacing nginx for repo server with apache
15:10:01 <noonedeadpunk> and there was a good idea to have an apache role
15:10:31 <jrosser> yes so i started with the infrastructure playbooks, that is looking good and there are patches
15:10:43 <jrosser> and it is testable with no extra jobs needed
15:11:03 <jrosser> tbh this is the thing i need for the magnum work
15:11:33 <noonedeadpunk> I'd say it's reasonable overall
15:11:37 <jrosser> i was going to look at the stuff in setup-hosts next as that is all testable too
15:11:42 <noonedeadpunk> gives us a chance to rename playbooks as well
15:11:55 <jrosser> then ceph playbooks which will need some new jobs
15:12:01 <jrosser> and finally the services
15:12:17 <noonedeadpunk> and then remove necessity to know the path to playbooks and need to cd to it before execution
15:12:24 <jrosser> ^ this last one is the most hard to wire up zuul jobs to without tons of extra jobs
15:14:43 <jrosser> kind of related for magnum i had some pretty good success on debian-12 and rocky-9
15:14:55 <noonedeadpunk> I wonder if there's a workaround for that
15:15:23 <jrosser> there could be, it's a bit like automatically determining the scenario
15:15:36 <noonedeadpunk> meaning - as we're gonna name playbook alike to scenario that will trigger it - maybe we can pass filename as part of scenario somehow
15:15:45 <jrosser> if its playbooks/install-<service>.yml then we want to add <service> to the scenario
15:16:03 <jrosser> iirc there is a changed files list on the zuul job?
15:16:30 <noonedeadpunk> yeah, but my idea was to rename in plugins to playbooks/<service>.yml
15:16:59 <noonedeadpunk> then file | basename | split('.') | first
15:17:04 <jrosser> well we can always check the name against a list of services
15:17:19 <jrosser> as there are many things in playbooks/ that are not services
15:17:25 <noonedeadpunk> as imo, having install-<service> is very confusing today
15:17:49 <noonedeadpunk> as it's way more then just install
15:18:10 <noonedeadpunk> but dunno - we can discuss naming for sure
15:19:02 <jrosser> so we might want to rename the playbook in https://review.opendev.org/c/openstack/openstack-ansible-plugins/+/924156
15:19:23 <noonedeadpunk> yeah, I'd just name it as `haproxy` frankly speaking
15:19:32 <jrosser> ^ this is the needed one for magnum which is why i did it first
15:19:41 <jrosser> thats fair
15:20:12 <noonedeadpunk> and then here https://opendev.org/openstack/openstack-ansible/src/branch/master/playbooks/haproxy-install.yml I'd just replace content with `import_playbook: openstack.osa.haproxy`
15:20:32 <noonedeadpunk> for keeping some compatability and give ppl to aclimatize
15:20:38 <noonedeadpunk> or how to say that...
15:21:17 <jrosser> i am already there https://review.opendev.org/c/openstack/openstack-ansible/+/924168
15:21:44 <jrosser> btw there are a bunch of patches to migrate the old playbooks to the common_tasks things now in the plugins repo
15:22:11 <jrosser> but i am getting ahead of those changes now with the patches putting the playbooks themselves into the plugins repo
15:22:23 <noonedeadpunk> aha, yeah, ok
15:23:10 <noonedeadpunk> I think most of them are merged actually?
15:23:42 <jrosser> seems  like we have a degree of risk here too particularly with the differences we foung with include/import_role vs include_tasks
15:23:51 <noonedeadpunk> oh yes
15:24:03 <jrosser> so i would like to get this all done sooner rather than later to find those things
15:24:31 <noonedeadpunk> ++
15:24:33 <noonedeadpunk> yes, totally
15:25:04 <jrosser> will 24.04 support be experimental for this release?
15:25:28 <noonedeadpunk> I think we're expected to have full support...
15:25:40 <noonedeadpunk> I'm not sure about repos, like for mariadb/rabbitmq though
15:25:59 <noonedeadpunk> what I'm inlined to leave experimental - is CentOS 10 Stream
15:26:14 <noonedeadpunk> but this all is another huge chunk of work...
15:26:32 <jrosser> given that debian-12 is working i am not expecting terrible surprised with 24.04
15:30:52 <opendevreview> Merged openstack/openstack-ansible-lxc_hosts master: Combine vars files for all debian derivatives  https://review.opendev.org/c/openstack/openstack-ansible-lxc_hosts/+/923166
15:31:15 <jrosser> the other thing was unmaintained branches
15:31:43 <jrosser> and specifically centos-8-stream, we're still stuck there on some of them - i think we approved two for frickler to look at merging?
15:33:40 <noonedeadpunk> yeah
15:33:45 <noonedeadpunk> and master should be fixed I think
15:33:56 <noonedeadpunk> I did some round there but didn\t have time since
15:35:41 <noonedeadpunk> and frankly - I'd really consider eol-ing branches at least before Xena
15:35:41 <jrosser> oh hmm sooo many job retries
15:35:53 <jrosser> this must be costing a ton of CI time
15:36:03 <noonedeadpunk> yeah, job retries is another issue, though likely not our fault...
15:36:14 <noonedeadpunk> but it's bad...
15:39:54 <noonedeadpunk> I think I will be able to find some next week for 24.04
15:42:43 <jrosser> looks like there are noble pacakges for rabbitmq at cloudsmith
15:43:52 <jrosser> and we look OK here too https://mirror.mariadb.org/repo/11.4/ubuntu/dists/
15:46:31 <noonedeadpunk> and ceph 19.2 comes from default repos
15:46:58 <noonedeadpunk> so this should be fine as well, just not use community repos as we did at first for jammy
15:47:22 <noonedeadpunk> https://packages.ubuntu.com/search?suite=noble&keywords=ceph-common
15:48:38 <jrosser> scarily new version of ceph :)
15:49:08 <noonedeadpunk> whaata
15:49:18 <noonedeadpunk> lol
15:49:34 <noonedeadpunk> I somehow thought it's reef
15:49:45 <noonedeadpunk> yeah, 19 is too fresh
15:50:29 <jrosser> of course ceph-ansible may be an issue there again
15:50:43 <noonedeadpunk> reef doesn't contain noble...
15:50:54 <noonedeadpunk> so yeah... it can be indeed end-up as experimental
15:51:10 <jrosser> we don't have to run the ceph job at all on noble
15:51:16 <jrosser> that can stay on jammy for a cycle
15:51:21 <noonedeadpunk> yes, totally
15:51:38 <noonedeadpunk> though a point for experimental state
15:51:40 <noonedeadpunk> at least for now
15:52:28 <noonedeadpunk> releasing of new ceph tags gets quite slow
15:53:23 <noonedeadpunk> it's almost year since reef release, and kinda just 2 "stable" releases of it
15:58:31 <frickler> jrosser: are there still things open for me to look at? I must have missed those, do you have links?
15:59:40 <jrosser> frickler: there are two here which noonedeadpunk approved https://review.opendev.org/q/topic:%22osa/remove-8-stream%22+status:open
16:00:34 <noonedeadpunk> #endmeeting