09:02:05 <jakeyip> #startmeeting magnum
09:02:05 <opendevmeet> Meeting started Wed Feb  7 09:02:05 2024 UTC and is due to finish in 60 minutes.  The chair is jakeyip. Information about MeetBot at http://wiki.debian.org/MeetBot.
09:02:05 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
09:02:05 <opendevmeet> The meeting name has been set to 'magnum'
09:02:15 <jakeyip> #link https://etherpad.opendev.org/p/magnum-weekly-meeting
09:02:40 <jakeyip> Please put your topics into to Agenda
09:02:44 <jakeyip> #topic Roll Call
09:02:49 <jakeyip> o/
09:02:52 <dalees> o/
09:04:27 <jakeyip> hm, I think mnasiadka is around, let's continue
09:04:33 <mnasiadka> o/
09:04:37 <jakeyip> hi! :)
09:04:37 <mnasiadka> (I am, but sleepy)
09:04:41 <jakeyip> #topic Review Action Items
09:04:57 <jakeyip> improve driver discovery implementation
09:04:59 <jakeyip> https://review.opendev.org/c/openstack/magnum/+/907297
09:05:38 <jakeyip> I've (finally) finished this. please review and comments.
09:05:51 <jakeyip> got attention of mnaser too, which is a good thing
09:06:09 <jakeyip> I
09:06:31 <jakeyip> I'll still need to respond to his comments
09:06:35 <dalees> thanks for the work on this jakeyip - I will review this week. It looks good and allows for extension if we want to expose driver names to the API.
09:07:10 <dalees> the second part I wasn't expecting, but it makes sense to build in.
09:07:24 <jakeyip> hm, what is the second part?
09:08:15 <dalees> sorry, by second part i just meant allowing extension into the API in future.
09:08:28 <jakeyip> oh I see, ok :)
09:09:00 <jakeyip> ok let's continue on
09:09:21 <jakeyip> Control plane resize https://review.opendev.org/c/openstack/magnum/+/906086
09:10:17 <jakeyip> dalees: I have left some comment for you. I remember your concern about exposing driver details to API. there's a different place we can do validation
09:11:07 <dalees> thanks for that, I've not revisited yet but need to get back to this patchset soon.
09:11:25 <jakeyip> no worries, thanks for proposing that
09:12:43 <jakeyip> alright next topic. Cilium driver + validation
09:13:38 <jakeyip> the two patches are in https://review.opendev.org/c/openstack/magnum/+/905427 https://review.opendev.org/c/openstack/magnum/+/904835
09:14:25 <jakeyip> I think needs to be reworked. it clashes with our driver improvement code. dalees was on this previously, what do you think?
09:15:21 <dalees> I'm not sure the second proposed patchset clashes? https://review.opendev.org/c/openstack/magnum/+/905427
09:16:42 <jakeyip> that one is ok, I'm not sure if this means https://review.opendev.org/c/openstack/magnum/+/904835 should be abandoned then?
09:17:00 <dalees> but broadly, I want to see Magnum defer as much of this to drivers as possible. For now this addition of 'cilium' seems like a reasonable inclusion to get out of the way of progress on the CAPI driver side, but it's not supported at all in Magnum core.
09:17:43 <dalees> yeah - I understand the second is an alternative to the first.
09:17:53 <mnasiadka> Well, basically with out of tree drivers - it would make more sense to defer this to drivers
09:18:14 <mnasiadka> and worst case fall back to some default in Magnum core
09:19:29 <jakeyip> ok I think I need to spend more time on https://review.opendev.org/c/openstack/magnum/+/905427/
09:19:51 <jakeyip> will comment on that later.
09:20:07 <dalees> yeah, this is another validation step that is done before the driver gets involved. Ideally we move that logic further towards the driver and then validate using it.
09:21:29 <jakeyip> yeah I will see if that is possible. baking less logic into magnum allows out of tree driver to work easier.
09:22:01 <jakeyip> any other reviews to discuss?
09:23:06 <dalees> yeah; failing that I am okay with adding cilium - it's a very common CNI. If we end up in this cycle more than once though, I'd push harder for the refactor. Right now I want to support the progress of the CAPI driver(s).
09:24:16 <jakeyip> ok
09:26:00 <dalees> I just want to quickly bring attention to this one that recently merged, which means Helm can now be upgraded: https://review.opendev.org/c/openstack/magnum/+/879157
09:26:47 <dalees> but not sure there's much to discuss on this; without this, the helm version on masters is quite old.
09:27:25 <opendevreview> Michal Nasiadka proposed openstack/magnum-tempest-plugin master: Switch to yoga job to unmaintained/yoga  https://review.opendev.org/c/openstack/magnum-tempest-plugin/+/908260
09:27:45 <jakeyip> hm LGTM, I missed this somehow
09:27:55 <mnasiadka> dalees: we will do some cleanup in the helm area when switching Calico to Tigera operator Helm chart
09:28:02 <jakeyip> looks like at least one cherry-pick failed, we should fix that
09:28:35 <mnasiadka> but again bumping defaults would probably need the treatment we discussed previously
09:29:49 <jakeyip> also needs cherry-pick to 2023.2, I will make that happen
09:30:37 <dalees> oh i hadn't noticed the cherry-picks
09:31:02 <jakeyip> mnasiadka: not sure what we are trying to bump. are you planning to propose a patch?
09:31:23 <mnasiadka> jakeyip: we could bump helm version, since the current one isn't really tested against fresh k8s
09:31:58 <mnasiadka> we will propose series of patches to get better with Calico, we can discuss in those patches what to do next
09:32:13 <jakeyip> ok cool
09:33:59 <jakeyip> mnasiadka: thanks for the work on CI, now it is in a much better state, and it gives us a good place to improve from
09:34:15 <mnasiadka> np, happy to have this merged :)
09:34:46 <mnasiadka> I will work on sonobuoy tests (and probably multi-master and multi-worker scenarios) in this cycle as well
09:34:52 <jakeyip> I think one improvement will be to figure out if all the pods are ready, currently we just see if the API reports Healthy but that doesn't catch calico pods not working
09:36:02 <dalees> yep, thanks for the work on CI mnasiadka. massive
09:36:07 <jakeyip> yeah I'll be interested in what sonobuoy says. I suspect we may need to set a few more things like occm versions
09:36:12 <mnasiadka> Question is do we want to backport the CI to stable branches
09:38:56 <jakeyip> I'm happy to put that at a lower priority, after we figure out the labels that we need and update the docs accordingly
09:39:20 <jakeyip> https://docs.openstack.org/magnum/latest/user/index.html#supported-labels
09:40:32 <jakeyip> there's this crazy bunch of labels I use when manually testing. things like `cloud_provider_tag` should ideally match the k8s version etc
09:45:03 <jakeyip> I'll send a change later if this is confusing
09:46:43 <jakeyip> anything else?
09:48:30 <dalees> none from me
09:49:38 <jakeyip> ok.
09:50:14 <jakeyip> I have a question, anybody tested fcos39 ?
09:51:58 <dalees> nope, but might need to soon.
09:52:25 <jakeyip> ok
09:53:13 <jakeyip> ok if there's nothing else I'll end the meeting
09:53:26 <jakeyip> #endmeeting