16:00:41 <bauzas> #startmeeting nova
16:00:41 <opendevmeet> Meeting started Tue Oct  8 16:00:41 2024 UTC and is due to finish in 60 minutes.  The chair is bauzas. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:41 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:00:41 <opendevmeet> The meeting name has been set to 'nova'
16:00:56 <Uggla> o/
16:00:58 <fwiesel> o/
16:01:09 <auniyal> o/
16:01:10 <elodilles> o/
16:01:26 <gibi> o/
16:01:58 <bauzas> #link https://wiki.openstack.org/wiki/Meetings/Nova#Agenda_for_next_meeting
16:02:06 <bauzas> sorry my internet seems to slow down
16:02:19 <bauzas> and I had two meetings just before
16:02:29 <bauzas> so I'm also slowing down
16:03:03 <bauzas> hah, internal error when updating the agenda
16:03:18 <bauzas> let's do it by hand
16:03:21 <bauzas> #topic Bugs (stuck/critical)
16:03:39 <bauzas> #info One Critical bug
16:03:46 <bauzas> #link https://bugs.launchpad.net/nova/+bug/2083518
16:03:56 <bauzas> #info Add yourself in the team bug roster if you want to help https://etherpad.opendev.org/p/nova-bug-triage-roster
16:04:40 <bauzas> tkajinam: around ? want to discuss distutils problem with py3.12 ?
16:07:00 <bauzas> apparently, nothing to say
16:07:33 <bauzas> I'm just not sure this is a critical bug per say, but we need to fix it as soon as we can
16:08:05 <dansmith> o/
16:09:04 <bauzas> okay, moving on anyway
16:09:15 <bauzas> #topic Gate status
16:09:22 <bauzas> #link https://bugs.launchpad.net/nova/+bugs?field.tag=gate-failure Nova gate bugs
16:09:27 <bauzas> #link https://etherpad.opendev.org/p/nova-ci-failures-minimal
16:09:35 <bauzas> #link https://zuul.openstack.org/builds?project=openstack%2Fnova&project=openstack%2Fplacement&pipeline=periodic-weekly Nova&Placement periodic jobs status
16:09:41 <bauzas> #info Please look at the gate failures and file a bug report with the gate-failure tag.
16:09:48 <bauzas> #info Please try to provide meaningful comment when you recheck
16:10:37 <bauzas> we got a couple of failures on the periodics
16:10:45 <bauzas> but :
16:10:51 <bauzas> - for py38 we should remove the job
16:11:13 <bauzas> - we know that c9s is currently failing
16:11:23 <bauzas> - ditto with the fips job
16:11:41 <bauzas> (that's because rdo hadn't yet delivered Dalmatian)
16:11:59 <bauzas> so, anything else people would want to discuss about CI ?
16:12:14 <gibi> I saw strange CI failure where the nova-compute complained that the cpus configured for VMs not available on the host. https://review.opendev.org/c/openstack/nova/+/928584/11#message-67b19ed6c1571b06bb03a4f24c903504521a88f4
16:12:18 <gibi> Invalid '[compute] cpu_shared_set' config: one or more of the configured CPUs is not online. Online cpuset(s): [0, 1, 2, 3], configured cpuset(s): [0, 1, 2, 3, 4, 5]
16:12:22 <gibi> Is this known?
16:12:53 <clarkb> gibi: cinder hit it too
16:13:07 <clarkb> I did some digging and I could only find incidence of it in rax-ord I think it was
16:13:22 <bauzas> yeah, looks like a node problem
16:13:26 <clarkb> when it occurred the product version for xen reported by ansible fact gathering is older than on successful jobs
16:13:44 <clarkb> I was going to followup with cardoe on it this week to see if rax has a bad hypervisor.
16:14:00 <clarkb> Oh also it seems to maybe only affect python3.12 so could be a combo of xen version and noble kernel
16:14:19 <gibi> I'm wondering how devstack generates cpu_shared_set config, or we just keep it empty ?
