14:00:31 <flaper87> #startmeeting Glance
14:00:32 <openstack> Meeting started Thu Feb 18 14:00:31 2016 UTC and is due to finish in 60 minutes.  The chair is flaper87. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:34 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:00:35 <flaper87> o/
14:00:37 <openstack> The meeting name has been set to 'glance'
14:00:41 <kragniz> o/
14:00:51 <abhishekk> o/
14:00:54 <rsjethani> o/
14:00:59 <dshakhray> o/
14:01:18 <rosmaita> o/
14:01:22 <nikhil> o/
14:01:26 <ninag> o/
14:01:35 <kairat_> o/
14:01:42 <flaper87> #topic Agenda
14:01:42 <jokke_> o/
14:01:44 <flaper87> #link https://etherpad.openstack.org/p/glance-team-meeting-agenda
14:01:48 <flaper87> That's our agenda for today
14:01:56 <sabari> o/
14:01:57 <flaper87> #topic Updates from Glare
14:02:03 <flaper87> mfedosin: kairat_ any updates ?
14:02:16 <flaper87> How's that spec coming along ?
14:02:19 <mfedosin> o/
14:02:30 <mfedosin> hey
14:02:38 <mfedosin> it's quite good
14:02:55 <mfedosin> we spent last days discussing and documenting our ideas
14:03:09 <mfedosin> so plan is still the same
14:03:19 <mfedosin> spec will be published on Monday
14:03:32 <mfedosin> 22nd of February
14:03:36 <flaper87> coolio
14:03:38 <flaper87> sounds good
14:03:42 <flaper87> Anything else?
14:04:03 <mfedosin> thanks nikhil for useful advice
14:04:09 <flaper87> ++
14:04:17 <nikhil> :)
14:04:26 <mfedosin> that's all I suppose
14:04:32 <flaper87> thanks for the updates
14:04:32 <kairat_> nikhil, we refined use case description
14:04:38 <kairat_> added statuses transition
14:04:40 <flaper87> #topic Updates Nova v1->v2
14:04:45 <nikhil> cool will look again
14:05:01 <mfedosin> nikhil: there are a lot of new things
14:05:02 <flaper87> #link http://lists.openstack.org/pipermail/openstack-dev/2016-February/086424.html
14:05:09 <nikhil> ack
14:05:10 <flaper87> So, mfedosin sent that email out
14:05:13 <flaper87> not much has happened
14:05:24 <flaper87> no response from the nova team, AFAICT
14:05:50 <mfedosin> jaypipes promised to send his response
14:05:51 <bunting> o/
14:05:53 <mfedosin> but...
14:06:19 <flaper87> I wouldn't get our hopes too high on this happening in Mitaka
14:06:36 <mfedosin> okay, let's do it in Newton
14:06:38 <flaper87> That said, as I mentioned in my reply, I do think we should move forward with the deprecation of v1
14:06:41 <mfedosin> but DO it
14:07:00 <flaper87> I'll work on a better and more formal proposal for that
14:07:02 <nikhil> well how do we deprecate it
14:07:06 <nikhil> if it's needed
14:07:09 <flaper87> All that to happen in newton, though
14:07:11 <flaper87> not now
14:07:21 <nikhil> k
14:07:21 <flaper87> nikhil: need to think this through a bit better, TBH.
14:07:26 <nikhil> surely
14:07:47 <flaper87> that's it, I guess
14:07:58 <flaper87> #topic Updates x-proj meeting
14:08:06 <nikhil> o/
14:08:15 <nikhil> There's just one
14:08:27 <nikhil> ommon policy scenario across all projects
14:08:34 <nikhil> Common*
14:08:42 <nikhil> #link https://review.openstack.org/#/c/245629
14:09:01 <nikhil> The proposal is merely to elaborate the roles in defaults shipped with the source
14:09:21 <flaper87> ++
14:09:24 <nikhil> Currently, that's not the case with most prjs. it's either admin or user
14:09:43 <nikhil> I think Glance has some good usecases where we can benefit the community with different types of roles
14:09:58 <flaper87> could you elaborate?
