15:02:36 <noonedeadpunk> #startmeeting openstack_ansible_meeting
15:02:36 <opendevmeet> Meeting started Tue Jul  9 15:02:36 2024 UTC and is due to finish in 60 minutes.  The chair is noonedeadpunk. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:02:36 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:02:36 <opendevmeet> The meeting name has been set to 'openstack_ansible_meeting'
15:02:40 <noonedeadpunk> #topic rollcall
15:02:45 <noonedeadpunk> sorry for being late
15:02:47 <noonedeadpunk> o/
15:08:37 <noonedeadpunk> #topic office hours
15:08:51 <noonedeadpunk> so, I didn't have much progress during this week and was mainly absent previous one
15:09:37 <noonedeadpunk> one of relatively importanti things - I've decided to write down policies for when we can land a change - please kindly review https://review.opendev.org/c/openstack/openstack-ansible/+/923725 and comment out there
15:09:39 <NeilHanlon> o/ i'm here
15:10:00 <noonedeadpunk> as that's kinda project governance thingy
15:10:16 <NeilHanlon> ++  will review
15:10:39 <noonedeadpunk> next to that - unmaintained/yoga seems healthy. I clean forgot to update release patch to transition Zed to unmaintained as well
15:10:53 <jrosser> o/ sorry just in another meeting, here soon
15:11:03 <noonedeadpunk> sure
15:11:26 <noonedeadpunk> but now we'd need to land https://review.opendev.org/c/openstack/openstack-ansible/+/923619 to fix 2023.1 upgrade jobs
15:11:55 <noonedeadpunk> we also should have issued new minor releases last week with regards to CVE-2024-32498
15:12:14 <noonedeadpunk> though, I was waiting for os_ironic bugfix to land, though it keeps timeouting on rocky upgrade job
15:12:29 <NeilHanlon> 🙉
15:12:41 <noonedeadpunk> I'm really about to set this job to NV specifically for Ironic...
15:13:07 <noonedeadpunk> Have zero idea why it's that slow....
15:13:15 <jrosser> it is curious indeed
15:14:06 <noonedeadpunk> to be also fair - upgrade from 2023.1 takes way longer then from 2023.2
15:14:24 <noonedeadpunk> and like - 30m longer
15:15:21 <jrosser> i hope it actually does the upgrade and we did not break that somehow
15:15:41 <noonedeadpunk> I'm just checking last 4 rechecks for ironic, and for ubuntu upgrade from 2023.2 is 2h 11m; 2h 09m; 2h 15m
15:16:25 <noonedeadpunk> while from 2023.1 jammy is 2h 49m; 2h 35m; 2h 56m
15:16:31 <noonedeadpunk> hehe
15:17:05 <jrosser> like if it upgrades to the same version that would be quicker /o\
15:17:47 <noonedeadpunk> I think it does: https://zuul.opendev.org/t/openstack/build/c0680e7fcd124a8ea02ed9a1bfe29ce9/log/job-output.txt#6478-6490
15:18:51 <noonedeadpunk> and then https://zuul.opendev.org/t/openstack/build/c0680e7fcd124a8ea02ed9a1bfe29ce9/log/job-output.txt#20066-20285
15:18:59 <noonedeadpunk> so we optimized something :D
15:19:11 <NeilHanlon> :D
15:19:33 <noonedeadpunk> I can recall you jrosser was puting some effort into that actually
15:19:36 <NeilHanlon> oh hmmm
15:19:41 <NeilHanlon> fatal: [localhost]: FAILED! => {"changed": false, "failures": [], "msg": "Depsolve Error occurred: \n Problem: problem with installed package curl-7.76.1-29.el9_4.x86_64\n  - package curl-minimal-7.76.1-29.el9_4.x86_64 from baseos conflicts with curl provided by curl-7.76.1-29.el9_4.x86_64 from @System\n  - package
15:19:41 <NeilHanlon> curl-minimal-7.76.1-29.el9_4.x86_64 from baseos conflicts with curl provided by curl-7.76.1-29.el9_4.x86_64 from baseos\n  - conflicting requests", "rc": 1, "results": []}
15:20:00 <NeilHanlon> did we miss a backport?
