14:00:19 <rosmaita> #startmeeting cinder
14:00:19 <opendevmeet> Meeting started Wed Nov 17 14:00:19 2021 UTC and is due to finish in 60 minutes.  The chair is rosmaita. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:19 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:00:19 <opendevmeet> The meeting name has been set to 'cinder'
14:00:33 <rosmaita> #topic roll call
14:00:40 <fungi> ahoy, mateys!
14:00:40 <fabiooliveira> hi
14:00:43 <walshh_> hi
14:00:45 <LeoCampelo> hi
14:01:05 <jungleboyj> o/
14:01:10 <tosky> o/
14:01:41 <geguileo> hi! o/
14:01:55 <rosmaita> good turnout!
14:02:03 <rosmaita> #link https://etherpad.opendev.org/p/cinder-yoga-meetings
14:02:08 <rosmaita> #topic announcements
14:02:23 <rosmaita> well, the cycle is moving right along ... this week is Milestone 1 for yoga
14:02:26 <eharney> hi
14:02:33 <e0ne> hi
14:02:38 <rosmaita> luckily, we aren't required to do an M-1 release any more
14:02:59 <rosmaita> but, keep in mind that we are on third of the way through the development cycle
14:03:08 <rosmaita> ok, some upcoming events
14:03:24 <rosmaita> the OpenInfra Live: Keynotes are today and tomorrow, 1500-1700 UTC
14:03:37 <rosmaita> if you're curious, the schedule is here:
14:03:45 <rosmaita> #link https://openinfra.dev/live/keynotes?utm_id=Keynotes#schedule
14:03:47 <fungi> starting right after this meeting concludes, in fact
14:03:52 <rosmaita> exactly
14:04:07 <rosmaita> you need to register to get a link to the video session, and it is free
14:04:12 <whoami-rajat> Hi
14:04:15 <rosmaita> #link https://openinfralivekeynotes.eventbrite.com/
14:04:24 <rosmaita> next item
14:04:33 <rosmaita> our monthly Festival of XS Reviews happens on Friday this week (1400-1600 UTC)
14:04:45 <rosmaita> full info here: http://eavesdrop.openstack.org/#Cinder_Festival_of_XS_Reviews
14:05:09 <rosmaita> i will be on limited bandwidth and may not be in meetpad for that one, but i will be in irc and watching the etherpad
14:05:26 <rosmaita> next
14:05:36 <rosmaita> the stable/ussuri branch is now in Extended Maintenance mode
14:05:47 <rosmaita> thanks to whoami-rajat for getting the final ussuri releases out the door in time
14:06:21 <jungleboyj> ++
14:06:32 <rosmaita> this means no more ussuri releases, though we may merge backports of bugfixes as a courtesy to people still using that branch
14:06:46 <rosmaita> another announcement
14:07:04 <rosmaita> the openstack-wide stable core team process has changed
14:07:07 <rosmaita> details here:
14:07:16 <rosmaita> #link http://lists.openstack.org/pipermail/openstack-discuss/2021-November/025741.html
14:07:53 <rosmaita> it doesn't really affect us, except in the sense that the openstack stable-core team can't -2 our releases if inappropriate stuff slips in
14:08:08 <rosmaita> so we need to continue to exercise due diligence
14:08:26 <rosmaita> ok, last announcement:
14:08:34 <rosmaita> cinder yoga midcycle coming up in 2 weeks on Wednesday 1 December 2021
14:08:42 <rosmaita> #link https://etherpad.opendev.org/p/cinder-yoga-midcycles
14:08:48 <rosmaita> please add topics to ^^
14:09:00 <rosmaita> if you are working on a spec, it would be a good place to get some feedback
14:09:29 <rosmaita> or if you have an idea for a feature, or a bugfix that requires a new config option or something, also a good place to get feedback
14:09:50 <jungleboyj> Are we thinking the usual time?
14:10:14 <rosmaita> was just going to ask that ... actually, first, is that day good (or at least not awful) for everyone?