14:10:03 <nikhil> viz. image (protected) properties, metadefs
14:10:24 <nikhil> We allow pretty fine grained access to individual elements
14:10:49 <nikhil> I think the work from sigmavirus24_awa in Liberty for using oslo.policy in full potential (haven
14:10:57 <nikhil> that I haven't tested*
14:11:14 <nikhil> should give us some narrow/corner case scenarios to think about
14:11:27 <flaper87> gotcha
14:11:31 <flaper87> that sounds interesting indeed
14:11:34 <nikhil> Plus we also have the discussion going with admin properties used/req by Nova
14:11:50 <nikhil> So, I encourage individual members to weigh in on that spec
14:12:08 <mclaren> Do you have a spec link nikhil?
14:12:12 <nikhil> I'm sure there are a lot amongst us who are working on specific cases more than others
14:12:15 <flaper87> #link https://review.openstack.org/#/c/245629/
14:12:16 <flaper87> mclaren: ^
14:12:20 <mclaren> gracias
14:12:23 <flaper87> de nada
14:12:46 <nikhil> (I posted the link above, hope all my msgs are going through)
14:12:54 <rosmaita> searchlight has some glance policy/prop protection config files they use for testing, i think they have several roles
14:13:00 <flaper87> nikhil: they are
14:13:09 <nikhil> flaper87: ty for ack
14:13:15 <nikhil> rosmaita: right you are!
14:13:16 <mclaren> oops I mean the link for the nova/glance properties
14:13:25 <nikhil> um
14:13:59 * flaper87 doesn't have it handy
14:14:07 <flaper87> It's a nova spec
14:14:15 <nikhil> me neither
14:14:18 <mclaren> nikhil: fwiw Niall put this in https://review.openstack.org/#/c/249950/
14:14:39 <nikhil> mclaren: ah interesting, will look
14:14:41 <bunting> Yup
14:14:50 <flaper87> nikhil: anything else?
14:14:53 <nikhil> will help with snaps
14:15:00 <nikhil> one small thing
14:15:05 <flaper87> shoot
14:15:30 <nikhil> Nova folks are very interested in the quotas work, so I think we should get good feedback prior to summit
14:15:57 <flaper87> nice! we've a topic for that coming up next
14:16:07 <nikhil> I forgot my other words :(
14:16:18 <flaper87> :D
14:16:20 <flaper87> :(
14:16:28 <nikhil> ok, I will update in the dedicated topic
14:16:34 <flaper87> ok, moving to the next topic
14:16:34 <nikhil> please move on for now
14:16:37 <nikhil> ty
14:16:43 <flaper87> #topic Quotas
14:16:52 <nikhil> :D
14:16:53 <flaper87> ninag nikhil
14:17:00 <flaper87> :D
14:17:28 <nikhil> oh yes, so the guy who has experience and wanted to write the spec is from yahoo
14:17:53 <nikhil> he is still interested in the work but due to recent stir in the valley, there has been delay in getting the spec up
14:18:11 <nikhil> I was told that first draft would be proposed later this week
14:18:28 <nikhil> (end of x-prj update)
14:18:35 <rosmaita> nikhil: this is a x-pj quotas spec?
14:18:35 <nikhil> For glance quotas
14:18:42 <nikhil> rosmaita: yeah
14:18:45 <nikhil> For glance quotas
14:18:46 <flaper87> rosmaita: you stole my question
14:18:48 <flaper87> :D
14:19:02 <jokke_> I don't think that being so big of an issue if the deprecation of v1 got pushed to Newton/O timeframe
14:19:07 <rosmaita> (see ... i am paying attention)
14:19:20 <nikhil> I had a discussion with rosmaita and ninag to bridge the gap between what goes in simple, nested and what are challenged in glance
14:19:41 <nikhil> jokke_: woohoo
14:19:56 <jokke_> so we have plenty of time to plan the quotas
14:19:59 <mclaren> which part of the quotas work is x-proj?
