13:00:16 <mnasiadka> #startmeeting kolla 13:00:16 <opendevmeet> Meeting started Wed Aug 21 13:00:16 2024 UTC and is due to finish in 60 minutes. The chair is mnasiadka. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:00:16 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:00:16 <opendevmeet> The meeting name has been set to 'kolla' 13:00:18 <mnasiadka> #topic rollcall 13:00:19 <mnasiadka> o/ 13:00:22 <SvenKieske> o/ 13:00:22 <mmalchuk> o/ 13:00:25 <mattcrees> o/ 13:00:28 <bbezak_alt> o/ 13:00:31 <IvanVnuko[m]> o/ 13:00:35 <mhiner> o/ 13:00:50 <frickler> \o 13:01:01 <r-krcek> o/ 13:02:03 <mnasiadka> #topic agenda 13:02:03 <mnasiadka> * Announcements 13:02:03 <mnasiadka> * CI status 13:02:03 <mnasiadka> * Current cycle planning 13:02:03 <mnasiadka> * Additional agenda (from whiteboard) 13:02:04 <mnasiadka> * Open discussion 13:02:07 <mnasiadka> #topic Announcements 13:02:08 <jovial> o/ 13:02:14 <darmach6> o/ 13:02:41 <mnasiadka> I've decided to fill in for the PTL role for another cycle, the patch has been merged already - https://review.opendev.org/c/openstack/election/+/926285 13:02:57 <SvenKieske> \o/ 13:03:12 <mnasiadka> #topic CI status 13:03:23 <mnasiadka> I haven't seen any obvious breakages, so I guess nothing to report here ;-) 13:03:28 <mnasiadka> #topic Current cycle planning 13:03:37 <jovial> Kayobe CI back to green as of this morning 13:03:51 <mnasiadka> ansible bump? 13:03:58 <jovial> Indeed 13:04:11 <mnasiadka> Ok, let's have a look on planned ''features'' this cycle 13:04:21 <mnasiadka> Ubuntu Noble - that's obvious, nearly getting there 13:04:43 <mnasiadka> OVN BGP agent - I haven't done anything - bbezak_alt - do you have some spare cycles? 13:05:02 <bbezak_alt> yeah, will revive that 13:05:13 <mnasiadka> great 13:05:19 <mnasiadka> Openinfra mirrors for some problematic packages - SvenKieske how's it going? 13:06:16 <SvenKieske> mhm nothing to report, as the downstream funding is running out I doubt I'll get servers/infra capacity from there. But I guess I can still bother opendev infra team if they can cough up some gigabytes? :) 13:06:26 <opendevreview> Merged openstack/kolla-ansible stable/2024.1: CI: drop RMQ reconfigure step in queue migrations https://review.opendev.org/c/openstack/kolla-ansible/+/926127 13:07:02 <mnasiadka> yeah, I think we wanted to discuss with openinfra - but would be good to have a list what we want to mirror and some capacity calculations 13:07:13 <mnasiadka> so if you could update that on whiteboard - would be nice 13:07:19 <mnasiadka> Multiple RMQ versions per cycle - mattcrees how's it going? 13:07:21 <SvenKieske> future funding of the downstream is a little unclear to me currently, so if we're lucky maybe there will still be capacity, but I guess I can only wait for end of year to report back 13:07:47 <SvenKieske> okay, I can certainly compile a list of currently external repositories :) 13:08:37 <mattcrees> We've got a topic set up now: https://review.opendev.org/q/topic:%22slurp-rabbitmq%22, been getting helpful reviews so far :). I'm making sure to respond asap so we can get these merged soon 13:09:18 <mnasiadka> Ok, I'll have a look into those later 13:09:27 <mattcrees> Cheers 13:09:28 <mnasiadka> if other cores can do the same - it would be nice 13:09:58 <mnasiadka> ok, I added that topic to whiteboard 13:10:23 <mnasiadka> I don't have anything more in the whitebaord, so let's move on 13:10:30 <mnasiadka> #topic Additional agenda (from whiteboard) 13:10:38 <mnasiadka> r-krcek [21th August] my wishlist :) 13:10:38 <mnasiadka> please review https://review.opendev.org/c/openstack/kolla-ansible/+/925714 and https://review.opendev.org/c/openstack/kolla/+/925712 13:10:38 <mnasiadka> This got us stumped with Sven, anyone with bandit knowledge? https://review.opendev.org/c/openstack/kolla-ansible/+/923110 13:10:38 <mnasiadka> Any feedback would be greatly appreciated https://review.opendev.org/c/openstack/kolla-ansible/+/924651 13:11:38 <SvenKieske> ah, thx for the reminder, need to check the bandit stuff.. but question is, if this function really needs to call arbitrary commands, afaik it's only used exactly once to call a single command iirc? 13:12:38 <kevko> bash to python ? it will be fun :D 13:12:42 <SvenKieske> ok I'm wrong, it's called twice.. maybe just use subprocess directly without a wrapper function? I mean it's two calls 13:13:17 <SvenKieske> not