14:00:09 <whoami-rajat> #startmeeting cinder 14:00:09 <opendevmeet> Meeting started Wed Oct 12 14:00:09 2022 UTC and is due to finish in 60 minutes. The chair is whoami-rajat. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:09 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:09 <opendevmeet> The meeting name has been set to 'cinder' 14:00:12 <whoami-rajat> #topic roll call 14:00:16 <jungleboyj> o/ 14:00:21 <rosmaita> o/ 14:00:28 <e0ne> hi 14:00:34 <rosmaita> e0ne: !!!!! 14:00:44 <eharney> hi 14:00:47 <aneeeshp1> hi 14:00:47 <felipe_rodrigues> o/ 14:00:57 <e0ne> rosmaita: I'm glad to see all of you too! 14:01:01 <harshailani> hi 14:01:02 <kgube> hi! 14:01:05 <simondodsley> hi 14:01:17 <rosmaita> e0ne: crazy times for sure, really happy to see you 14:01:18 <mubeen> hi 14:01:35 <whoami-rajat> we've shifted etherpad since we're into the antelope release now 14:01:37 <whoami-rajat> #link https://etherpad.opendev.org/p/cinder-antelope-meetings 14:01:38 <enriquetaso> hi 14:02:00 <TusharTgite> hi 14:02:30 <sfernand> hi 14:03:53 <whoami-rajat> good amount of people have shown up, let's get started 14:03:58 <whoami-rajat> #topic announcements 14:04:05 <whoami-rajat> Welcome to the first meeting of Antelope Cycle! 14:04:25 <whoami-rajat> and related announcement, OpenStack Zed is released 14:04:32 <whoami-rajat> #link https://lists.openstack.org/pipermail/openstack-announce/2022-October/002061.html 14:04:44 <rosmaita> \o/ 14:05:09 <whoami-rajat> thanks everyone for the contributions!! 14:05:18 <harshailani> yay :) 14:05:35 <rosmaita> whoami-rajat: thanks for the leadership! congratulations on your first release as PTL! 14:05:41 <jungleboyj> We made it through the Alphabet! 14:06:05 <rosmaita> i am still bummed that we are not re-starting with Aardvark 14:06:24 <whoami-rajat> rosmaita, yes, feels good, thanks for the guidance throughout, wouldn't have been possible without that :) 14:07:00 <jungleboyj> rosmaita: ++ 14:07:45 <whoami-rajat> rosmaita++ jungleboyj++ 14:07:55 <whoami-rajat> so moving on, PTG upcoming 14:07:58 <sfernand> rajat+++ 14:08:06 <whoami-rajat> we've PTG next week 14:08:18 <whoami-rajat> sfernand, thanks 14:08:27 <whoami-rajat> #link https://etherpad.opendev.org/p/antelope-ptg-cinder-planning 14:09:01 <whoami-rajat> we've some topics added to the etherpad, which is good but please try to add it as soon as possible since we've very less time 14:09:15 <whoami-rajat> if anyone has any topic i mean ^ 14:09:38 <whoami-rajat> I've allocated 4 hours everyday in the PTG schedule 14:10:13 <whoami-rajat> the last hour will be flexible if the discussions extend or for other purposes 14:10:18 <whoami-rajat> #link https://ptg.opendev.org/ptg.html 14:11:10 <whoami-rajat> I still need to update few things like the etherpad and video meeting link but generally we will be meeting 1300-1700UTC everyday from 18-21 Oct 14:11:59 <harshailani> okie 14:12:18 <rosmaita> the TC & Community meetup is 1500-1700 UTC on monday 14:12:47 <whoami-rajat> also we won't have upstream meeting next week for obvious reasons but i will send out a mail for that next week 14:14:06 <whoami-rajat> rosmaita, yes 14:14:15 <whoami-rajat> this is the etherpad for the same 14:14:16 <whoami-rajat> #link https://etherpad.opendev.org/p/tc-leaders-interaction-2023-1 14:14:55 <whoami-rajat> so go through the PTG schedule for the sessions you would like to attend 14:15:41 <whoami-rajat> next is a related announcement, we've shifted the operator hour from 1400 UTC to 1500 UTC on tuesday since there was a conflict with nova 14:17:11 <whoami-rajat> that's all i had for the announcements 14:17:23 <whoami-rajat> does anyone have anything else to announce? 