15:00:20 <noonedeadpunk> #startmeeting openstack_ansible_meeting 15:00:20 <opendevmeet> Meeting started Tue Jul 1 15:00:20 2025 UTC and is due to finish in 60 minutes. The chair is noonedeadpunk. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:20 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:20 <opendevmeet> The meeting name has been set to 'openstack_ansible_meeting' 15:00:24 <noonedeadpunk> #topic rollcall 15:00:33 <noonedeadpunk> o/ 15:00:44 <anfimovir> hello ;) 15:01:35 <damiandabrowski> hi! 15:02:23 <DavidGomez> o/ 15:02:36 <NeilHanlon> O/ 15:02:51 <NeilHanlon> Mobile... be home in ten mins 15:03:29 <noonedeadpunk> sure, no rush :) 15:03:33 <noonedeadpunk> #topic office hours 15:04:11 <noonedeadpunk> at the moment potentially our gates on master are broken, due to desynced mirrors for ubuntu, raising conflict for librbd/qemu 15:04:51 <noonedeadpunk> though it seems, that upstream mirrors are fine, so it;s likely limited to infra ones. 15:05:14 <noonedeadpunk> I was going to wait regular 24h until next sync before raising that to infra team 15:07:12 <noonedeadpunk> Another thing, is that we _really_ need to do some releases 15:07:23 <noonedeadpunk> #link https://review.opendev.org/q/topic:%22bump_osa%22+status:open 15:08:05 <noonedeadpunk> as we haven't done bugfix release for couple of months now 15:08:17 <damiandabrowski> ack, I'll have a look 15:08:46 <noonedeadpunk> thanks, much appreciated 15:09:23 <noonedeadpunk> There is also a topic regarding our deploy guide translation possibility 15:09:53 <noonedeadpunk> As currently, only docs/ folder is parsed and pushed to zanata 15:10:08 <noonedeadpunk> And there are 2 approaches to add deploy-guide to it 15:11:14 <noonedeadpunk> first would be to adjust scripting and CI jobs for translations to also look for deploy-guide and parse them. Though I don't really know, if/how that is gonna work, as things are spread quite thin - some managed by l18n, some in project-config scripts. 15:11:34 <noonedeadpunk> But then we'd fix it for everyone, despite the fact they want it or not 15:11:52 <noonedeadpunk> Another approach would be to move our deploy-guide under docs/ folder 15:11:54 <noonedeadpunk> #link https://review.opendev.org/c/openstack/openstack-ansible/+/949497 15:12:30 <noonedeadpunk> which looks like a straightforward thing to do... 15:12:43 <NeilHanlon> okay back now :) 15:12:53 <noonedeadpunk> do you have a history?:) 15:13:14 <NeilHanlon> yes but to tell it, beer pairings are involved 15:13:30 <NeilHanlon> we start with a 12% tequilla-barrel finished sour 15:13:31 <noonedeadpunk> hehe 15:14:31 <damiandabrowski> :D 15:14:35 <noonedeadpunk> so yeah, it would be good to hear wdyt about deploy guide what approach to select and if we wanna get it potentially translated or not 15:15:01 <anfimovir> @noonedeadpunk may be we just move deploy directory in docs... https://docs.openstack.org/2025.1/deploy/ - just blank... 15:17:08 <noonedeadpunk> Um, I'd say no? As it's really a different issue to deal with... 15:17:46 <noonedeadpunk> I wonder though why https://review.opendev.org/c/openstack/openstack-manuals/+/953312 did not help 15:18:27 <noonedeadpunk> as it's here obviously: https://ef4886d127b9d5e50b2a-46ed996c4c88287cea630d62dd5380de.ssl.cf1.rackcdn.com/openstack/37a9b4537fc9495c99974bf011d09f00/docs/2025.1/deploy/ 15:18:50 <damiandabrowski> i think it's nice to be present in https://docs.openstack.org/2025.1/deploy/ , I assume it can help a lot in getting new contributors 15:19:04 <anfimovir> It's just that the entire current structure is needed for the sake of this page, I think a translation would be more useful... I may be wrong, but for now it seems so ;) 15:19:06 <damiandabrowski> so the question is: can we move deploy guide to docs/ and still be present there? 