16:14:58 <gibi> clarkb: thanks
16:15:10 <clarkb> oh wait this is different
16:15:21 <clarkb> gibi: the problem you have is because rax-flex only has 4vcpu instead of 8
16:15:40 <clarkb> this is known because even with 4vcpu the jobs are running faster and there is no need to ask for an 8vcpu variant
16:16:41 <gibi> then we need to change our devstack config as it is explicitly uses 0-7 pcpus for nova
16:16:56 <clarkb> side note: I always ask people link to the job logs it makes my life much easier when debugging if I don't have to dig through gerrit change info to find the specific build in question
16:17:26 <gibi> clarkb: https://zuul.opendev.org/t/openstack/build/93e5dca3b79348e5a8d3f5fecda82382/log/controller/logs/etc/nova/nova-cpu_conf.txt
16:17:57 <gibi> anyhow I think I understand the root cause now. So we can move on
16:18:24 <gibi> we might need a smarter devstack config generation to take into account the available cpus
16:18:27 <bauzas> gibi: would you imagine some devstack check that would lookup the cpus ?
16:18:32 <bauzas> hah
16:18:41 <gibi> yeah :)
16:18:50 <bauzas> that sounds a good idea to me
16:19:13 <bauzas> anyway, let's not solve that problem now
16:19:29 <bauzas> the wikipage's agenda is now updated \o/
16:19:54 <bauzas> #topic Release Planning
16:20:01 <bauzas> #info Dalmatian GA last Wednesday
16:20:03 <bauzas> huzzah
16:20:08 <elodilles> \o/
16:20:11 <bauzas> #link https://releases.openstack.org/epoxy/schedule.html
16:20:31 <bauzas> #info as a reminder, nova Epoxy deadlines will be defined at the PTG
16:20:48 <bauzas> #topic Review priorities
16:21:42 <bauzas> #link https://etherpad.opendev.org/p/nova-2025.1-status
16:21:50 <bauzas> nothing to report
16:21:56 <bauzas> #topic PTG planning
16:22:02 <bauzas> #info as a reminder, we'll meet (virtually) at the PTG on Oct 21-25 2024
16:22:10 <bauzas> #info Please register yourself to the virtual PTG
16:22:17 <bauzas> #link https://etherpad.opendev.org/p/nova-2025.1-ptg
16:22:24 <bauzas> the etherpad is horribly empty
16:23:13 <bauzas> I know some of us could wait a bit before adding topics, but maybe some people could already know things that they're sure that they could be working
16:23:14 <gibi> then we will have a slow ptg week :)
16:23:34 <bauzas> gibi: I'm afraid this will just mean we'll have a short time for organizing ourselves
16:23:43 <bauzas> with a packed week
16:24:15 <bauzas> I also got a request from the horizon team to have a cross-project session
16:24:37 <gibi> agility :)
16:24:53 <bauzas> I'll reply to tatiana asking her when they're available during the week
16:25:06 <bauzas> gibi: please, don't say gross words
16:25:48 <bauzas> given the current agenda, I'm cool with taking one hour from the 16 I booked
16:26:23 <sean-k-mooney1> gibi: yes i know what causes that
16:26:29 <bauzas> oh, also I'll probably need to handover some Friday discussions as I would attend the TC sessions
16:26:38 <sean-k-mooney1> i shoudl go and revert the patch that added that until i can fix the it properly
16:27:07 <gibi> sean-k-mooney1: ack
16:27:09 <bauzas> anyway, moving on
16:27:11 <sean-k-mooney1> ah just caught up
16:27:16 <bauzas> #topic Stable Branches
16:27:36 <bauzas> elodilles: report :)
16:27:45 <elodilles> not much to report actually :)
16:27:48 <elodilles> #info stable/202*.* gates seem to be OK
16:27:58 <elodilles> and that's all for now :)
16:28:38 <bauzas> ++
16:28:50 <bauzas> #topic vmwareapi 3rd-party CI efforts Highlights
16:28:53 <bauzas> fwiesel: around ?
