14:01:15 <abhishekk> #startmeeting glance
14:01:16 <openstack> Meeting started Thu Jul  9 14:01:15 2020 UTC and is due to finish in 60 minutes.  The chair is abhishekk. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:01:17 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:01:19 <openstack> The meeting name has been set to 'glance'
14:01:20 <abhishekk> #topic roll call
14:01:24 <jokke> o/
14:01:29 <abhishekk> #link https://etherpad.openstack.org/p/glance-team-meeting-agenda
14:01:31 <smcginnis> o/
14:01:31 <abhishekk> o/
14:01:44 <abhishekk> welcome all :D
14:01:51 <abhishekk> short agenda
14:01:58 <abhishekk> lets start
14:02:16 <abhishekk> #topic release/periodic jobs update
14:02:27 <abhishekk> V2 is 2 weeks away
14:02:50 <abhishekk> We need to review all the specs for Victoria on priority basis
14:03:26 <abhishekk> Image encryption is merged, sparse image upload and dansmith's policy specs are in good shape
14:03:47 <dansmith> yes please :)
14:03:59 <abhishekk> We need to focus on Duplicate downloads, cinder multiple stores support and remaining specs in order
14:04:59 <abhishekk> Regarding periodic job, 4 tips job of py36 unit tests failed yesterday for glance_store, fix is already merged, green now
14:05:08 <abhishekk> So all good at the moment
14:05:25 <abhishekk> moving ahead
14:05:43 <abhishekk> #topic Should we change the spec approval policy?
14:05:57 <abhishekk> I want your opinion on this
14:06:40 <abhishekk> Earlier we started this because we have enough of contributors to take care of work
14:07:00 <abhishekk> situation is different for last 2-3 cycles
14:07:52 <rosmaita> the way it is now, only the PTL can merge a spec ... i think leave it up to the PTL how many +2s to require
14:07:57 <abhishekk> everyone is busy with enough work in hand which delaying the review process which might result in loosing interest of new contributors
14:08:12 <rosmaita> i mean, on a per-spec basis
14:08:20 <smcginnis> Yeah, I think it's probably good to not require every core's +2/
14:08:20 <rosmaita> there may be uncontroversial specs
14:08:34 <rosmaita> but also some where you want everyone to have looked at them
14:08:55 <abhishekk> yes, in that case I will personally ping all the members to have a look
14:09:01 <dansmith> we do that in nova, controversial specs require lots of buy-in
14:09:20 <abhishekk> such as unified limits, duplicate downloads etc
14:09:30 <dansmith> my spec has had four cores vote +2 at various times across minimal changes, but not all at once
14:09:52 <dansmith> getting them all to +2 at the same time seem like it might be difficult
14:10:08 <abhishekk> correct
14:10:33 <rosmaita> i think it makes sense to be flexible about this
14:11:04 <jokke> I think we've all been there. I'm fine leaving it to the PTL, just want to make sure rosmaita & smcginnis don't feel left out. It was indeed bit different store when I introduced the requirement when we were all more or less fulltime glance
14:11:27 <jokke> or full time openstack at least with more time for glance ;)
14:11:48 <abhishekk> So IMO we should come back to traditional way of having two +2's for simple/straight forward specs and more than 2 for some controversial specs
14:11:50 <jokke> s/store/story/
14:12:24 <abhishekk> yes no offense here, I know everyone is giving their best here
14:12:39 <smcginnis> abhishekk: That sounds like a good plan to me.
14:12:47 <abhishekk> smcginnis, thank you
14:13:07 <abhishekk> rosmaita, jokke ??
14:13:18 <rosmaita> sounds good to me
14:13:29 <abhishekk> Cool
14:14:26 <abhishekk> Moving ahead
14:14:41 <jokke> abhishekk: yes I'm good with the plan. Maybe make sure the spec exists at least over a weekly meeting so everyone interested will have heads up and time to review. So how about no 1day specs where quick read and merge?
14:15:15 <abhishekk> jokke, sounds good to me
14:15:22 <smcginnis> jokke: ++
14:15:51 <abhishekk> I will make sure to highlight important specs and status in our weekly meeting (Starting from next)
14:16:08 <jokke> abhishekk: sounds great.
14:16:28 <abhishekk> ok, moving to open discussion
14:16:35 <abhishekk> #topic Open discussion
14:17:14 <abhishekk> As discussed in PTG, I have noted down TODO's list in glance
14:17:24 <jokke> #link https://review.opendev.org/#/c/738675/
14:17:32 <abhishekk> #link https://docs.google.com/spreadsheets/d/1akuBhqmJfAC13Oi8-PjVZCwpX4KjotworWU_2w03QZA/edit?usp=sharing
14:17:45 <abhishekk> I need to provide access though
14:17:46 <jokke> smcginnis: rosmaita: ^^ we would need you two as that was collab between myself and abhishekk
14:18:05 <jokke> I just updated the commit message as I had bug open for it, just forgot the tag
14:18:27 <abhishekk> I will try to copy this spreadsheet using my personal mail id and will open for all
14:18:27 <rosmaita> ok, i can look at that today
14:18:36 <abhishekk> Will add TODO's from store and client also
14:18:55 <dansmith> jokke: rosmaita smcginnis: yeah, please, I need that to be merged before my full stack of 9000 dependencies will run :D
14:19:08 <abhishekk> We will discuss more about this in our next meeting
14:19:25 <smcginnis> Ah, just bug link added. I could have sworn I already reviewed that.
