16:01:40 <bauzas> #startmeeting nova 16:01:40 <opendevmeet> Meeting started Tue Sep 10 16:01:40 2024 UTC and is due to finish in 60 minutes. The chair is bauzas. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:01:40 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:01:40 <opendevmeet> The meeting name has been set to 'nova' 16:01:58 <fwiesel> o/ 16:02:07 <elodilles> o/ 16:02:10 <dansmith> o/ 16:02:11 <sean-k-mooney> o/ 16:02:12 <gmann> o/ 16:02:17 <bauzas> hey everyone 16:02:20 <bauzas> nice to see you 16:02:24 <sean-k-mooney> welcome back 16:02:35 <bauzas> sorry, was away last week for the OIF Summit Asia 16:02:36 <tkajinam> o/ 16:03:26 <auniyal> o/ 16:03:28 <bauzas> #topic Bugs (stuck/critical) 16:03:50 <bauzas> #info No Critical bug 16:03:58 <bauzas> #info Add yourself in the team bug roster if you want to help https://etherpad.opendev.org/p/nova-bug-triage-roster 16:04:14 <bauzas> any bug report people would want to discuss ? 16:04:48 <dansmith> maybe one 16:04:58 <dansmith> I filed the bug against oslo and nova, 16:05:11 <bauzas> shoot 16:05:15 <dansmith> but sean-k-mooney, are you still working on excluding the safety check for the ephemeral disks? 16:05:16 <sean-k-mooney> https://bugs.launchpad.net/nova/+bug/2079850 16:05:41 <sean-k-mooney> i have a draft of that locally the unit test are failing so im tryign to fix that 16:05:52 <dansmith> the oslo fix is +2 but it'll take a release and u-c bump before we get it 16:05:54 <dansmith> sean-k-mooney: okay cool 16:06:16 <bauzas> ergh 16:06:42 <bauzas> yet another reason to work on the image backend refactoring if you want MHO 16:06:45 <dansmith> bauzas: basically, if you have nova format your ephemerals as vfat for you, you'll hit this fail 16:07:00 <dansmith> yes, it's really stupid that we're even doing this, and it's at least partially because of the design of the imagebackend 16:07:16 <opendevreview> sean mooney proposed openstack/nova master: [WIP] Add functional repoducer for ephemeral disks https://review.opendev.org/c/openstack/nova/+/928310 16:07:17 <opendevreview> sean mooney proposed openstack/nova master: [WIP] adapt to vfat support in oslo.utils https://review.opendev.org/c/openstack/nova/+/928462 16:07:17 <opendevreview> sean mooney proposed openstack/nova master: only saftey check bootable files created form glance https://review.opendev.org/c/openstack/nova/+/928829 16:07:25 <sean-k-mooney> thats what im working on ^ 16:07:30 <dansmith> sean-k-mooney: cool thanks 16:07:34 <bauzas> formatting ephemerals to vfat is like "I don't care about my fs, dude" 16:07:45 <sean-k-mooney> still failing unit test but i should get those passing today 16:07:55 <dansmith> bauzas: right, it's pointless for us to even do that anyway, because nobody would actually use them as fat anyway 16:08:05 <dansmith> bauzas: double pointless behavior :) 16:08:13 <bauzas> I assume this would be for additional disks ? 16:08:18 <dansmith> ephemerals 16:08:30 <sean-k-mooney> i propoaded not formatign ephemeral disk by default before 16:08:42 <sean-k-mooney> we have a blueprint for that that i think we shoudl do in epoxy 16:08:47 <dansmith> yeah, sean-k-mooney I'm on board with that too 16:09:21 <bauzas> ring me a bell, I know what ephemeral disks are in nova, but I wonder *why* people would want vfat 16:09:21 <sean-k-mooney> but for right now just skipign the safty check if the disk is not bootable and thats only set when we chreat ephermal/swap disks in my patch above 16:09:42 <sean-k-mooney> bauzas: its our default when you dont set a config option or image property 16:09:45 <bauzas> probably because they think "ephemeral" means the disks are not persisted (which is not the