15:20:12 <noonedeadpunk> I _think_ it just landed to yoga
15:20:31 <noonedeadpunk> if you're talking about https://review.opendev.org/c/openstack/openstack-ansible/+/923619
15:20:36 <NeilHanlon> yeah
15:20:36 <noonedeadpunk> it's re-checking right now
15:20:39 <NeilHanlon> ahh okay
15:20:47 <NeilHanlon> sorry i misunderstood--thought that was the timeout one
15:21:25 <noonedeadpunk> doh, it failed :(
15:21:37 <noonedeadpunk> nah, timeout one is https://review.opendev.org/c/openstack/openstack-ansible-os_ironic/+/923594
15:23:08 <jrosser> that is timeout during log upload?
15:23:16 <jrosser> i feel like we have been there before
15:23:26 <noonedeadpunk> nah
15:23:40 <noonedeadpunk> timeout is actually somewhere between tempest and rally
15:23:44 <noonedeadpunk> so it's just being slow
15:24:13 <noonedeadpunk> ok, it succeeded just now in gates
15:24:26 <NeilHanlon> of course
15:24:31 <noonedeadpunk> so hopefully it will land now....
15:24:34 <NeilHanlon> it's because it was being observed
15:24:45 <noonedeadpunk> sure, and it just got scared of you :D
15:25:04 <noonedeadpunk> it always works like that - once you look at a thing it get scared and behaves
15:25:25 <NeilHanlon> I believe in computer/technology "Mana"
15:25:33 <NeilHanlon> if you don't have enough mana with something, it won't respect you
15:25:47 <NeilHanlon> until someone with more mana comes aroud
15:26:15 <noonedeadpunk> haha, yeah, probably :D
15:26:41 <noonedeadpunk> so, we still have ansible-core 2.17 and mariadb 11.4 thingy
15:26:51 <noonedeadpunk> will try to check on mariadb this week
15:27:00 <jrosser> ansible-core just needs a release note on ceph-ansible
15:27:06 <jrosser> and a fix for the upgrade issues
15:27:33 <noonedeadpunk> yeah
15:27:43 <noonedeadpunk> which maybe worth doing as a follow-up?
15:27:55 <noonedeadpunk> ok, I guess it's time for another etherpad
15:28:10 <jrosser> that would be helpful yes
15:29:58 <noonedeadpunk> #link https://etherpad.opendev.org/p/osa-epoxy-ptg
15:30:08 <noonedeadpunk> I propose to start building up ptg one right away
15:31:16 <NeilHanlon> the name i vote for is never chosen... :P
15:35:26 <noonedeadpunk> frankly - all names were terrible this time
15:35:27 <noonedeadpunk> imo
15:37:22 <noonedeadpunk> we have actually 2 more things regarding new ansible-core
15:37:44 <noonedeadpunk> first - we need to drop `ANSIBLE_COLLECTIONS_PATHS` - this raises the warning in 2.17
15:37:56 <noonedeadpunk> https://paste.openstack.org/show/bomMOIV12Nki1ffQNstj/
15:38:13 <noonedeadpunk> and the second - kinda time to think about what to do regarding DEFAULT_GATHER_SUBSET
15:38:32 <noonedeadpunk> as that is /o\
15:39:40 <opendevreview> Jonathan Rosser proposed openstack/openstack-ansible-openstack_hosts master: Manage apt repositores and keys using deb822_repository module  https://review.opendev.org/c/openstack/openstack-ansible-openstack_hosts/+/907434
15:40:50 <jrosser> i have a patch for the collections path
15:41:12 <jrosser> https://review.opendev.org/c/openstack/openstack-ansible/+/921928
15:42:33 <opendevreview> Dmitriy Rabotyagov proposed openstack/openstack-ansible-rabbitmq_server master: Manage apt repositores and keys using deb822_repository module  https://review.opendev.org/c/openstack/openstack-ansible-rabbitmq_server/+/907833
15:42:47 <noonedeadpunk> oh nice
15:42:59 <noonedeadpunk> sorry, I haven't reviewed all things from previous week yet :(
15:43:45 <noonedeadpunk> huh, I wonder if https://review.opendev.org/c/openstack/openstack-ansible-plugins/+/923403 might help us with facts as well
15:44:06 <jrosser> yeah i was thinking about something like that too
15:44:36 <jrosser> though the thing is DEFAULT_GATHER_SUBSET also affects all 3rd party codde we use, which is a super nice side effect
15:44:45 <noonedeadpunk> yeah
15:45:13 <noonedeadpunk> I wonder if it's worth rasing within Ansible community
15:45:19 <noonedeadpunk> and what's the thinking behind that
15:45:23 <jrosser> but perhaps module_defaults on all playbooks will do the same?