really worth the wrapper imho, would get rid of the bandit warning I guess 13:13:24 <mnasiadka> ok, the first two look heavy, second one also is not a light one - I added my review on the last one 13:13:33 <mnasiadka> bbezak_alt: willing to have a look in the last one as well? 13:14:13 <mnasiadka> SvenKieske: let's discuss in the patch, shall we? 13:14:22 <SvenKieske> fine :) 13:14:22 <bbezak_alt> 924651 that one? 13:14:32 <r-krcek> Thank you guys for the feedback :) 13:14:35 <bbezak_alt> will check 13:14:37 <mnasiadka> bbezak_alt: yup 13:14:40 <mnasiadka> ok, next one 13:14:42 <mnasiadka> ivnucko [21th August] 13:14:42 <mnasiadka> adding TLS for other than internal/external and backend (proxy to service API) 13:14:42 <mnasiadka> like galera replication - https://review.opendev.org/c/openstack/kolla-ansible/+/925317 or 13:14:42 <mnasiadka> rabbitmq internode - https://review.opendev.org/c/openstack/kolla-ansible/+/921381 13:14:42 <mnasiadka> ...handle them as part of backend - enable them with kolla_enable_tls_backend and use the same cert from certificate role? 13:14:44 <mnasiadka> ...or a separate type, create a new play in certificates role and enable to have distinct certificate for them? 13:14:44 <mnasiadka> is there a use case for separate certificates for MariaDB to ProxySQL and for inter-node replication, is it worth the complication (cannot use service-cert-copy)? 13:15:43 <mnasiadka> I don't think it's worth any complication - I would handle them as part of backend 13:16:06 <mnasiadka> if anybody needs anything more sophisticated in future - he can add that functionality in a separate patch? 13:16:10 <mnasiadka> Any opinions? 13:16:31 <IvanVnuko[m]> It's mainly about galera, rabbitmq needs distinct certificates for it uses names 13:17:13 <IvanVnuko[m]> but thax, I did it as a part of backend, but maybe galera replication is another beast, so I'm asking for advice... 13:17:33 <SvenKieske> agreed 13:17:51 <IvanVnuko[m]> and a bit of promotion for the TLS encryption change requests we have put up:) 13:17:52 <SvenKieske> I would use the same certs for a start where possible. 13:17:54 <IvanVnuko[m]> thank you 13:17:59 <SvenKieske> we can make it complicated later ;) 13:19:08 <IvanVnuko[m]> yeah, was my initial idea also, just that there are some rumors about a bit complicated restart of galera cluster when new certificates are set up, so... 13:19:40 <mnasiadka> ok, then agreed - keep it simple as possible 13:19:42 <mnasiadka> next one 13:19:45 <mnasiadka> jovial [21st] Should we lower minimum ansible as we should technically support python 3.9 in 2024.2 13:19:45 <mnasiadka> https://governance.openstack.org/tc/reference/runtimes/2024.2.html 13:19:45 <mnasiadka> 2.15.x is last version support on 3.9 13:19:45 <mnasiadka> Also affects kayobe 13:20:23 <jovial> Am I right in thinking we should be supporting 3.9 still? 13:20:29 <mnasiadka> we are supporting 3.9 13:20:30 <frickler> can't you use 3.11 on rocky? 13:20:32 <mnasiadka> on the destination hosts 13:20:38 <mnasiadka> you can use 3.11 or 3.12 13:20:49 <mnasiadka> my problem with relaxing the requirement is that we would need to have that running in CI 13:20:56 <mnasiadka> so that's a bit more jobs that we currently have 13:21:07 <mnasiadka> and we never supported more than two versions per release 13:22:10 <jovial> Just wondering if I should follow suit in Kayobe and drop 3.9 really 13:22:34 <SvenKieske> mhm, maybe ask TC (frickler? ;) ) if it's a breakage of the requirements? I mean technically you can run openstack with python 3.9, it's just not what we test or deploy? use another deployment tool? ;) 13:22:40 <mnasiadka> well, as I mentioned - Ansible requires 3.10+ only on the host where you're running Ansible 13:23:22 <frickler> yes, I don't think the PTI applies in this case 13:23:42 <mnasiadka> Python 3.9 (available as default in Debian 11 and Centos 9 Stream) Python 3.9 is the the minimum supported/required version for 2024.2. This does not imply that Debian 11 is the minimum supported distribution but it may be used for tox testing. supporting Python 3.9 does not require full tempest testing, but py39 unit tests are expected as a minimum requirement for all Python projects. The minimal requirement for testing jobs 13:23:42 <mnasiadka> against Python versions above is to ensure language compatibility, having more extensive testing is allowed. 