case) 16:09:56 <sean-k-mooney> we choose it as the only format that would work for every os 16:10:23 <sean-k-mooney> not it was just because we needed something that works on linux and windows 16:10:32 <dansmith> vfat would only be for windows guests so they see something on the disk instead of blank 16:10:33 <sean-k-mooney> if you set os_type in teh image we use ntfs or ext4 16:10:37 <dansmith> but makes no actual sense 16:10:52 <dansmith> sean-k-mooney: we actually support ntfs? 16:10:54 <sean-k-mooney> or if you set the config option we use what you set as the fallback 16:11:01 <sean-k-mooney> yep 16:11:11 <dansmith> either way, we're not partitioning the disk properly anyway.. I guess I'm surprised windows is happy with a whole-disk ntfs 16:11:11 <sean-k-mooney> im not saying you should use it 16:11:15 <sean-k-mooney> but the code is there 16:11:33 <sean-k-mooney> https://docs.openstack.org/nova/latest/configuration/config.html#DEFAULT.default_ephemeral_format 16:11:34 <dansmith> yeah, I would never use linux-created ntfs for production if I was a windows user 16:12:06 <sean-k-mooney> well cloud model. as a end user you should not know if its hyperv or libvirt/qemu 16:12:30 <bauzas> but you know the image you use 16:12:34 <sean-k-mooney> its in the default section because this applies to other drivers 16:12:51 <dansmith> I mean as the operator I would never configure that, 16:13:03 <dansmith> but you're right, users may not realize they are getting ntfs created by linux that they shouldn't trust :) 16:13:31 <bauzas> agreed on not configure it if I was admin, I'd just assume nova would create me ephemerals that my instance can read 16:13:32 <sean-k-mooney> anyway we can carry this discussion forward to the reveiws 16:13:36 <dansmith> yeah 16:13:42 <sean-k-mooney> but next cycle we can add unformated and set that as the default 16:13:59 <sean-k-mooney> https://blueprints.launchpad.net/nova/+spec/default-ephemeral-format-unformated 16:14:01 <dansmith> with docs and renos suggesting to leave it that way 16:14:21 <sean-k-mooney> we can flag the option with advnaced if we like 16:14:52 <sean-k-mooney> https://docs.openstack.org/oslo.config/latest/reference/defining.html#advanced-option 16:15:39 <sean-k-mooney> anyway we can likely move on 16:15:42 <dansmith> yup 16:15:46 <bauzas> okay 16:15:58 <bauzas> action items are on reviewing the oslo.utils patch, right? 16:16:08 <dansmith> no, 16:16:21 <dansmith> that's on the oslo people and tkajinam has already +2d and asked the other cores to hit it 16:16:35 <dansmith> but again, it won't solve our problem until release/bump 16:16:53 <dansmith> the action item on us should be getting sean's actual fix working,reviewed,merged 16:16:55 <dansmith> (IMHO) 16:17:16 <sean-k-mooney> i have a patch at the end for compatiblity with the oslo change too 16:17:17 <bauzas> are we talking of https://review.opendev.org/q/topic:%22ephmeral-and-swap-backing-files%22 ? 16:17:50 <sean-k-mooney> the first 2 patches yes but ill change the topic of that to the bug topic later 16:17:58 <bauzas> cool 16:18:12 <bauzas> I guess then we can move on 16:18:22 <tkajinam> I can push the oslo.utils to gate if it doesn't get attention from the other cores and hard blocks nova release but note that we may need a few more steps (backport to 2024.1, new release in 2024.1 and req bump) 16:18:30 <bauzas> but do we feel this as a regression bug ? 