14:10:24 <rosmaita> (somehow I seem to remember someone saying there might be a conflict)
14:11:07 <jungleboyj> The day should be fine for me.
14:11:29 <rosmaita> ok, hearing no negative responses, we will stick with 1 december
14:11:37 <rosmaita> OK, next question is the time
14:11:42 <rosmaita> I'd like to do our usual thing where we use 1 hour from the weekly meeting time
14:11:49 <rosmaita> (since everyone has that blocked off already)
14:11:55 <jungleboyj> ++
14:11:56 <rosmaita> and add an hour either before or after the meeting
14:12:04 <rosmaita> let's try to pick a time right now
14:12:15 <rosmaita> options are:
14:12:15 <rosmaita> A 1300-1500 UTC
14:12:15 <rosmaita> B 1400-1600 UTC
14:12:15 <rosmaita> C either one of the above
14:12:15 <rosmaita> D need more options
14:12:27 <rosmaita> #startvote time for the R-17 (1 December) midcycle? A, B, C, D
14:12:27 <opendevmeet> Begin voting on: time for the R-17 (1 December) midcycle? Valid vote options are A, B, C, D.
14:12:28 <opendevmeet> Vote using '#vote OPTION'. Only your last vote counts.
14:12:43 <rosmaita> (everybody clear about how this works?)
14:12:58 <jungleboyj> #vote B
14:13:02 <eharney> #vote C
14:13:04 <rosmaita> #vote B
14:13:07 <walshh_> #vote B
14:13:29 <fabiooliveira> #vote C
14:13:38 <ssuerd> #vote B
14:14:57 <rosmaita> one more minute to vote!
14:15:01 <geguileo> #vote C
14:15:20 <lucasmoliveira059> #vote C
14:16:09 <rosmaita> ok, let's see what the results are
14:16:14 <rosmaita> #endvote
14:16:14 <opendevmeet> Voted on "time for the R-17 (1 December) midcycle?" Results are
14:16:14 <opendevmeet> B (4): jungleboyj, walshh_, rosmaita, ssuerd
14:16:14 <opendevmeet> C (4): lucasmoliveira059, fabiooliveira, geguileo, eharney
14:16:33 <rosmaita> OK, looks like 1400-1600 UTC it is
14:17:04 <rosmaita> by the way, if you are constrained to 1400-1500, put a note on the midcycle etherpad when you add your topic
14:17:15 <rosmaita> and we can make sure we discuss it during that time frame
14:17:29 <rosmaita> ok, thanks everyone ... on with the meeting
14:17:43 <rosmaita> #topic Ceph: add option to keep only last n snapshots per backup
14:17:52 <rosmaita> #link https://review.opendev.org/c/openstack/cinder/+/810457
14:17:57 <rosmaita> jhartkopf: that's you
14:18:32 <jhartkopf> yes
14:18:58 <jhartkopf> so basically we'd like be able to only keep a specific number of backup snapshots of a volume
14:19:25 <rosmaita> for the RBD driver
14:19:47 <rosmaita> well, ceph backup driver, i mean
14:19:53 <jhartkopf> so this patch adds a new config option to configure how many snapshots to keep
14:20:02 <jhartkopf> exactly, for the Ceph driver
14:20:49 <rosmaita> seems like a good idea, but i asked jhartkopf to put it on the agenda because it requires a new configuration option
14:21:03 <geguileo> I disagree, I don't like this
14:21:21 <rosmaita> please elaborate
14:21:24 <geguileo> that's my first impression, looking more on the patch
14:21:37 <geguileo> rosmaita: the reason for having backup snapshots is to be able to do incremental backups
14:22:06 <geguileo> if we remove the snapshot without removing the related backup then we can reach a point were we can't do an incremental backup and end up doing a full backup
14:23:01 <rosmaita> that is a good point
14:23:23 <geguileo> literally what the referenced bug was fixing
14:23:27 <geguileo> https://bugs.launchpad.net/cinder/+bug/1703011
14:24:26 <jhartkopf> as far as I understand, we still can do incremental backups if both snapshots (the same ones) exist on both sides
14:25:10 <geguileo> but if you remove it due to the new config option then we can't
14:25:39 <geguileo> we have a 1-to-1 relationship between cinder backups and RBD snapshots
14:25:43 <rosmaita> that's quite a commit message on https://review.openstack.org/486678