14:20:27 <nikhil> mclaren: xprj will be nested work
14:20:33 <nikhil> actually that's the plan
14:20:45 <nikhil> but I think we will know the practicalities as things evolve
14:20:59 <mclaren> ok sounds good
14:21:01 <nikhil> so, I guess the proposal(s) is(are) pretty thorough wrt use cases and API description
14:21:58 <nikhil> I was raised some concerns on how things will work for entities that are not storage related.
14:22:21 <jokke_> nikhil: like?
14:22:25 <nikhil> I guess we can have some planning of how to span out the work and if it will or will not affect
14:23:01 <nikhil> jokke_: right, so the concerns are premature atm. we need some (virtual)whiteboarding to figure out if the concerns stand tall.
14:23:19 <nikhil> image-members for the most part
14:23:23 <flaper87> Planning Newton will be interesting. Basically, most of the things that were planned for mitaka got pushed to Newton. Image import refactor and nova v1 - v2
14:23:32 <flaper87> Scheduling those 2 with quotas will be.... fun
14:23:55 <nikhil> I am just curious if we can work out a plan to get them done in n-1
14:24:05 <flaper87> nikhil: quotas?
14:24:07 <nikhil> and focus on improvements for later phases?
14:24:11 <jokke_> flaper87: easy ... we start planning/discussing quotas now and implement in O where we hopefully have v1 finally deprecated
14:24:38 <nikhil> flaper87: n-1 = nova v1->v2 and import
14:24:46 <rosmaita> nikhil: +1
14:24:47 <jokke_> and the above points implemented
14:24:59 <flaper87> nikhil: That would be ideal
14:25:07 <nikhil> so, this is a really really important work we need in N
14:25:17 <flaper87> jokke_: We'll see, definitely not something we need to discuss right now.
14:25:25 <nikhil> and I think the planning (even work) has been started already
14:25:38 <flaper87> actually, sorry for mentioning that, I didn't mean to hijack the conversation
14:25:58 <nikhil> I would like to discuss the possibility in more detail.
14:26:39 <rosmaita> man, i love working in open source
14:26:47 <nikhil> So, my point of getting quotas right starting from Mitaka was for the same reason
14:26:49 <jokke_> rosmaita: ;)
14:27:16 <flaper87> I think we need to freeze the image import refactor spec and then we can start scheduling everything
14:27:16 <nikhil> we discuss all the edge cases here and get the xprj feedback early so that we don't have surprises
14:27:27 <flaper87> nikhil: right
14:27:39 <flaper87> nikhil: so, I take we'll be going with your and ninag proposal, right?
14:27:43 <flaper87> Not flwang's
14:28:19 <nikhil> flaper87: so, we would love to describe how they all work so that it's clear that fei long's proposal is not discredited with ninag's
14:28:37 <flaper87> absolutely
14:29:18 <nikhil> if we need a more bandwidth conversations with those who have concerns, it would be totally possible
14:29:48 <flaper87> Ok, I think the next step is for folks to read the specs and start commenting there
14:29:54 <nikhil> ++
14:30:03 <flaper87> That way we can start raising questions and have more specific conversations
14:30:26 <flaper87> cool, should we move on?
14:30:32 <nikhil> that would be super helpful
14:30:39 <nikhil> ninag:  any comments?
14:30:46 <ninag> nope, am good
14:30:52 <nikhil> thanks
14:30:54 <flaper87> #topic Remove env tools
14:30:59 <flaper87> #link https://review.openstack.org/#/c/269414/
14:31:16 <flaper87> So, we still have `run_tests` and some scripts under `tools` to create venvs
14:31:24 <flaper87> Is anyone using those?
14:31:34 <flaper87> I'm certainly not and the gate isn't either
14:31:48 <flaper87> I'd like to remove them. Last time we tried, some folks were using them. Trying again now
14:32:14 <kairat_> are these tools x-proj?
14:32:27 <flaper87> They used to be in oslo-incubator
14:32:33 <flaper87> oslo0incubator is now gone
14:32:41 <flaper87> and several projects have removed them
14:33:08 <bunting> https://bugs.launchpad.net/glance/+bug/1546664 says they are broken
14:33:08 <openstack> Launchpad bug 1546664 in Glance "Unable to run tox tests, install_command error" [Undecided,New]
14:33:15 <bunting> So does that mean that nobody is using them?