16:28:56 <fwiesel> #info https://review.opendev.org/c/openstack/nova/+/910627 merged
16:29:07 <fwiesel> So, in theory, we should be down to two failing tests...
16:29:31 <fwiesel> Theory, as the pipeline is still flaky. I have a lead, but no fix yet for that.
16:29:46 <fwiesel> After that, I tackle the last two failing tests.
16:29:56 <fwiesel> That's from my side, unless there are any questions.
16:30:12 <bauzas> cool, thanks
16:30:25 <fwiesel> (Thanks to sean-k-moonkey1 and dansmith for reviewing it)
16:30:29 <bauzas> #topic Open discussion
16:30:35 <bauzas> (tkajinam): Python 3.8 support being removed
16:30:41 <bauzas> is tkajinam around ?
16:31:31 <bauzas> oh wait, I'm burned
16:31:38 <bauzas> we discussed that last week
16:31:46 <bauzas> this just ringed a bell to my tired mind
16:31:59 <bauzas> moving on to the other item in the agenda
16:32:19 <bauzas> (auniyal) Request to approve blueprint https://blueprints.launchpad.net/nova/+spec/nova-manage-improvement
16:32:22 <bauzas> is auniyal here ?
16:32:26 <auniyal> yep
16:32:44 <auniyal> its an improvement in nova-manage cmd
16:33:06 <auniyal> to return  output in more formats such as json
16:33:26 <dansmith> doesn't this need a spec given some of the complexities around making sure there is consistency among all the commands?
16:33:29 <sean-k-mooney> so that eaiser said then done
16:33:44 <sean-k-mooney> unlike osc we are using raw prints
16:33:51 <sean-k-mooney> in nova manage
16:34:42 <gibi> at least on of the use case is coming from here https://github.com/openstack-k8s-operators/nova-operator/blob/7e7ba33aab515809fb23ce20cb6a55e50da3f457/templates/nova-manage/bin/ensure_cell_mapping.sh#L20-L25 and for that having a --cell_name param would be enough
16:35:00 <sean-k-mooney> we cnat rely on click/cliff to help use with that  generically
16:35:14 <sean-k-mooney> we could do it on a per comand basis yes
16:35:21 <bauzas> AFAIK, originally nova-manage wasn't intended to be scripted
16:35:27 <bauzas> hence the poor interface
16:35:45 <auniyal> sure, I can write a spec,  the output is formatted by prettytable mostly
16:35:54 <gibi> bauzas: deployment tooling needs to add cells, there is no other interface for that
16:36:05 <auniyal> which supports other format as well
16:36:19 <bauzas> gibi: I wasn't disagreeing your usecase, I was just explaining why we basically have that
16:36:51 <sean-k-mooney> auniyal: so for succsful output yes alot goes via prettytbale
16:36:52 <gibi> and my use case does not need json output per se, it needs a way to easily manipulate cells by name and not by uuid
16:36:58 <sean-k-mooney> error tend to just use print directly
16:37:02 <bauzas> but I agree with dansmith, there could be some concerns about ensuring that we keep the same formatted output
16:37:23 <sean-k-mooney> so lets move forward with a spec and review the proposal
16:37:28 <bauzas> agreed
16:37:29 <dansmith> gibi: ack, the cell_name thing is easy to just do without a spec, no doubt.. the json thing not so much, IMHO
16:37:52 <dansmith> I think we can just tuck the cell_name thing under the covers without a lot of paperwork, IMHO, if it's as easy as I expect
16:37:59 <dansmith> so perhaps just do that as a one-off and spec for the json stuff?