16:18:32 <tkajinam> which might mean that RC1 release of nova might be delayed because of it 16:18:46 <bauzas> heh, jinxed by tkajinam 16:18:53 <tkajinam> s/2024.1/2024.2/g 16:19:01 <tkajinam> :-) 16:19:04 <sean-k-mooney> tkajinam: its not a hard block really but it is a regression that might require an rc2 16:19:11 <tkajinam> ok 16:19:12 <dansmith> it's not an oslo regression 16:19:20 <bauzas> it's a nova regression 16:19:26 <sean-k-mooney> yep 16:19:28 <dansmith> it is a nova one, but again, it feels like we need to fix our own code anyway 16:19:32 <bauzas> so we need to merge sean's patches before RC1 16:19:47 <sean-k-mooney> ideally but i need to get it passing first 16:19:48 <tkajinam> ok 16:19:51 <bauzas> or deliver RC1 and adress them in RC2 16:20:16 <dansmith> so we could aim to get sean's stuff in to fix, 16:20:22 <bauzas> sean-k-mooney: fwiw, I'm just adding to that bug report the dalmatian-rc-potential taag 16:20:27 <dansmith> but if we can't for some reason, just getting nova to use the updated oslo will paper over the problem 16:20:42 <tkajinam> makes sense 16:20:45 <bauzas> and I'm flagging that bug in the rc etherpad I'm just gonna speak about in a few 16:20:48 <sean-k-mooney> cool. im going to monitor the meetign but im goign to go back to working on this in parallel 16:21:21 <bauzas> sean-k-mooney: don't freak out, we didn't had RC2s for a while, but I feel brave enough to ship one this time :) 16:22:58 <dansmith> let's move on for real now 16:23:41 <bauzas> okay, did the paperwork 16:23:51 <bauzas> (sorry, took more time than estimate) 16:24:00 <bauzas> #topic Gate status 16:24:06 <bauzas> #link https://bugs.launchpad.net/nova/+bugs?field.tag=gate-failure Nova gate bugs 16:24:12 <bauzas> #link https://etherpad.opendev.org/p/nova-ci-failures-minimal 16:24:18 <bauzas> #link https://zuul.openstack.org/builds?project=openstack%2Fnova&project=openstack%2Fplacement&pipeline=periodic-weekly Nova&Placement periodic jobs status 16:24:29 <bauzas> nothing to relate except nova-emulation zed again 16:24:47 <bauzas> my bad, nova-emulation on antelope 16:25:09 <bauzas> but since master job run worked like a charm, let's not look at it 16:25:17 <bauzas> #info Please look at the gate failures and file a bug report with the gate-failure tag. 16:25:22 <bauzas> #info Please try to provide meaningful comment when you recheck 16:25:34 <bauzas> anything else about gate stability ? 16:26:25 <bauzas> looks not 16:26:33 <bauzas> #topic Release Planning 16:26:40 <bauzas> #link https://releases.openstack.org/dalmatian/schedule.html 16:26:41 <opendevreview> Merged openstack/os-vif stable/2024.2: Update .gitreview for stable/2024.2 https://review.opendev.org/c/openstack/os-vif/+/928353 16:26:46 <bauzas> #info Dalmatian RC1 planned this week 16:27:14 <bauzas> so, again, I'd appreciate if people chime in if they find regressions 16:27:22 <elodilles> ( there is the generated patch: https://review.opendev.org/c/openstack/releases/+/928551 ) 16:27:29 <elodilles> ( rc1 patch for nova) 16:27:43 <bauzas> we now have an etherpad for RC tracking 16:27:46 <bauzas> #link https://etherpad.opendev.org/p/nova-dalmatian-rc-potential 16:28:02 <bauzas> elodilles: thanks, just adding your nova rc1 patch to the etherpad :) 16:28:14 <elodilles> ++ 16:29:18 <bauzas> also, 16:29:43 <bauzas> maybe people forgot here that as soon as we branch master with RC1, then master becomes Epoxy :) 16:29:50 <bauzas> accordingly 16:30:04 <bauzas> #info Epoxy development phase will start as soon as we branch RC1 16:30:21 <bauzas> #link https://releases.openstack.org/epoxy/schedule.html 16:30:44 <bauzas> and accordingly, we'll use another etherpad for tracking all the work we'll be doing for Epoxy : 16:30:54 <bauzas> #link https://etherpad.opendev.org/p/nova-2025.1-status 16:31:23 <bauzas> feel free to add any blueprints you would want us to look at or any bug report you'd want to fix by Epoxy timeframe 16:32:04 <bauzas> based on my discussions at the OIF Summit Asia, I was also convinced to update our docs to mention our tracking system 16:32:25 <bauzas> #action bauzas to update the contributor docs to mention the etherpad and how to reach the nova community 16:33:11 <bauzas> that's it for me for release planning 16:33:20 <bauzas> anything else to add ? 