15:45:29 <noonedeadpunk> maybe they just don't recognize the usecase...
15:45:51 <jrosser> would need to check if module_defaults will propagate down into a called role
15:46:04 <noonedeadpunk> that is _very_ annoying....
15:49:01 <opendevreview> Dmitriy Rabotyagov proposed openstack/openstack-ansible unmaintained/yoga: Bump role SHAs for unamitained/yoga  https://review.opendev.org/c/openstack/openstack-ansible/+/923768
15:49:19 <noonedeadpunk> this is another thing that's probably needed for Zed upgrade job...
15:49:32 <noonedeadpunk> as it needs more modern openstack_hosts role
15:50:25 <jrosser> i'm not sure that the work i did on auto-handling stable/<> vs unmaintained/<> landed in all the branches either
15:50:45 <jrosser> fixing the unmaintained branches is on my todo list but just no time
15:50:53 <noonedeadpunk> well, it pulls the right thing on Zed
15:51:03 <noonedeadpunk> Yoga does not have that patch though
15:51:10 <noonedeadpunk> but upgrades are NV there anyway
15:51:17 <noonedeadpunk> actually about that
15:51:29 <noonedeadpunk> maybe we should just eol everything below Yoga?
15:51:44 <noonedeadpunk> or Xena
15:52:04 <noonedeadpunk> (I'd keep Xena maybe...)
15:52:11 <jrosser> we do still get queries on the ML about upgrades from old stuff
15:52:20 <jrosser> so yes we should EOL *really* old stuff
15:52:54 <noonedeadpunk> so we have Victoria, Wallaby, Xena and Yoga
15:53:27 <noonedeadpunk> like neither of them feel *really* old....
15:53:44 <noonedeadpunk> (while they kinda are)
15:54:14 <opendevreview> Dmitriy Rabotyagov proposed openstack/openstack-ansible unmaintained/yoga: Bump role SHAs for unamitained/yoga  https://review.opendev.org/c/openstack/openstack-ansible/+/923768
15:54:57 <noonedeadpunk> and I highly doubt we'll backport anything below Yoga...
15:55:08 <noonedeadpunk> Though I know couple of folks are still running Xena...
16:03:10 <opendevreview> Jonathan Rosser proposed openstack/openstack-ansible-ops master: Add support for deploying mcapi control plane k8s on debian-12  https://review.opendev.org/c/openstack/openstack-ansible-ops/+/923586
16:04:38 <jrosser> NeilHanlon: i have the starting point of making magnum cluster api work on rocky here https://review.opendev.org/c/openstack/openstack-ansible-ops/+/923447/8
16:04:56 <NeilHanlon> ooh, nice!
16:05:01 <jrosser> though if it does not "just work" i'm not sure i have time to chase it
16:05:21 <jrosser> so just fyi if you're interested in magnum stuff
16:05:54 <jrosser> it will be kerrnel meets lxc meets bpf meets cilium == brok, most likely
16:06:22 <NeilHanlon> alright cool, appreciate the heads up
16:11:02 <noonedeadpunk> #endmeeting