15:19:18 <noonedeadpunk> so my guess would be that it's some kind of cache preventing new content from showing up 15:19:37 <noonedeadpunk> damiandabrowski: I think we should be, yes 15:19:53 <NeilHanlon> i think translating would be a good thing, especially to encourage contributors 15:20:02 <noonedeadpunk> as it's matter of producing an artifact and running promote job on it 15:20:16 <noonedeadpunk> NeilHanlon: anfimovir let's do both, huh? 15:20:31 <anfimovir> You right ;) 15:20:37 <NeilHanlon> yes :) 15:20:55 <noonedeadpunk> I will check with infra team what's wrong with the page as well. 15:21:15 <noonedeadpunk> but I would expect moving content to satisfy both usecases. 15:21:57 <noonedeadpunk> anfimovir: I also don't like to convert deploy guide to just docs, as this means that only kolla will stay on https://docs.openstack.org/2025.1/deploy/ which might send a very wrong signal 15:22:15 <noonedeadpunk> (once it's fixed) 15:22:42 <anfimovir> @noonedeadpunk, hmmm 15:22:50 <anfimovir> may be problem in..... 15:23:27 <anfimovir> https://review.opendev.org/c/openstack/openstack-manuals/+/946951/12/www/.htaccess - 160 line. 15:24:56 <noonedeadpunk> shouldn't be... https://opendev.org/openstack/openstack-manuals/src/branch/master/tools/www-generator.py#L81 15:25:02 <noonedeadpunk> only 2025.2 is in development 15:25:48 <noonedeadpunk> and also https://docs.openstack.org/2025.2/deploy/ 15:26:15 <anfimovir> Yes, problem only with current release 15:26:22 <anfimovir> 2024.2 fine too. 15:26:37 <noonedeadpunk> I'd bet on some cache tbh, as patch was merged less then 24h ago 15:26:58 <noonedeadpunk> or well.. 15:27:03 <noonedeadpunk> it's more? 15:27:15 <noonedeadpunk> anyway, will check with folks 15:27:36 <anfimovir> Just if you test - https://docs.openstack.org/latest/deploy/ - not working too.. 15:29:13 <noonedeadpunk> another thing - I put some love into our aio bootstrap role. It resulted in quite huge series of patches 15:29:17 <noonedeadpunk> #link https://review.opendev.org/c/openstack/openstack-ansible/+/953577 15:29:36 <noonedeadpunk> first, my intention with networks, was to offload all systemd stuff to openstack_hosts role 15:30:21 <jrosser> o/ hello sorry i'm late 15:30:45 <noonedeadpunk> but when I came to refactoring loopback devices, I realized I can't do that, as for instance with zfs or nfs or smth else - we need to have stuff prepared and mounted before lxc-containers-create execute 15:31:03 <noonedeadpunk> and we don't have any hook in between of openstack-hosts and lxc-container create 15:31:36 <noonedeadpunk> but still as a result I saw almost 2 times improvement in bootstrap-aio.sh runtime speed 15:32:12 <noonedeadpunk> and also I hope I fixed intermittent failures for LVM/Ceph scenario after AIO VM reboot 15:32:17 <jrosser> when we made the hooks it was always open to add more, it was just not obvious where thry would be needed 15:33:04 <noonedeadpunk> so the hook which would be useful to offload everything would be placed somewhere here: https://opendev.org/openstack/openstack-ansible-plugins/src/branch/master/playbooks/setup_hosts.yml#L30 15:33:17 <noonedeadpunk> but I'm not sure it makes sense outside of AIO... 15:33:32 <noonedeadpunk> Or maybe it does, and AIO is a justification that it's needed there 15:34:34 <noonedeadpunk> but also I already used templated networking today, as had to adjust some parameters for a sandbox 