15:00:28 <noonedeadpunk> #startmeeting openstack_ansible_meeting
15:00:28 <opendevmeet> Meeting started Tue Jun  4 15:00:28 2024 UTC and is due to finish in 60 minutes.  The chair is noonedeadpunk. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:28 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:28 <opendevmeet> The meeting name has been set to 'openstack_ansible_meeting'
15:00:32 <noonedeadpunk> #topic rollcall
15:00:34 <noonedeadpunk> o/
15:03:50 <hamburgler> o/ - observing :)
15:07:37 <hamburgler> Is the floor open for questions if things are quiet at the moment?
15:10:37 <noonedeadpunk> yes, suire
15:10:48 <noonedeadpunk> fire it up
15:11:05 * noonedeadpunk got severely distracted from leading the meeting anyway
15:11:32 <noonedeadpunk> I guess the main thing for it was releasing of Caracal that we're doing right now
15:12:00 <noonedeadpunk> which should be cut from SHA produced on merge of https://review.opendev.org/c/openstack/openstack-ansible/+/921271/2
15:12:17 <hamburgler> Cool, haven't dug through the code yet - so apologies if it is an easy answer. I know there is new role integration to manage resources within OSA using Ansible modules - is this optional via flag? Currently we have our own that we would likely keep.
15:12:35 <NeilHanlon> o/ am here sorry for being late
15:12:44 <hamburgler> howdy :)
15:13:09 <NeilHanlon> o/
15:13:09 <noonedeadpunk> hamburgler: so, it is/will be used inside our roles, like octavia, magnum, trove, tempest, etc.
15:13:23 <noonedeadpunk> And there is a separate playbooks that can be used standalone
15:13:33 <noonedeadpunk> So if you don't define variables - it will not do anything
15:13:43 <hamburgler> excellent, great :)
15:13:49 <noonedeadpunk> https://opendev.org/openstack/openstack-ansible/src/branch/master/playbooks/openstack-resources.yml
15:14:11 <noonedeadpunk> if you want to - you can migrate to it. if not - just ignore.
15:14:18 <noonedeadpunk> it's not enforced or anything like that
15:14:19 <hamburgler> wonderful, ty
15:14:46 <noonedeadpunk> #topic office hours
15:14:59 <noonedeadpunk> I actually don't have any agenda except doing release this week
15:15:11 <noonedeadpunk> and by know - things looks fine
15:15:16 <noonedeadpunk> *by now
15:15:40 <noonedeadpunk> the only role we know is kinda broken  - is adjutant. Though it's kinda adjutant fault
15:15:49 <noonedeadpunk> also Masakari upgrade known to be broken
15:16:12 <hamburgler> Have not been able to stay up to date as I'd like - but is magnum w/ vexxhost capi driver looking to be rolled out in Caracal?
15:16:33 <noonedeadpunk> hamburgler: yes, and jrosser also did a CI testing for this one
15:17:04 <noonedeadpunk> hamburgler: though doc for it is in slightly unusual place: https://docs.openstack.org/openstack-ansible-ops/latest/mcapi.html
15:17:14 <jrosser> oh meeting - doh
15:17:18 <jrosser> o/ hello
15:17:21 <NeilHanlon> heya jrosser
15:17:59 <hamburgler> noonedeadpunk: ty - yes i wouldn't have looked in the ops repo
15:18:03 <opendevreview> Dmitriy Rabotyagov proposed openstack/openstack-ansible-openstack_hosts stable/zed: Allow openstack_hosts role to configure git safe.directory entries  https://review.opendev.org/c/openstack/openstack-ansible-openstack_hosts/+/921037
15:18:13 <jrosser> hamburgler: you should be able to make an AIO with capi driver pretty easily
15:18:28 <noonedeadpunk> folks, seems I forgot that we do upgrade jobs also in gates
15:18:37 <noonedeadpunk> so ^ should be re-voted now
15:18:40 <jrosser> (btw it would be great for someone != me to have a go at that in case the docs are missing something obvious)
15:18:56 <noonedeadpunk> and this kinda last part left to transition Zed to unmaintained which long overdue now
15:19:03 <hamburgler> jrosser: is it limited to AIO atm? or can be used in HA config?
