15:03:28 <noonedeadpunk> #startmeeting openstack_ansible_meeting
15:03:28 <opendevmeet> Meeting started Tue Feb 24 15:03:28 2026 UTC and is due to finish in 60 minutes.  The chair is noonedeadpunk. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:03:28 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:03:28 <opendevmeet> The meeting name has been set to 'openstack_ansible_meeting'
15:03:32 <noonedeadpunk> #topic rollcall
15:03:33 <noonedeadpunk> o/
15:03:43 <NeilHanlon> o/
15:03:51 <anfimovir> hi all
15:03:53 <damiandabrowski> hi!
15:04:02 <noonedeadpunk> long time no see NeilHanlon - how are you doing?
15:04:14 <noonedeadpunk> courtesy ping jrosser
15:04:19 <NeilHanlon> heya - i'm doing alright.. head above water
15:04:34 <NeilHanlon> how y'all been?
15:06:06 <noonedeadpunk> quite /o\ with some personal matters lately
15:06:23 <noonedeadpunk> but I guess summary would be the same - head above the water :)
15:06:31 <noonedeadpunk> #topic CI state
15:07:07 <noonedeadpunk> CI seems to become better lately, though one issue pop ups from time to time, regarding failure to symlink /var/lib/libvirt/save folder from nova
15:07:40 <noonedeadpunk> #link https://zuul.opendev.org/t/openstack/build/6d136e1bdad74a48923d0bc441aa783d/log/job-output.txt#16761-16772
15:08:03 <noonedeadpunk> also nice to see you anfimovir on the meeting :)
15:08:49 <anfimovir> =)
15:09:36 <noonedeadpunk> so the issue seems to be some kind of race condition, as we explicitly do mv of the folder in advance: https://opendev.org/openstack/openstack-ansible-os_nova/src/branch/master/tasks/drivers/kvm/nova_compute_kvm_install.yml#L39-L40
15:10:09 <noonedeadpunk> From my local testing, force does a job to avoid the failure, so I'd expect this patch to be solving current failures
15:10:17 <noonedeadpunk> #link https://review.opendev.org/c/openstack/openstack-ansible-os_nova/+/977797
15:11:09 <noonedeadpunk> #topic ansible core 2.19
15:11:28 <noonedeadpunk> the topic has quite significant progress
15:11:31 <noonedeadpunk> #link https://review.opendev.org/q/topic:%22osa/core-2.19%22
15:11:45 <noonedeadpunk> though I have not tested/patched all roles so far
15:12:11 <noonedeadpunk> I think we have agreed to merge this patch first
15:12:17 <noonedeadpunk> #link https://review.opendev.org/c/openstack/openstack-ansible/+/971756
15:12:27 <noonedeadpunk> and then go through all other roles with fixes
15:13:48 <noonedeadpunk> and that good test would be the PKI patches, as they are needed across all repos
15:14:32 <noonedeadpunk> #topic hashi vault driver for pki
15:15:05 <noonedeadpunk> #link https://review.opendev.org/q/topic:%22osa_hashi_vault%22
15:15:27 <noonedeadpunk> previously we agreed with looking at os_cinder patch first, and review it
15:15:49 <noonedeadpunk> in case it's passing the review, damiandabrowski was planning to push patches to all other projects
15:17:16 <damiandabrowski> yeah, exactly! so we start with https://review.opendev.org/c/openstack/openstack-ansible-os_cinder/+/949427
15:17:18 <damiandabrowski> when it's merged, I'll prepare similar patches for all other roles
15:17:36 <noonedeadpunk> just let's make them dependent on https://review.opendev.org/c/openstack/openstack-ansible/+/971756 ok?
15:20:02 <noonedeadpunk> or well.
15:20:15 <noonedeadpunk> let's get more healthy state of 971756 first maybe
15:20:25 <damiandabrowski> ack
15:20:39 <noonedeadpunk> but I kinda like the idea to ensure state of all roles with 2.19 already in
15:20:52 <noonedeadpunk> rather then put time of doing that separately
15:21:03 <noonedeadpunk> #topic office hours
15:21:30 <noonedeadpunk> I am trying to solve ordring issue in os_swift
15:21:44 <noonedeadpunk> as it fails badly in case of aio lxc even
15:21:57 <noonedeadpunk> and we test it only on metal, so issue is not visible
15:22:32 <noonedeadpunk> so I'm going also to switch testin from metal to lxc for it
15:24:00 <noonedeadpunk> and this kinda block work on os_freezer work right now
15:24:20 <noonedeadpunk> os_watcher role has not been started yet
15:24:36 <noonedeadpunk> and then I think the last thing for the cycle would be ubuntu 26.04 support
15:24:48 <noonedeadpunk> I'd expect it to be quite trivial, given Trixie works nicely
15:24:55 <anfimovir> I remove today API v1 for Manila.
15:25:18 <noonedeadpunk> #link https://review.opendev.org/c/openstack/openstack-ansible-os_manila/+/977803
15:26:07 <noonedeadpunk> anfimovir: I think we should leave non-versioned variables
15:26:19 <noonedeadpunk> and make them to be v2
15:26:48 <noonedeadpunk> also I think it's missing clean-up of the catalog for v1
15:27:19 <noonedeadpunk> so maybe my first statement about unversioned vars is not great in this aspect
15:28:48 <anfimovir> No problem, I update my MR today, just wait finish tests for current results for receive logs from manila side.