14:25:45 <geguileo> that's how we can do incremental
14:26:01 <geguileo> if we break that 1-to-1 relationship we break incremental backups in some cases
14:26:54 <jhartkopf> but if we only remove older snapshots, we keep the more recent ones which can then be used to do incremental backups
14:27:21 <geguileo> let me show you the issue with an exageration
14:27:27 <geguileo> set the new conf option to 2
14:27:32 <geguileo> do 4 incremental backups
14:27:41 <geguileo> the oldest 2 snapshots are removed
14:27:45 <geguileo> now you delete the last 2 backups
14:27:50 <geguileo> now you can't do incremental backups
14:28:06 <geguileo> because the new one will be a full backup
14:28:20 <geguileo> (the removal of the snapshots are done by the new feature)
14:30:15 <rosmaita> so it looks like if the motivation for this is to save space on the backend, this approach is counterproductive
14:30:25 <jhartkopf> yes that seems like a situation to consider
14:30:39 <geguileo> and that was the situation that the related bug was fixing
14:31:07 <rosmaita> jhartkopf: why don't you take some time to think about this, and we could discuss again at the midcycle if you have some new ideas
14:31:18 <geguileo> jhartkopf: what was the original reason for the feature?
14:31:28 <geguileo> performance? save space?
14:32:06 <jhartkopf> mainly to save disk space because keeping all snapshots can quickly take up much space
14:32:33 <geguileo> jhartkopf: I assume you don't want to delete the backups that match those snapshots
14:34:05 <jhartkopf> yes we'd like to keep the backups
14:35:24 <geguileo> I don't like the idea of removing the source snapshots... Though I understand your issue
14:36:24 <rosmaita> ok, let's think about this some more and pick up the discussion, maybe at the midcycle, and work through the issues
14:36:34 <rosmaita> thanks geguileo and jhartkopf
14:36:50 <rosmaita> #topic Implications of new Glance spec-lite on Cinder image encryption spec rewrite
14:37:02 <fungi> i'll make this quick, it's a national holiday for the secustack folks, but during the image encryption pop-up team meeting on monday, Luzi mentioned wanting to inquire in the cinder meeting whether anyone's had time to mull over the implications of the glance "spec-lite" on the former cinder spec, and whether a new spec would be required for the cinder side of things or if changes
14:37:02 <rosmaita> fungi: you have the floor
14:37:04 <fungi> might be needed to the old cinder spec for it
14:37:20 <fungi> looks like jhartkopf had some feedback on the agenda etherpad already
14:37:36 <rosmaita> actually, that was me
14:37:44 <fungi> ahh, okay
14:37:56 <rosmaita> so the change in Glance is to not wait for the Barbican Secret Consumers API to be implemented
14:38:03 <fungi> right
14:38:04 <rosmaita> but go ahead without it, and add it later
14:38:13 <fungi> that's my understanding, yes
14:38:26 <rosmaita> i believe that cinder wasn't in the workflow for using the Secret Consumers API
14:38:31 <rosmaita> so that doesn't affect us
14:39:13 <rosmaita> i think there may be some issues that are going to come up with how to handle the optimizations when glance is using the cinder glance_store
14:39:23 <fungi> so the old cinder spec is still applicable, or are you saying there's no need to any work in cinder to support it?
14:39:24 <rosmaita> and also the cinder image cache
14:39:47 <fungi> er, no need to do any
14:40:16 <rosmaita> i guess what i'm saying is that some stuff will have to be worked out during the implementation of the cinder side
14:40:35 <rosmaita> but i don't think we need a new spec for that
14:40:51 <fungi> so doesn't rise to the level of complexity to require a spec. got it
14:40:54 <fungi> thanks!