16:38:16 <auniyal> ack
16:38:19 <gibi> we can do both but I prefer to have --cell_name as that feels easier
16:38:27 <bauzas> yeah that should be two blueprints
16:38:46 <bauzas> one for adding a new param to the command, the other one about formatting outputs
16:39:05 <dansmith> idk that we need a blueprint for a single simpleish patch for cell_name, personally, but whatever the PTL says :)
16:39:13 <bauzas> and as dansmith, I'm cool with stamping over the cell param addition bp as a specless, if that helps
16:39:36 <bauzas> then the PTL says : dude, write the patch and consider it as just a techdeby
16:39:45 <dansmith> ++
16:39:49 <sean-k-mooney> if we want a tracker i guess we can use a wishlist bug
16:39:52 <bauzas> tech debt reduction
16:40:02 <dansmith> sean-k-mooney: sure
16:40:03 <sean-k-mooney> yep
16:40:13 <bauzas> let's not paper it, we now have a status etherpad for tracking such patches
16:40:25 <bauzas> but I'm cool with a wishlist bug
16:41:17 <dansmith> moving on?
16:41:23 <bauzas> yah
16:41:31 <gibi> cool
16:41:34 <bauzas> that was it in the agenda
16:41:45 <dansmith> if I can, I'd like to beg for review of this: https://review.opendev.org/c/openstack/nova/+/930754
16:41:48 <bauzas> anything else or can I just take a breath before my 4th meeting ?
16:42:00 <dansmith> definitely could use some in-depth sanity checking of my findings
16:42:10 <dansmith> but it needs to go in before glance can start enforcing image formats
16:42:19 <bauzas> I'll
16:42:52 <bauzas> I already have the context
16:43:00 <dansmith> thanks
16:43:34 <haleyb> can i ask for a review of https://review.opendev.org/c/openstack/nova/+/811521 ? just need one more +2 to finally complete this BP
16:44:15 <sean-k-mooney> oh that ya i ment to loop back to it. im also +2 on haleyb change
16:44:33 <bauzas> dansmith: I just have one open question, this will change the uploaded image type, right?
16:44:35 <haleyb> sean-k-mooney: thanks for the +2, i figured i'd wait until we were in epoxy
16:44:51 <bauzas> if so, shall we document that in a relnote ?
16:45:28 <dansmith> bauzas: it doesn't upload an image
16:45:40 <dansmith> bauzas: it just creates one and then does a zero-byte upload to make it active
16:46:02 <dansmith> bauzas: since it claims it will be qcow2, zero bytes is not a valid qcow2 file, so glance rejects it
16:46:09 <bauzas> ah I see
16:46:14 <dansmith> we really shouldn't be doing this at all, IMHO, but alas
16:46:19 <dansmith> (I thought you already had context? :D)
16:46:29 <bauzas> we just ping glance just for the sake of flipping something on our side, naice
16:46:43 <dansmith> we just LIE to glance, but yeah :D
16:46:58 <bauzas> dansmith: I had the context, but I missed that detail
16:47:06 <sean-k-mooney> its not really that, we are tryign to make bfv snapshot look like boot form image snapshots
16:47:09 <sean-k-mooney> but they are not
16:47:12 <sean-k-mooney> the same
16:47:19 <bauzas> and trust me, I refrain myself as hard as a I can from stepping into that image mud
16:47:25 <dansmith> that's the reason for the lie yeah
16:48:01 <bauzas> okay, now I have everything I need for reviewing that patch
16:49:25 <dansmith> thanks
16:49:54 <sean-k-mooney> bauzas: gibi  woudl either of ye have bandwith to review haleyb patch
16:50:10 <sean-k-mooney> bauzas: you have seen that before too a few months ago
16:50:23 <bauzas> yeah
16:50:26 <bauzas> I'll
16:50:27 <sean-k-mooney> cool
16:50:48 <bauzas> that's it for me
16:50:55 <bauzas> anything else last minute ?
16:51:08 <auniyal> sean-k-mooney, can you please loop back at this too https://review.opendev.org/c/openstack/nova/+/929858
16:51:31 <sean-k-mooney> yes i can try
16:51:42 <auniyal> thanks
16:51:53 <sean-k-mooney> i asked you to ping me if i didnt do it last week so thansk for the reminder
16:51:55 <gibi> sean-k-mooney: I can try too
16:52:12 <bauzas> okay, that's it for the meeting then
16:52:14 <bauzas> thanks all
16:52:19 <bauzas> #endmeeting