13:23:56 <mnasiadka> not reminding that the PTI claims that you need to have passing unit tests on Py3.9 ;-) 13:24:03 <mnasiadka> and nothing more 13:24:13 <frickler> IMO kolla isn't a python project in the sense of that whole document 13:24:29 <mnasiadka> well, kolla-ansible more likely 13:24:38 <frickler> kolla* even 13:25:02 <mnasiadka> I don't think we should add additional work of supporting older Ansible release 13:25:17 <frickler> +1 13:25:18 <SvenKieske> +1 13:25:32 <kevko> hmm, are u sure that https://review.opendev.org/c/openstack/kolla-ansible/+/921743 didn't break anything ? 13:25:32 <mnasiadka> And for those sticking to SLURP - they will still need to upgrade to Rocky 10 before jumping from 2024.1 to 2025.1 13:26:15 <mnasiadka> kevko: I'll fix the rocky upgrade - missed that, but thanks for reminding 13:26:56 <SvenKieske> shortly afk.. 13:27:04 <mnasiadka> well, actually it's not that patch that broke it 13:27:27 <jovial> OK, I will following suit in Kayobe then to keep the test matrix smaller (and it makes sense if we are aligned to kolla). Was just checking that it wasn't an oversight. 13:28:02 <mnasiadka> jovial: no, I was thinking about that in the past, but it just doesn't make any sense if you can install python3.11/3.12 on Rocky 9 and get it working that way 13:28:33 <kevko> mnasiadka: it's actually this one https://review.opendev.org/c/openstack/kolla/+/924245 13:28:43 <kevko> revert ? or what ? 13:29:02 <mnasiadka> what did it break? 13:29:10 <kevko> build rocky 13:29:19 <opendevreview> Merged openstack/kolla-ansible stable/2023.1: Fix post-config of OVS for manila-share servers https://review.opendev.org/c/openstack/kolla-ansible/+/926030 13:29:19 <kevko> kolla-toolbox 13:29:28 <mnasiadka> where? 13:29:34 <mnasiadka> what's the error? 13:30:29 <kevko> mnasiadka: validating ..give me 5 minutes 13:30:46 <mnasiadka> kevko[21.st aug] https://review.opendev.org/q/owner:michal.arbet@ultimum.io+-status:abandoned+-status:merged+branch:master+(repo:openstack/kolla+OR+repo:openstack/kolla-ansible) 13:30:50 <mnasiadka> 5 minutes is too much 13:30:58 <mnasiadka> I'll leave it here - since kevko did not add any comments 13:31:35 <kevko> :D 13:31:42 <mnasiadka> #topic Open discussion 13:31:46 <mnasiadka> anybody anything? 13:31:51 <kevko> thank you very much 13:31:58 <SvenKieske> the usual begging for reviews ;) 13:32:07 <mmalchuk> orphaned backports 13:32:09 <mmalchuk> https://review.opendev.org/c/openstack/kolla-ansible/+/926186 13:32:13 <mmalchuk> https://review.opendev.org/c/openstack/kolla-ansible/+/924573 13:32:36 <mmalchuk> and also please review: 13:32:36 <SvenKieske> one question though, regarding my WIP prometheus config validation: https://review.opendev.org/c/openstack/kolla-ansible/+/926209 13:32:37 <mmalchuk> https://review.opendev.org/c/openstack/kolla/+/926084 13:32:41 <mmalchuk> https://review.opendev.org/c/openstack/kolla-ansible/+/926190 13:33:23 <SvenKieske> would it be ok for a start to only validate the prometheus.yml inside the main prometheus-server container? because we have dozens of exporters and constructing the loop for that makes my head hurt :) 13:33:34 <mnasiadka> it's unmaintained, it should be orphaned 13:34:14 <mnasiadka> SvenKieske: I would say we don't care about exporters - that should be validated in the CI if we're able to scrape from them 13:34:39 <mmalchuk> mnasiadka not all 13:34:44 <SvenKieske> alright, so that will simplify the validation a lot, I'll just check the main prometheus server configs then :) 13:34:47 <mmalchuk> https://review.opendev.org/c/openstack/kolla-ansible/+/924573 13:34:54 <mmalchuk> is the A 13:35:16 <mmalchuk> https://review.opendev.org/q/Ic8e2dd42b95a8f5c2141a820c55642a3ed7beabd 13:35:34 <mmalchuk> month ago merged all but A lost 13:36:50 <mnasiadka> bbezak_alt: can you have a look ^^? 13:37:04 <bbezak_alt> looking 13:37:16 <bbezak_alt> some already done 13:37:31 <mmalchuk> thanks a lot 13:38:06 <mmalchuk> still lack of Kayobe reviews :( 13:38:08 <mmalchuk> https://review.opendev.org/c/openstack/kayobe/+/921628 13:38:17 <mmalchuk> 2 months( 13:38:21 <mnasiadka> jovial: ^^ 13:38:59 <mmalchuk> what about new core reviewers for Kolla/Kayobe ? outside the StackHPC ? 13:40:11 <mnasiadka> I'll leave that to jovial since he's more responsible for Kayobe 13:40:32 <mnasiadka> Ok, unfortunately I need to run - see you all next week! 13:40:35 <mnasiadka> #endmeeting