14:18:29 <whoami-rajat> guess not 14:18:33 <whoami-rajat> let's move to topics then 14:18:38 <whoami-rajat> #topic extending attached volumes for remotefs drivers 14:18:41 <whoami-rajat> kgube, that's you 14:18:53 <kgube> alright 14:19:24 <kgube> remotefs drivers resize their volume files with qemu-img 14:19:49 <kgube> which does'nt work for attached volumes because qemu locks the files 14:20:20 <kgube> (which is a good thing, concurrent updates to qcow2 can corrupt the file) 14:21:20 <kgube> the qemu monitor supports a block-resize command that can be used to resize attached disk images 14:22:37 <kgube> nova already makes this available through the external server event API 14:22:38 <felipe_rodrigues> we have a patch: https://review.opendev.org/c/openstack/cinder/+/739079/ 14:23:00 <kgube> but there is no feedback on the success of the resize operation 14:23:27 <kgube> I've submitted a nova-spec that proposes a new synchronous API that would give feedback 14:23:47 <kgube> but I'm not sure if this the correct approach 14:24:04 <kgube> https://review.opendev.org/c/openstack/nova-specs/+/855490 14:24:10 <kgube> this is the spec 14:25:56 <eharney> it seems like a reasonable idea, similar to how we do assisted snapshots 14:26:48 <kgube> one big question is, is it okay to do this synchronously? 14:27:26 <whoami-rajat> I don't think cinder has any mechanism like external events in nova to accept updates asynchronously 14:28:09 <eharney> we made one for assisted snapshots, could have a similar mechanism for this 14:28:32 <whoami-rajat> ok 14:28:59 <kgube> there are callbacks, like os-migrate_volume_completion 14:29:01 <eharney> i'm unsure about it being synchronous when it comes to handling NFS errors that could take minutes to recover from 14:29:07 <eharney> yeah 14:29:53 <whoami-rajat> kgube, yeah, forgot about the migration case, we can do it like that as well 14:30:03 <kgube> if we need a callback then it might not even be necesserary to add a new nova API 14:30:30 <kgube> and we could keep the external server event to trigger it 14:31:26 <kgube> eharney, are NFS errors a common thing? 14:32:22 <eharney> probably common enough that we should consider what cleanup is required if this fails part-way through 14:32:54 <kgube> qemu disables preallocation for block-resize, so it should always be fast otherwise 14:34:45 <kgube> and then there is a pretty dirty alternative, which would be polling "qemu-img info" on the volume file until the size is the expected one 14:35:24 <kgube> but it wouldn't require any API change at all 14:37:24 <kgube> but failure would only be detectable by timeout 14:39:10 <eharney> is there a path in nova that resizes the volume if it's attached to nova but the instance isn't running? iirc we have that for other operations 14:39:31 <whoami-rajat> i think we use the polling mechanism for online snapshot creation but I think there were concerns raised to that approach and moving it to external events 14:40:08 <eharney> yes, there are better ways to handle the info updates about assisted snapshots that didn't exist when it was created, it would be possible to rework that 14:41:01 <kgube> eharney, there is no path yet, afaik 14:41:45 <kgube> The patch that felipe_rodrigues posted tries always to resize attached volumes 14:42:12 <kgube> and only handles them differently when they fail with the right error message 14:42:33 <kgube> s/they fail/this fails/ 14:42:56 <whoami-rajat> kgube, but nova can check if the guest is active and perform operations based on that right? 14:43:09 <kgube> yeah 14:43:19 <eharney> i'm not sure what happens when you resize the volume file while nova already has it partially attached.. 14:43:26 <eharney> seems like it would cause problems ? 14:43:56 <sfernand> Felipe proposed the idea to let Cinder handles the resize if the nova instance is shutdown by just trying and in case of a error check if qcow2 is locked 14:44:08 <sfernand> what do yu mean by partially attached? 14:44:33 <eharney> the stage between when nova gets info/metadata about the volume and when the attachment process is actually completed 14:44:40 <eharney> or when the instance is shutoff but Nova still has an attachment record 14:45:08 <eharney> something presumably has to let it know that the volume file size changed, because i think it's stored in the BDM 14:45:49 <sfernand> i don't know for stage between metadata but we tested for the case where Nova was shutoff and it works fine 14:46:02 <kgube> Yeah, thats the original purpose of the volume-extended external server event 14:46:07 <sfernand> when nova is put back online the vm get the new size 14:46:54 <kgube> eharney, to notify nova of a changed volume size, i mean 14:47:05 <eharney> if you are using the code in patch 739079, does it send an external server event? 