Tuesday, 2026-02-17

daleeshello, anyone around for the meeting this evening?08:00
dalees(or morning)08:00
jakeyiphi dalees I am here08:03
daleeshi jakeyip , well let's start a meeting and see where we get to.08:03
dalees#startmeeting magnum08:03
opendevmeetMeeting started Tue Feb 17 08:03:41 2026 UTC and is due to finish in 60 minutes.  The chair is dalees. Information about MeetBot at http://wiki.debian.org/MeetBot.08:03
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.08:03
opendevmeetThe meeting name has been set to 'magnum'08:03
dalees#topic roll call08:03
daleeso/08:03
jakeyipo/08:03
daleesanyone else?08:05
dalees#topic open discussion08:05
daleesnothing particular on the agenda; did you have anything to bring up, Jake?08:06
mnasiadkaHola!08:06
daleeshi mnasiadka 08:06
jakeyiphi mnasiadka 08:06
mnasiadkaI’m still off, but will be back in March - just lurking :)08:06
jakeyipI don't have much, need to talk abt PTL?08:09
BigMaxGood morning o/08:09
daleesYes sure; two things about PTL - firstly thanks for putting your name forward mnasiadka for next cycle. 08:10
mnasiadkaNo problem :)08:11
daleessecondly, I'm moving on from my current role in a couple of weeks and won't have time for Magnum contributions. Do you want to pick up the PTL role early, mnasiadka ?08:11
mnasiadkaYes, that’s fine08:12
mnasiadkadalees: is there anybody picking up your old role at Catalyst and we should expect them being active here?08:12
daleesthank you, are you okay with mnasiadka picking up PTL for the remainder of this cycle, jakeyip ?08:13
jakeyipyes definitely08:13
daleesmnasiadka: unsure at this stage, mattnorthcott who contributed a feature last year will hopefully make more appearances in patchsets.08:14
jakeyipcatalyst will still be running magnum?08:15
dalees#action dalees to pass PTL to mnasiadka for remainder of this cycle, over the next week or two.08:16
jakeyipI wonder how to attract more ops08:16
daleesjakeyip: yes, definitely08:16
daleesjakeyip: yeah, it's tricky. I will try and encourage them to join in. Timezones are one thing (it's 9pm) but also the priority of features vs upgrades/ops.08:19
jakeyipdalees: not saying catalyst, but operators of clouds in general. surely can't be catalyst, nectar, stackhpc running magnum?08:19
mnasiadkaRunning magnum-capi-helm rather08:20
mnasiadkaAnd I can’t promise helm charts from Azimuth will be the same quality, since johnthetubaguy is moving on as well08:21
mnasiadkaI think we need to be more integrated with other drivers out there to have a connection to users08:21
jakeyipi wonder how vexxhost is doing08:23
jakeyipmnasiadka: how's the future for magnum-capi-helm looking? should we look at magnum-cluster-api too? 08:25
mnasiadkaI have no clue, but I think magnum-cluster-api has more traction in the wider community08:26
daleesseems to, though only a few companies contributing to its development. I see Vexxhost and BBC08:27
mnasiadkaI’ll try to get more people working on capi-helm in SHPC, but it won’t help getting it more widely adopted, just a bit more maintained08:27
jakeyipwell, every project has issues, but it's better tackled together than separately08:27
daleesthe magnum-capi-helm driver is fairly stable. There are certainly features I'd like to add, but they are also for Magnum core. Otherwise the charts are the things that need maintenance now, to keep them moving and relevant.08:28
mnasiadkaWell, obviously ideally we should have one driver that fits all, but that’s not the case08:28
jakeyipdalees: hm, true, I wonder if vexxhost also suffers from the nodegroups issues08:30
mnasiadkajrosser: around? How’s your experience with magnum-cluster-api? ^^08:31
mnasiadkaI think SCS also ships with magnum-cluster-api but they are not active here08:34
daleesjakeyip: which nodegroups issues? I know there are some just wondering which you mean.08:35
jakeyipthe big effort to make nodegroups be able to inherit and have separate labels, instead of using clusters'08:36
daleesoh yes, I see. mutable labels being the other feature goal, that that would help enable.08:39
jakeyipyeah it's a core piece I think08:42
jakeyipanyway, not your problem now I think! :P08:43
jrossermnasiadka: andrewbonney is doing a lot of work here with magnum-cluster-api, maybe can comment more accurately than me08:43
andrewbonneyaside from specifics it's working for us at present. we have some current issues moving forward with upgrades due to deployment assumptions in the driver which we've reported upstream08:47
opendevreviewGrzegorz Bialas proposed openstack/magnum-capi-helm master: Add config to add helm values  https://review.opendev.org/c/openstack/magnum-capi-helm/+/95196608:50
daleesjakeyip: no I won't be continuing now, but I'll pass my PoC on. I don't think it's the right Magnum to Driver interface yet but perhaps I should upload to gerrit anyway as 'draft' or 'closed' so it exists somewhere.08:54