14:41:14 <rosmaita> yeah, i am in favor of the glance spec-lite because i think it will move the effort along
14:41:18 <fungi> i mainly wanted to make sure the folks working on this weren't blocked for a week waiting for another cinder meeting
14:41:29 <fungi> much appreciated
14:41:36 <rosmaita> once there is end-to-end testing of the feature, i imagine a bunch of issues will areise
14:41:39 <rosmaita> *arise
14:41:45 <fungi> sure, as ever
14:41:52 <fungi> "that's software"
14:41:57 <rosmaita> yes, so on the cinder side, the first priority is to get the gpg stuff into os-brick
14:42:51 <rosmaita> #link https://review.opendev.org/709432
14:43:35 <fungi> yeah, looks like it needs rebasing at the very least
14:43:36 <rosmaita> thanks fungi ... anyone else have comments or concerns about this?
14:44:28 <fungi> okay, cool. that answers the questions i think. thanks again!
14:44:29 <rosmaita> fungi: yes, if Luzi could get it out of merge conflict, that would be a good motivator for reviewers
14:44:44 <fungi> agreed
14:45:28 <rosmaita> ok, thanks ... let's move on
14:45:35 <rosmaita> #topic mypy patch of the week
14:46:03 <rosmaita> last week's patch was this one
14:46:12 <rosmaita> #link https://review.opendev.org/c/openstack/cinder/+/733622
14:46:32 <rosmaita> we're still working on it ... it's very close, i think
14:46:47 <rosmaita> so that will remain the mypy patch of the week for this week
14:47:08 <eharney> yep, i'll update that again shortly
14:47:16 <rosmaita> and we can carry over the same reviewers
14:47:39 <rosmaita> also, thanks to walshh_, who did a thorough review
14:48:06 <walshh_> you are welcome
14:49:07 <rosmaita> #topic volume driver API function of the week
14:49:16 <rosmaita> i think we are stalled on this one
14:49:37 <rosmaita> still need a solidfire dev to leave an OK on https://review.opendev.org/c/openstack/cinder/+/814898
14:50:32 <rosmaita> and that blocks the next patch, that needs a second +2
14:50:40 <rosmaita> #link https://review.opendev.org/c/openstack/cinder/+/815659
14:51:47 <rosmaita> i had a nit on the next patch, but iirc it's a sphinx formatting thing, not content
14:51:56 <rosmaita> #link https://review.opendev.org/c/openstack/cinder/+/815660/3
14:52:06 <rosmaita> so don't let that stop you from looking at it
14:52:47 <rosmaita> so it looks like we remain the same for volume driver API function of the week
14:53:04 <geguileo> rosmaita: I'll update the patches and start working on a new patch
14:53:22 <rosmaita> geguileo: thanks!
14:53:24 <geguileo> I've been busy on the new quota stuff, which has turned to be a bit more complex than I had anticipated
14:53:31 <geguileo> (but making progress)
14:53:44 <rosmaita> that is understandable!
14:54:10 <rosmaita> #topic open discussion
14:54:10 <fabiooliveira> > still need a solidfire dev to leave an OK on https://review.opendev.org/c/openstack/cinder/+/814898
14:54:10 <fabiooliveira> i'll take a look :D
14:54:11 <geguileo> I should have the spec and patches for the midcycle
14:54:17 <geguileo> fabiooliveira: thanks!
14:54:27 <rosmaita> fabiooliveira: thank you
14:55:58 <rosmaita> I just looked at the calendar, next week is the last meeting of the month, so will be held in video
14:56:16 <rosmaita> but, thursday and friday next week are holidays in some parts of the world
14:56:40 <rosmaita> just wondering if we will have a critical mass of attendance for wednesday's meeting?
14:57:30 <rosmaita> anyone who will not be attending?
14:58:15 <rosmaita> ok, sounds like everyone will be around ... great!
14:58:31 <rosmaita> don't forget the Festival of XS Reviews on Friday this week
14:59:40 <rosmaita> it has gotten awfully quiet in here ... thanks for attending, and have a good rest of the day
14:59:44 <rosmaita> see you on friday!
14:59:59 <rosmaita> #endmeeting