15:19:18 <jrosser> HA is possible too
15:19:24 <noonedeadpunk> jrosser: I am expected to go and have some hands-on with CAPI stuff in next couple of weeks
15:19:34 <hamburgler> jrosser: will likely start looking at Caracal upgrade in lab and will follow the docs and see in the coming weeks too
15:19:44 <jrosser> this would be great
15:20:04 <jrosser> i am currently making some improvements to the upstream k8s image-builder tool for baking your own capi images
15:22:23 <noonedeadpunk> but frankly - I don't have much for the meeting today.
15:22:37 <noonedeadpunk> still catching up after quite crazy May...
15:25:04 <NeilHanlon> yeah, same
15:25:19 <NeilHanlon> and i'm going to a conference this weekend so.. here's to a crazy June..
15:26:00 <noonedeadpunk> hehe
15:26:07 <noonedeadpunk> true
15:26:40 <NeilHanlon> hey at least we all do the same thing and lie to ourselves that next month will be better ;)
15:27:12 <hamburgler> Previously saw discussion around vpnaas w/ ovn, I think? Is this still on the roadmap?
15:27:24 <noonedeadpunk> It's done for Caracal as well
15:27:36 <noonedeadpunk> And it works for our sandbox at least with OVN :D
15:27:49 <noonedeadpunk> as well as ovn-bgp-agent
15:28:04 <noonedeadpunk> might need some polishing ofc, but overall should be fine
15:28:32 <noonedeadpunk> though project itself is quite raw up to my taste...
15:28:37 <hamburgler> :D - Well it looks like it's a party then. I'm super curious to play with the ovn-bgp-agent - I have the underlying BGP setup in my lab w/ frr and networking gear so hopefully I can make it play with the neutron/frr role vars etc.
15:29:02 <noonedeadpunk> yeah, so we have FRR role now as part of that
15:29:15 <noonedeadpunk> though I have no idea if you can just skip using it...
15:30:20 <noonedeadpunk> kinda - not at the moment: https://opendev.org/openstack/openstack-ansible-os_neutron/src/branch/master/tasks/neutron_install.yml#L55-L68
15:30:42 <hamburgler> Will have to take a look, not opposed to seeing if I can just use that instead of my internal one. Lab is a bit janky, l3 for 'external' connectivity and l2 internal
15:31:13 <noonedeadpunk> well, you kinda need frr on net/compute nodes anyway
15:31:15 <noonedeadpunk> so...
15:31:39 <noonedeadpunk> and frr config is pretty much for you to define through a var
15:32:04 <hamburgler> Yeah - currently running there. But will see lol :D. I had trouble integrating it with LXC last year.
15:33:53 <hamburgler> Does it run in northd container as well? or strictly hosts?
15:36:55 <noonedeadpunk> no, it does not in northd
15:37:07 <noonedeadpunk> basically, it ideally should run only on gatway hosts
15:37:16 <noonedeadpunk> and that's how tings are with SB DB driver.
15:37:56 <noonedeadpunk> BUT, NB DB has line of bugs, where it assumes that you always should have distributed routing, meaning - no gateway/net nodes, just computes
15:38:32 <noonedeadpunk> and also obviously haven't tried/looked into ovn exposure method
15:38:48 <noonedeadpunk> (which is non-default)
15:39:11 <noonedeadpunk> as there you'd need to build a separate OVN cluster on each compute in addition to central OVN cluster
15:39:21 <hamburgler> Okay, for some reason I thought it had to be able to read from DB in OVN container. Ah yes everything setup is DVR right w/ collapsed compute, no dedicated gateway nodes.
15:40:02 <noonedeadpunk> So, it's an ovn-bgp-agent that connects to OVN DB yes, but it does through networks
15:40:12 <noonedeadpunk> so it doesn't need to run on northd
15:40:21 <noonedeadpunk> and then it manages local to itself FRR
15:40:53 <hamburgler> ah great :)
15:43:11 <hamburgler> ty :) that's all from me with the question parade
15:48:03 <noonedeadpunk> I guess I will conclude this meeting as well
15:48:06 <noonedeadpunk> #endmeeting