15:34:59 <noonedeadpunk> and allowed me to add second node more easily to aio deployment 15:35:35 <noonedeadpunk> so templating out systemd vars and doing them as part of openstack_hosts appears to be handy :D 15:36:37 <jrosser> i had some small comments on the network patch i think 15:36:38 <noonedeadpunk> About proceeding with refactoring - I guess I wanna see initial feedback/reviews, as chain is already going out of control 15:36:49 <noonedeadpunk> I have not seen them yet 15:36:58 <jrosser> imho starting with CIDRs would perhaps make quite some simplification 15:37:45 <jrosser> also ansible.utils.nthhost is useful 15:38:05 <noonedeadpunk> I never used ansible.utils.nthhost - would need to check what is that :D 15:38:36 <jrosser> "192.168.0.0/24" | nthhost(100) -> "192.168.0.100 15:38:40 <jrosser> " 15:39:16 <noonedeadpunk> ah, right 15:39:18 <jrosser> ^ so both things there together, start with the CIDR and end up with (example) the vip address 15:39:32 <jrosser> in one step 15:40:06 <noonedeadpunk> aha, so `| ansible.utils.ipaddr('100') | ansible.utils.ipaddr('address')` -> `| nthhost(100)` 15:40:08 <noonedeadpunk> makes sense 15:40:34 <jrosser> yes exactly, and it is happy with either cidr or not notation as input 15:41:32 <noonedeadpunk> ok, yes, that would simplify a lot of things in there 15:42:37 <noonedeadpunk> regarding if we even need `_mgmt_network` to be not network address right away - is also a good question 15:43:31 <noonedeadpunk> potentially not... 15:44:14 <noonedeadpunk> But I guess I used these new _mgmt_network too much 15:44:50 <jrosser> i think it was mostly ok, just could be more compact 15:45:04 <jrosser> and not to have the `netmask` var at all 15:45:23 <noonedeadpunk> so the netmask var was already used in couple of places 15:45:54 <noonedeadpunk> like here https://review.opendev.org/c/openstack/openstack-ansible/+/953675/7/tests/roles/bootstrap-host/templates/user_variables_ceph.yml.j2 15:46:42 <noonedeadpunk> so I decided to keep var format, but probably you're right and I could just change that as well 15:47:24 <noonedeadpunk> ok, will check on that 15:48:36 <noonedeadpunk> anything else? 15:48:43 <noonedeadpunk> other topics?:) 15:50:00 <damiandabrowski> it would be nice to get more reviews in https://review.opendev.org/q/topic:%22osa_hashi_vault%22 :D 15:50:02 <noonedeadpunk> oh, right 15:50:14 <noonedeadpunk> I started checking it, though I did not finish for sure 15:50:47 <noonedeadpunk> one thing I wanted to try out, is how to be sparter while defining _install_certificates 15:51:00 <anfimovir> I think no for today, we going back to deploy to Rocky 10 ;))) Thanks guys 15:51:31 <noonedeadpunk> and if there's a way to reduce amount of backend-specific stuff in there 15:51:36 <jrosser> i think we need to be very careful about the choice between vars and defaults 15:51:36 <noonedeadpunk> oh, btw 15:51:53 <jrosser> if there has been problems with not being able to override things via group vars, then the design is wrong 15:52:45 <noonedeadpunk> I just have quite limited experience with vault 15:53:11 <noonedeadpunk> so not always obvious for me what one may/may not want to override 15:53:41 <noonedeadpunk> btw about rocky 10 - support was added to dib and dib was tagged with a new release 15:54:00 <noonedeadpunk> so I'd assume that adding image to nodepool should be possible now 15:54:16 <noonedeadpunk> and I also confirmed that rocky 10 element is working 15:56:26 <noonedeadpunk> so hopefully we'll get some CI around it soon 15:56:47 <noonedeadpunk> but if that's it, then 15:56:51 <noonedeadpunk> #endmeeting