14:19:42 <jokke> abhishekk: good job, will you just open the doc or do you want us to request access from that link?
14:19:54 <jokke> smcginnis: you did indeed
14:20:05 <jokke> smcginnis: and yes, just the missing bug link changed
14:20:14 <abhishekk> jokke, I am going to open it for all
14:20:25 <jokke> abhishekk: ++
14:21:02 <abhishekk> we have plenty of time, so I would like to finalize how we should proceed on race condition thing
14:21:10 <abhishekk> #link Fix race condition in copy image operation
14:21:19 <abhishekk> #link https://review.opendev.org/737596
14:22:39 <abhishekk> Unique exception comment is easy to fix though, problem is where do we fix the race condition :D
14:23:05 <dansmith> yeah, I feel pretty strongly that it's important for the API to be consistent here,
14:23:42 <dansmith> because when I was trying to write the client side of this, it's difficult to detect the difference between "things are slow" and "the API told me I should wait a long time for a thing to be done, because I got my 20x"
14:24:35 <dansmith> obviously fixing the race is the important part for correctness, but multi-boot of lots of instances at a time *will* hit the situation where two novas ask and don't realize that one request is being silently ignored
14:25:50 <abhishekk> other way to deal with this is to add revert to each task which might fail and cleanup those properties there
14:26:19 <jokke> abhishekk: I will be very generous with my -2 for that :P
14:26:25 <abhishekk> :D
14:27:55 <abhishekk> smcginnis, rosmaita could you please also have a look at this patch?
14:28:33 <smcginnis> Will do.
14:28:39 <rosmaita> ok, will put it on my list
14:28:42 <abhishekk> smcginnis, thank you
14:29:14 <rosmaita> abhishekk: you mean https://review.opendev.org/737596 ?
14:29:20 <abhishekk> rosmaita, yes
14:29:22 <jokke> So smcginnis rosmaita just for context. The problem is that the copy-image has race, currently we're just in swamp with it. abhishekk & dansmith want to put locking in place, which I'm not fond of but it looks like one of the best options to tackle this.
14:29:41 <jokke> The question is when and how do we trigger that lock and clean it
14:29:50 <rosmaita> ok
14:30:22 <abhishekk> You will also get the clear picture when you will go through the comments
14:30:39 <smcginnis> OK, thanks. That helps.
14:30:56 <abhishekk> Cool that's it for today
14:31:26 <rosmaita> quick question: how are we looking on prospective new glance cores?
14:32:30 <abhishekk> rosmaita, that's difficult to say as we don't have much new reviewers at the moment
14:32:45 <rosmaita> i was afraid you'd say that :(
14:32:48 <abhishekk> I am trying to connect some guys
14:33:18 <jokke> or even old reviewers outside of this group
14:33:55 <abhishekk> will ask them if they are interested for the same
14:33:57 <jokke> rosmaita: feeling like getting too busy to keep up with glance anymore?
14:34:07 <rosmaita> i am right at this minute!
14:34:31 <abhishekk> lets revisit this again two weeks from now
14:34:36 <jokke> ouch ... please try to avoid burning yourself out
14:35:32 <abhishekk> rosmaita, I would suggest to stay onboard at least for this cycle
14:35:50 <rosmaita> i can do that
14:35:50 <abhishekk> I will ping you for important patches only
14:36:00 <rosmaita> just want to see light at the end of the tunnel
14:36:01 <jokke> rosmaita: we try to keep your load as minimum as we can
14:36:25 <abhishekk> I will stand holding torch there :d
14:36:46 <abhishekk> jokke, ++
14:36:57 <abhishekk> Just FYI, I will not be around tomorrow, my little sister is getting engaged this weekend
14:37:02 <jokke> I'll keep abhishekk supplied with fresh batteries for his torch :D
14:37:10 <rosmaita> :)
14:37:11 <abhishekk> lol
14:37:27 <abhishekk> most probably not able to check mails as well
14:37:48 <jokke> abhishekk: enjoy and relay our best wisher for her
14:37:54 <abhishekk> thank you
14:37:55 <jokke> wishes
14:38:09 <abhishekk> could you please restore your +2 on Dan's spec
14:38:53 <abhishekk> any thing else guys?
14:39:26 <jokke> not from me, I'll reread that spec
14:39:39 <abhishekk> ack
14:39:45 <abhishekk> thank you all
14:39:50 <abhishekk> have a nice weekend
14:40:28 <jokke> thanks all
14:40:45 <abhishekk> #endmeeting