jakeyipdalees: yeah title with WIP is good too, we may be able to pick it up.08:55
daleesalso that helm config patch would be good to get some eyes on, it would have helped with a recent issue where k8s dashboard helm chart repo moved locations.08:55
daleeshelm config patch: https://review.opendev.org/c/openstack/magnum-capi-helm/+/95196608:56
* dalees will close the meeting shortly.08:59
daleesthanks for joining and coding with us!09:02
dalees#endmeeting09:02
opendevmeetMeeting ended Tue Feb 17 09:02:12 2026 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)09:02
opendevmeetMinutes:        https://meetings.opendev.org/meetings/magnum/2026/magnum.2026-02-17-08.03.html09:02
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/magnum/2026/magnum.2026-02-17-08.03.txt09:02
opendevmeetLog:            https://meetings.opendev.org/meetings/magnum/2026/magnum.2026-02-17-08.03.log.html09:02
opendevreviewNikita Vokhmintsev proposed openstack/magnum master: resolve resource names to UUIDs when creating cluster templates  https://review.opendev.org/c/openstack/magnum/+/97139309:06
jakeyipthanks dalees . hm interesting, it's either use that patch, or roll a new CT with the updated values and upgrade clusters to it?09:06
daleesjakeyip: injecting helm config overrides in lets you make customisations to values without actually forking and publishin your own charts (we do, it's okay with some pipelines). However the k8s dashboard would have required us to change every published chart. Adding a helm values override would have let us fix it in a Magnum config file.09:08
jakeyiphow many charts will need to be changed in your env? and they should be updated anyway? is this a temp patch thing while those charts are rebuild or will you carry this change for a long time?09:12
jakeyipsorry, it's a bunch of Q, can answer another day, I know it's late on your end09:12
daleesit's a good question though. We'll carry it for now, and just remove it when we remove k8s dashboard in k8s v1.35. so it'll phase out. Not many really installed it except our CI, which started failing.09:13
daleesthe issue, in case you didn't encounter: https://github.com/azimuth-cloud/capi-helm-charts/issues/65209:14
dalees(by CI I meant Tempest, which runs regularly)09:15
* dalees departs for now. Happy to reply to other things tomorrow!09:16
mnasiadkaandrewbonney: do you have some link to that report?09:56
andrewbonneysure, original report in https://github.com/vexxhost/magnum-cluster-api/issues/868, which is now covered by https://github.com/vexxhost/magnum-cluster-api/issues/88109:57
mnasiadkaandrewbonney: ah, that,s interesting - probably comes from the fact that vexxhost runs their stack inside a k8s cluster10:07
andrewbonneyyes, we assumed the same10:11
opendevreviewNikita Vokhmintsev proposed openstack/magnum master: resolve resource names to UUIDs when creating cluster templates  https://review.opendev.org/c/openstack/magnum/+/97139310:20
jrossermnasiadka: something that vexxhost have put a lot of effort into is getting magnum-cluster-api into the community deployment tools11:42
jrosseri think that this is probably why from an ops/end user perspective there seems to be a lot of traction. certainly from an OSA point of view its just working out-of-the-box, including deployment of the management k8s cluster11:43
mnasiadkajrosser: maybe in OSA and OSH case, but not really in K-A case11:43
mnasiadkaBut yes, I agree11:44
jrosserwhilst i have little/no experience with the stackhpc one, my impression is you need to have more in depth k8s understanding as an operator to get things working11:44
mnasiadkaI would like them to be more part of the community, instead of being a bit outside, but that will probably include cooperation in other direction11:45
jrosserandrewbonney has been looking at what we have already got in OSA to re-use for deploying the stackhpc driver11:45
jrosserit looks like relatively little change on top of what we have already11:45
mnasiadkawell, it's not really stackhpc driver, because it's in opendev and has some contributions from other companies11:46
jrosseryes of course11:46
mnasiadkaBut I think first we need magum-core running CI against these two drivers, which I'll work on in the coming cycle11:46
jrosseryeah11:47
jrosserits not just the driver tbh11:47
jrosserhaving some good story about connectivity between the management cluster and workload clusters is very important for us11:47
jrosseri.e real world architecture rather than devstack11:48
jrosseri put a lot of effort into making that be sensible https://github.com/openstack/openstack-ansible-ops/tree/master/mcapi_vexxhost11:49
mnasiadkajrosser: well, devstack still is better than assuming your openstack services run in the same kubernetes cluster12:43

Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!