15:35:05 <opendevreview> Ivan Anfimov proposed openstack/openstack-ansible-os_manila master: Remove support for Manila v1 API  https://review.opendev.org/c/openstack/openstack-ansible-os_manila/+/977803
15:35:22 <noonedeadpunk> ok, cool
15:35:51 <noonedeadpunk> we're running short on time for implementing all of these, but I think it should be doable after all
15:36:19 <noonedeadpunk> anything else to discuss?
15:37:48 <mgariepy> is there a doc that explains how to patch like a dependency like install if from a fork or something similar ?
15:37:51 <opendevreview> Ivan Anfimov proposed openstack/openstack-ansible-os_manila master: Remove support for Manila v1 API  https://review.opendev.org/c/openstack/openstack-ansible-os_manila/+/977803
15:39:09 <anfimovir> Please look it - https://review.opendev.org/c/openstack/openstack-ansible/+/949497
15:39:21 <anfimovir> I think we can merge it in current cycle.
15:40:03 <noonedeadpunk> mgariepy: I'm not sure we have that....
15:41:05 <mgariepy> anyone have some notes for this then ? i can submit patch for our doc but i'm not 100% sure how to do it.
15:41:29 <noonedeadpunk> so, we have a deployment installed fully from a forked/local mirrors
15:41:51 <noonedeadpunk> but I guess it depends on what exactly you are looking for
15:42:23 <anfimovir> @mgariepy please submit it for look.
15:42:23 <mgariepy> i want to patch ossa-2026-1
15:42:24 <noonedeadpunk> and yes - 949497 looks like a good candidate to me personally
15:42:28 <noonedeadpunk> at least smth worth trying
15:43:27 <noonedeadpunk> mgariepy: so you need to change these 2 vars then: https://opendev.org/openstack/openstack-ansible/src/branch/master/inventory/group_vars/nova_all/source_git.yml#L39-L40
15:44:11 <mgariepy> nah not these. like if i want to pin keystonemiddleware to a specific sha.
15:44:28 <noonedeadpunk> ah
15:44:39 <noonedeadpunk> right
15:44:53 <noonedeadpunk> I think then you need to have a own upper-constraints repo
15:45:21 <noonedeadpunk> and define own fork as https://opendev.org/openstack/openstack-ansible/src/branch/master/inventory/group_vars/all/source_git.yml#L19-L20
15:46:04 <noonedeadpunk> as it's coming from here: https://opendev.org/openstack/requirements/src/branch/master/upper-constraints.txt#L341
15:46:29 <jrosser> o/ argh sorry i'm very late
15:46:41 <mgariepy> for tagged release it's easy.. but if there are no tags?
15:47:03 <mgariepy> hey jrosser good monring :)
15:47:06 <noonedeadpunk> hm...
15:47:14 <noonedeadpunk> can't you have an sha there?
15:47:33 <noonedeadpunk> I'm not sure tbh
15:47:59 <mgariepy> it's kinda not 100% clear for me also :)
15:48:39 <mgariepy> i'll dig around and submit doc patches when i find a good way to do it.
15:49:07 <noonedeadpunk> well, I mean... you ofc can add keystonemiddleware to all service requirements, and drop them from u-c, but that would be really /o\
15:49:35 <jrosser> well we can have custom upper_constraints in /etc/openstack_deploy/ but that only counts for released versions from pypi
15:51:19 <jrosser> i think that this is a case that we have also come across, needing to patch a thing thats a dependancy of other python packages which isnt directly a requirement listed in the roles, or controlled by our git sha variables
15:51:22 <mgariepy> this particular on desn't have a new rel for the verion.
15:51:28 <jrosser> it would be nice to have a way to do that universally
15:52:29 <mgariepy> i'll see what i can find.
15:52:37 <mgariepy> for direct project it's really easy and awesome but for dependency like that it's not that trivial.
15:53:09 <noonedeadpunk> mgariepy: I think, you can set `venv_default_pip_packages: ""git+https://opendev.org/openstack/keystonemiddleware@master#egg=keystonemiddleware"
15:53:30 <noonedeadpunk> but important to drop constraint from u-c
15:54:04 <jrosser> here is how we patched glance_store https://paste.opendev.org/show/b3QdCW6sXwEwbrTzaMoa/
15:55:05 <noonedeadpunk> venv_default_pip_packages can be defined per group as well, as not all roles has `_user_pip_packages`
15:55:07 <mgariepy> so i guess we can do something similar to that for anything.
15:55:12 <jrosser> which yeah is a bit different because it's in a specific service
15:55:29 <noonedeadpunk> but actually most of them do
15:55:40 <jrosser> so i think you can edit u-c with an expression like the one in my paste
15:55:44 <noonedeadpunk> venv_default_pip_packages should be universal though
15:55:55 <jrosser> then use venv_default_pip_packages to specify the version you actually want
15:56:02 <opendevreview> Merged openstack/openstack-ansible-ceph_client master: Update outdate information in bindep.txt  https://review.opendev.org/c/openstack/openstack-ansible-ceph_client/+/972883
15:57:45 <mgariepy> i'll do some tests and probably add a documentation on our doc.
15:58:17 <noonedeadpunk> ++ would be really nice to have that
15:58:54 <mgariepy> yep.
15:59:12 <noonedeadpunk> #endmeeting