14:33:16 <flaper87> I don't think we need to make this a x-prj effort. It's my opinion that tox does everything we need to do and that can be done with run_tests
14:33:33 <flaper87> bunting: open source, one never knows :P
14:33:45 <kragniz> flaper87: +1 on removing them
14:33:58 <flaper87> Unless someone screams, I'll send a patch to remove them
14:34:08 <kairat_> +1, wondering about e-mail to others
14:34:20 <flaper87> sure, emails and everything will go out
14:34:49 <flaper87> ok, that's it on this topic
14:34:59 <jokke_> make a lite spec, we discuss about it few months and we are safe to disagree ;)
14:35:12 <kairat_> heh
14:35:24 <flaper87> jokke_: speaking of lite specs, weren't you going to propose a plan or something? :P
14:35:27 * flaper87 ducks
14:35:41 <flaper87> #topic Return request id to caller (rsjethani)
14:35:47 <flaper87> rsjethani: around?
14:35:53 <rsjethani> yup
14:36:00 <rsjethani> I think the work on return-request id to caller is in good shape.
14:36:08 <rsjethani> https://review.openstack.org/#/c/261288
14:36:16 <jokke_> flaper87: on-off on it when ever I have energy hopefully I have the proposalt together soon enough
14:36:38 <rsjethani> It seems a good candidate to be merged in mitaka time frame
14:36:48 <flaper87> rsjethani: sounds good!
14:37:03 <mclaren> I'll try to take a look
14:37:07 <kairat_> just my 5 cents, we have a cross-proj spec about that
14:37:16 <kairat_> and unfortunately we don't follow it
14:37:21 <kairat_> because of our validation
14:37:23 <flaper87> rsjethani: for future cases like this. We always try to keep review requests for Open discussions.
14:37:41 <flaper87> rsjethani: nothing bad with your request, though.
14:37:51 <rsjethani> flaper87: ok
14:37:53 <flaper87> kairat_: does that patch fix such inconsistency?
14:38:03 <flaper87> I haven't looked into it with enough details
14:38:14 <kairat_> Nope, we still doing request id in "glance" way
14:38:29 <kairat_> I don't see anything bad with it
14:38:55 <kairat_> but looks like our validation leads to troubles from time to time
14:38:56 <flaper87> kairat_: ok, something to fix. It'd be great to have a bug opened for that so we can discuss this further
14:39:16 <kairat_> I better describe it in review
14:39:21 <flaper87> ok
14:39:33 <jokke_> flaper87: the mentioned was discussed when the req-id cross project work was done
14:39:46 <jokke_> flaper87: unifying us basically breaks our API
14:39:56 <flaper87> jokke_: ahhh now I remember
14:40:01 <flaper87> thanks for the heads up
14:40:22 <flaper87> ok, anything else?
14:40:32 <mclaren> interestingly there was a mail suggesting the other projects do it our way...but too late
14:40:36 <rsjethani> nope. thats it. Ty all
14:40:45 <flaper87> #topic Heirarchial Image access ( nikhil )
14:40:48 <nikhil> o/
14:41:08 <nikhil> I got a WIP spec for this https://review.openstack.org/#/c/281599/ (
14:41:23 <nikhil> please take a look at the tasks
14:41:36 <nikhil> that should explain the amount of work is pretty small
14:42:04 <nikhil> I thought the only reason this sort of work would need a full spec is for API change in the schema
14:42:42 <nikhil> otherwise, it's straightforward approach
14:42:44 <flaper87> fair enough. I'd agree changes on the schema requires a full spec
14:42:51 <flaper87> coolio, thanks for putting that up
14:43:15 <nikhil> np
14:43:15 <mclaren> nikhil any chance of an example problem in the spec?