16:33:30 <bauzas> oh 16:33:35 <bauzas> I forgot 16:33:53 <bauzas> in the rc tracking etherpad, I have a couple of patches I'd beg cores to review 16:34:12 <gibi> I went through that list so we need one more cores 16:34:20 <bauzas> particularly the highlights 16:34:23 <gibi> s/cores/core/ 16:34:27 <bauzas> gibi: thanks ! 16:34:57 <gmann> I +w few of them 16:35:13 <bauzas> gmann: thanks too 16:35:26 <bauzas> okay, I'll chase the patches and bug people eventually 16:35:29 <bauzas> moving on then 16:35:38 <bauzas> #topic Review priorities 16:35:59 <bauzas> this is currently punted to once we branch RC1 16:36:35 <bauzas> the current priorities are now clearly bugfixes, particularly if they are regression ones :) 16:36:47 <bauzas> #topic Stable Branches 16:36:55 <bauzas> elodilles: take the baton 16:36:59 <elodilles> o7 16:37:02 <elodilles> #info stable/202*.* gates seem to be OK 16:37:16 <elodilles> #info stable/2024.2 branch were cut for libraries 16:37:44 <elodilles> #info stable branch status / gate failures tracking etherpad: https://etherpad.opendev.org/p/nova-stable-branch-ci 16:38:19 <elodilles> please add gate errors in the pad if anyone finds one ^^^ 16:38:35 <elodilles> and that's all about stable branches from me 16:39:14 <bauzas> all good 16:39:34 <bauzas> #topic vmwareapi 3rd-party CI efforts Highlights 16:39:37 <fwiesel> #info No updates. 16:39:38 <bauzas> fwiesel: your time 16:39:42 <bauzas> cool with me 16:39:53 <bauzas> #topic Open discussion 16:40:14 <bauzas> we have one time that was carried over from last week, I'd say 16:40:16 <bauzas> (whoami-rajat) Add multipath_id to the BDM in case of iSCSI 16:40:25 <bauzas> whoami-rajat: are you available now ? 16:41:21 <sean-k-mooney> gibi and i have looked at the patch 16:41:40 <sean-k-mooney> effectivly they want to do this to make debuging with db dubps simplere 16:41:58 <sean-k-mooney> apprently only nova and the network backend has the multipath_id 16:42:10 <bauzas> orly? 16:42:25 <sean-k-mooney> and this is useful for them to be able to corralate the instance and the volume on the storage backend 16:42:51 <sean-k-mooney> im a little reluctant to add the multipath id just for debuging but not 100% agisnt it 16:43:01 <sean-k-mooney> i left a comment to that effect on the patch 16:43:18 <bauzas> and I guess the BDM is a blob ? 16:43:37 <bauzas> no database update needs ? 16:44:02 <sean-k-mooney> https://review.opendev.org/c/openstack/nova/+/889257 16:44:15 <sean-k-mooney> its a json blob i belive yes 16:44:46 <bauzas> I don't get why we need to persist it 16:44:55 <bauzas> couldn't we show it in the logs ? 16:45:00 <gibi> It feels wrong that nova perists a data that nova never uses and never passes along 16:45:30 <sean-k-mooney> hum https://github.com/openstack/nova/blob/master/nova/db/main/models.py#L684-L754 16:45:30 <bauzas> agreed with gibi 16:45:48 <sean-k-mooney> i guess this would be in connection_info 16:45:54 <bauzas> yeah 16:46:01 <bauzas> that's where they're stuffing it into 16:46:09 <bauzas> and that's a text filed 16:46:12 <bauzas> field 16:46:12 <sean-k-mooney> ya so im not a fan of that either 16:46:18 <bauzas> but I don't like persisting it 16:46:27 <sean-k-mooney> this is also in teh vm xml 16:46:35 <bauzas> for multiple reasons, the first one being that this info couldn't be reliable 16:46:50 <sean-k-mooney> the storage folk apparently use db dumps for debugging more then we do 16:46:58 <sean-k-mooney> i value having it in the logs and xml more 16:47:03 <gibi> hm, if this is in the xml then a proper sos report has enough info to correlate multipath_id with volume and instance 16:47:07 <sean-k-mooney> becuase we normally dont have db dumps 16:47:10 <bauzas> even with all we know about orphaned BDMs ? 