14:47:33 <felipe_rodrigues> If I remember correctly, there is a problem with that approach when the VM is stopped ... The lock keeps there but the Nova does not extend 14:47:42 <kgube> the event is always sent when extending attached v9olumes 14:48:20 <kgube> felipe_rodrigues, oh, thats a problem 14:49:10 <whoami-rajat> I don't feel good interrupting the discussion but I think this is something worth discussing in a PTG session with all the members around 14:50:00 <whoami-rajat> since we've limited time here, and also another topic to discuss, it would be good to continue this during PTG 14:50:01 <felipe_rodrigues> do we have a slot for this ? 14:50:03 <whoami-rajat> #link https://etherpad.opendev.org/p/antelope-ptg-cinder-planning 14:50:12 <eharney> yeah, i'm happy to help on this area where i can 14:50:36 <whoami-rajat> felipe_rodrigues, a topic needs to be added and i will arrange a slot/schedule for topics with their respective date and time 14:50:46 <jungleboyj> ++ 14:51:38 <kgube> ++ 14:52:19 <whoami-rajat> great, thanks for initiating this again kgube, let's move on to the next topic 14:52:35 <whoami-rajat> #topic Outreachy applications started 14:52:38 <whoami-rajat> enriquetaso, that's you 14:52:55 <enriquetaso> hey 14:53:18 <enriquetaso> Outreachy applicants period started this week (until the end of November). 14:53:31 <enriquetaso> Because of that cinder channel is going to have some messages from potential interns. 14:53:47 <enriquetaso> Most probably they are pinging me or Rajat but Please feel free to reply to them and help them with their cinder/devstack questions :D 14:53:56 <jungleboyj> Cool. 14:53:59 <enriquetaso> Thanks! 14:54:05 <enriquetaso> that's all :P 14:54:24 <enriquetaso> jungleboyj++ 14:54:28 <harshailani> sure 14:54:51 <whoami-rajat> thanks enriquetaso for mentoring the outreachy interns! 14:55:05 <enriquetaso> thanks whoami-rajat for also mentoring! 14:55:37 <whoami-rajat> :) 14:56:00 <whoami-rajat> since we don't have any more topics, let's move to open discussion for last 5 minutes 14:56:05 <whoami-rajat> #topic open discussion 14:56:15 <sfernand> maybe a onboard session in ptg to would help them 14:56:41 <jungleboyj> sfernand: That is a good idea. We haven't done an onboarding session for people in a while. 14:56:59 <enriquetaso> when was the last time? 14:57:07 <sfernand> some similar to what we did in the openstack day I don't know 14:57:24 <sfernand> we could try defining a few topics, I could help with a quick overview on devstack 14:57:25 <whoami-rajat> sfernand, jungleboyj good idea 14:57:26 <jungleboyj> Well, the last time I did one was the first Berlin I think? 14:57:54 <jungleboyj> I don't know that there has been one since? 14:58:00 <simondodsley> jungleboyj: was that recorded and is it available? 14:58:13 <rosmaita> we may have done one at shanghai but no one showed up 14:58:15 <enriquetaso> sounds good, i'll look for a reference 14:58:23 <enriquetaso> oh :/ 14:58:26 <jungleboyj> I recorded one in the past. Let me Check YouTube. 14:58:34 <enriquetaso> sfernand++ 14:58:36 <sfernand> rosmaita: yes I remember that :P 14:58:51 <rosmaita> other than that, i never did one as PTL 14:59:02 <whoami-rajat> rosmaita, yeah, i remember doing one but don't remember any discussion from it, maybe that's why :D 14:59:03 <rosmaita> so it's been quite a while 14:59:23 <whoami-rajat> surely we can have a quick onboarding towards the start of PTG 14:59:51 <whoami-rajat> and if sfernand would like to add a session with introduction to devstack or similar, we can find a slot for that as well 15:00:00 <whoami-rajat> s/towards/at 15:00:22 <sfernand> yes I can do that 15:00:31 <whoami-rajat> great 15:00:53 <whoami-rajat> so again reminding everyone to add topics to the planning etherpad 15:00:54 <enriquetaso> thanks sfernand 15:00:57 <whoami-rajat> https://etherpad.opendev.org/p/antelope-ptg-cinder-planning 15:00:59 <jungleboyj> Sounds good. Even just a time at the beginning to see who is new. 15:01:01 <enriquetaso> i can help with that 15:01:13 <enriquetaso> maybe a cinder 101 too i don't know, 15:01:16 <sfernand> enriquetaso: :) 15:01:25 <whoami-rajat> jungleboyj, ++ enriquetaso ++ 15:01:36 <whoami-rajat> time is up, thanks everyone, see you at the PTG! 15:01:38 <simondodsley> I have people in India who would love those sessions, so please record them 15:01:45 <whoami-rajat> #endmeeting