14:43:42 <nikhil> mclaren: sure, I will put that up. working on the formal description of the problem statement and one corner case
14:43:48 <mclaren> what's there is a bit too abstract for me :-)
14:43:54 <mclaren> thanks
14:44:06 <nikhil> I can understand
14:44:21 <nikhil> that's it from me
14:44:31 <flaper87> oki
14:44:38 <flaper87> #topic Open Discussion
14:44:47 <flaper87> We've plenty of time for open discussions today
14:44:54 <flaper87> FYI, glance_store 0.11.0 is out
14:45:02 <flaper87> Please, keep an eye on gates and possible failures
14:45:21 <jokke_> I'd like to make cry out now
14:45:28 <flaper87> shoot
14:45:40 <jokke_> Please people start filling the reno files with your commits
14:46:13 <jokke_> I realized that we had next to none on that 0.11.0 release and they will be required for the releases so we get release notes out
14:46:38 <flaper87> ++
14:46:41 <flaper87> Yup
14:46:46 <nikhil> wait, I thought the release notes were pretty thorough
14:46:46 <flaper87> and reviewers, keep an eye on that
14:46:53 <jokke_> I think review wise now is good time to start -1 things that does not have release notes included
14:46:54 <kairat_> Ok
14:47:00 <nikhil> umm, may be not!?
14:47:09 <flaper87> everything that requires upgrade changes, deprecations, new features, etc is worth a release note
14:47:30 <jokke_> + any user impacting bugfixes
14:48:37 <flaper87> ok, anything else?
14:48:45 <flaper87> otherwise I'll close the meeting earlier
14:48:55 <kairat_> flaper87, open discussion
14:48:55 <rosmaita> hang on
14:49:03 <rosmaita> got something
14:49:06 <flaper87> kairat_: we are in open discussion
14:49:09 <rosmaita> kairat_: you go first
14:49:17 <kairat_> ah, ok
14:49:22 <kairat_> guys please review https://review.openstack.org/#/c/251851/
14:49:32 <kairat_> I would like to finish work with trusts in Mitaka
14:49:46 <kairat_> But doesn't have any valuable reviews
14:49:54 <kairat_> So if you have some time
14:50:00 <flaper87> ah that reminds me! Yes, ppl, please, provide reviews on features
14:50:05 <kairat_> it would perfect
14:50:06 <flaper87> we need to land those asap
14:50:10 <kairat_> especially mclaren
14:50:18 <mclaren> heh, ok
14:50:23 <flaper87> I've been pinging some folks on some reviews to get feedback on
14:50:29 <flaper87> pls, keep an eye on those
14:50:30 <kairat_> Ok, that;s all from me
14:50:48 <rosmaita> image import update patch is available for your +1/+2 pleasure: https://review.openstack.org/#/c/278086/
14:50:51 <rosmaita> also
14:50:53 <mfedosin> when it's merged we can start using swiftclient service in our driver
14:50:54 <flaper87> rosmaita: ++
14:51:06 <rosmaita> this is related to a metadefs patch, https://review.openstack.org/#/c/265152/
14:51:09 <rosmaita> the names of http headers are case-insensitive according to the standard
14:51:12 <rosmaita> glance v1 transmits image properties in headers
14:51:14 <rosmaita> v1 client gives you image property names in all lowercase
14:51:17 <rosmaita> if you use v1 client, not http directly, you expect image property names to be all lowercase
14:51:20 <rosmaita> v2 transmits image properties in response body, so they are cased however they were defined by the user who created the property
14:51:23 <rosmaita> so for property "CamelCasedProp", v1 -> camelcasedprop , v2 -> CamelCasedProp, and in nova API -> camelcasedprop
14:51:26 <rosmaita> two questions:
14:51:28 <rosmaita> (1) mfedosin: what will happen to nova api response with your v1->v2 work?
14:51:31 <rosmaita> (2) should we be advising people always to treat glance image properties as case-insensitive?
14:51:34 <rosmaita> (3) or is this not a big deal?
14:51:51 <rosmaita> so i guess that's 3 questions
14:52:33 <flaper87> I'm for (2)
14:52:36 <jokke_> rosmaita: are they case-insensitive in database?