16:47:39 <gibi> or they need historical data about delete VMs? 16:47:51 <bauzas> logs will tell you tho 16:47:58 <gibi> yeah we dump the xml to debug 16:48:01 <sean-k-mooney> bauzas: to move this forward can i suggest reaching out to rajat 16:48:09 <gibi> yeah we need rajat for this 16:48:10 <bauzas> also 16:48:16 <sean-k-mooney> or putting it on the ptg topic list 16:48:32 <bauzas> this would require a blueprint at least, maybe a spec if we consider the DB change 16:48:55 <bauzas> the fact that nothing in nova reads that value makes me very reluctant at least 16:49:13 <sean-k-mooney> ya i was -1 on thi initally when i spoke to them on slack and asked them ot bring it upstream 16:49:17 <bauzas> I don't want us to create a precedent with any text field just being a brown bag for DB bumps 16:49:21 <bauzas> dumps* 16:49:24 <sean-k-mooney> i agree this is not a bug but a minor feature 16:49:51 <sean-k-mooney> to me this is out os scope but i think we should let them present there case 16:50:07 <bauzas> then we can write an AI 16:50:18 <opendevreview> Balazs Gibizer proposed openstack/nova master: [doc]Developer doc about PCI and SRIOV testing https://review.opendev.org/c/openstack/nova/+/928834 16:51:07 <bauzas> #agreed this case sounds a feature request and requires a blueprint, which would describe the reason for that change and the solution 16:51:09 <dansmith> a db change for a thing we don't actually need would require a lot of justification for me to get behind :) 16:51:42 <bauzas> #agreed this blueprint would have to be discussed in a later nova meeting to see whether it would be specless or requiring a spec 16:51:52 <bauzas> dansmith: I think the meeting logs are clear :) 16:52:21 <bauzas> tbc, if nothing reads that BDM value, this is tech debt 16:52:22 <dansmith> yep, just chiming in 16:52:43 <bauzas> cool 16:52:52 <bauzas> I hope whoami-rajat will read the logs 16:52:57 <bauzas> and again, I'm open to chat 16:53:15 <bauzas> I think we're settled 16:53:29 <bauzas> any carried over item from last week or any new meat to digest ? 16:53:38 <bauzas> or can I call it a wrap and take a beer ? 16:54:07 <opendevreview> Balazs Gibizer proposed openstack/nova master: [doc]Developer doc about PCI and SRIOV testing https://review.opendev.org/c/openstack/nova/+/928834 16:54:25 <sean-k-mooney> we had one other item 16:54:38 <sean-k-mooney> can we reappove the shared security group specless blueprint 16:54:41 <sean-k-mooney> for expoy 16:54:48 <sean-k-mooney> https://blueprints.launchpad.net/nova/+spec/shared-security-groups 16:55:00 <bauzas> thanks 16:55:09 <bauzas> I haven't seen it in the agenda 16:55:18 <bauzas> I have no objection for the reapproval 16:55:26 <sean-k-mooney> i think the current state of the patch is mergable, this is the one i pinged about at the start of the meeting 16:55:38 <sean-k-mooney> so it would be a easy win after rc1 16:55:51 <bauzas> sounds a quickwin as soon as RC1 branches 16:56:00 <bauzas> yeah 16:56:12 <bauzas> okay, don't hear any disagreement, so... 16:57:00 <bauzas> #agreed https://blueprints.launchpad.net/nova/+spec/shared-security-groups is accepted again as specless bp for Epoxy 16:57:09 <bauzas> that's it for me 16:57:14 <bauzas> and we're at time 16:57:22 <bauzas> thanks all 16:57:27 <bauzas> #endmeeting