14:52:41 <mclaren> we probably can't change the behaviour now
14:52:55 <nikhil> I think we take them as is
14:53:01 <flaper87> We can't change it and we've kinda treated them case-insensitive on glance
14:53:05 <mclaren> would 'v1 is case insensitive' 'v2 is case sensitive' be ok?
14:53:09 <jokke_> does glance allow you to set CamelCasedProp and camelcasedprop to same image?
14:53:11 <rosmaita> jokke_: actually, it's a mess in the db
14:53:20 <flaper87> hahahaha
14:53:31 <rosmaita> at least in my devstack, dupes are blocked by an index
14:53:43 <rosmaita> i need to file a bug
14:54:01 <rosmaita> so if you have CamelCase and then delete that prop
14:54:06 <nikhil> would this be affected by the table type and or DB type?
14:54:06 <rosmaita> you cannot add camelcase
14:54:12 <rosmaita> but you can add CamelCase back
14:54:14 <flaper87> rosmaita: pls do, that way we can discuss it there in more detail
14:54:23 <rosmaita> nikhil: possibly, have not investigated enough yet
14:54:27 <rosmaita> flaper87: ok
14:54:39 <nikhil> I think it might be worth asking PG folks too
14:54:53 <mfedosin> rosmaita: oh, I don't lower them
14:55:01 <jokke_> this might be rabbit hole we do not want to dig too deep :P
14:55:01 <rosmaita> mfedosin: thanks
14:55:04 <mfedosin> I think I should
14:55:20 <nikhil> agree, we need to
14:55:21 <mfedosin> but how can we send it back?
14:55:35 <mfedosin> from Nova to glance v2
14:55:57 <rosmaita> i only came across this because of the patch i linked above
14:56:04 <rosmaita> i always use lowercase
14:56:09 <nikhil> rosmaita: etherpad would be nice with your findings :)
14:56:18 <mclaren> We may just end up documenting the behaviour here. Any changes could potentially trip someonce up.
14:56:22 <rosmaita> ok, i will put something together and send to ML
14:56:22 <flaper87> I'd prefer a bug
14:56:26 <nikhil> I am still not getting mfedosin and would like this full usecase
14:56:27 <flaper87> rosmaita: or ML
14:56:29 <flaper87> :D
14:56:29 <jokke_> mclaren: ++
14:56:30 <rosmaita> ok but it is
14:56:37 <rosmaita> *bug
14:56:59 <flaper87> 4mins left
14:57:01 <flaper87> anything else?
14:57:04 <nikhil> Can we keep a review day on Wed, Mar 9th?
14:57:17 <rosmaita> question though, what to do about this in the meantime:
14:57:20 <rosmaita> #link https://review.openstack.org/#/c/265152/
14:57:26 <nikhil> #startvote Can we keep a review day on Wed, Mar 9th? (yes, no, maybe)
14:57:27 <kairat_> nikhil, Is it hackaton day?
14:57:28 <openstack> Only the meeting chair may start a vote.
14:57:34 <nikhil> kairat_: yeah
14:57:46 <kairat_> seems reasonable
14:57:50 <nikhil> I prefer triaging cores reviewing them on the final hack day
14:57:54 <flaper87> nikhil: I think we can, I was thinking we could do it on Monday but if we can align with the hakaton day, that's better
14:57:57 <jokke_> I won't be much available on 9th
14:58:06 <kairat_> ah, sorry
14:58:06 <mclaren> metadefs are v2 only, so are they affected?
14:58:30 <rosmaita> mclaren: it's got to do with the property names they "suggest"
14:58:31 <nikhil> flaper87: sounds good
14:59:05 <kairat_> ah, ok, it is working day in Russia
14:59:20 <flaper87> ok, time's up
14:59:23 <flaper87> Thanks everyone
14:59:26 <flaper87> tty next week
14:59:27 <nikhil> kairat_: cool!
14:59:28 <jokke_> thanks all
14:59:30 <kairat_> thanks
14:59:32 <nikhil> thx
14:59:33 <flaper87> rosmaita: let us know when the bug is ready
14:59:38 <flaper87> I'll look for your email on the ML
14:59:41 <flaper87> #endmeeting