17:01:11 <nikhil> #startmeeting glance_artifacts_sub_team
17:01:12 <mfedosin> o/
17:01:14 <openstack> Meeting started Mon Jan 25 17:01:11 2016 UTC and is due to finish in 60 minutes.  The chair is nikhil. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:01:16 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:01:18 <nikhil> #chair mfedosin
17:01:19 <openstack> The meeting name has been set to 'glance_artifacts_sub_team'
17:01:20 <openstack> Current chairs: mfedosin nikhil
17:01:31 <nikhil> #topic agenda
17:01:40 <nikhil> there's no official agenda listed
17:01:52 <nikhil> so, if there's nothing we will move to free form open discussion
17:01:52 <mfedosin> okay, I have some news
17:01:56 <nikhil> excellent
17:02:07 <nikhil> #topic mfedosin: NEWS
17:02:30 <docaedo> o/
17:02:31 <mfedosin> First one, I'm responsible for artifacts project from this week
17:02:48 <mfedosin> Alex Tivelkov now fully in Murano
17:03:29 <nikhil> congrats Alex!
17:03:34 <mfedosin> I knew it last Friday, so I have not time to prepare :)
17:03:35 <nikhil> Good luck Mike!
17:03:43 <mfedosin> thanks Nikhil
17:04:14 <mfedosin> the second news is that Flavio wants us to move to separate project
17:04:31 <mfedosin> openstack/glare
17:04:47 <mfedosin> and I agree with him
17:04:59 <mfedosin> in that case we can move forward much faster
17:05:18 <mfedosin> he promised to send an email to ML tomorrow
17:06:00 <mfedosin> in long-term it's a good solution
17:06:19 <mfedosin> nikhil, I know you don't like the idea
17:06:26 <nikhil> I don't
17:06:43 <nikhil> We had this discussion many times and people had agreed otherwise
17:06:47 <kfox1111> will glance then depend on glare eventually?
17:07:18 <nikhil> It doesn't seem to align with other discussion and seems like a radical move TBH
17:07:24 <mfedosin> we need to talk about it, but I wish glare will be a replacement for glance in OS
17:07:31 <kfox1111> I was really hoping to get to artefact like nova images at some point, but that's seeming less and less likely. :/
17:08:19 <kfox1111> I proposed years ago, a symlink like thing for images and the answer came back, oh, just use artefacts for them.
17:08:19 <mfedosin> anyway glance and glare are now separate projects
17:08:37 <nikhil> I am ok if like mfedosin is saying that Artifacts will replace glance but that needs to be clearly pointed
17:08:42 <mfedosin> and we're talking about repo
17:09:11 <mfedosin> okay, not projects, services
17:09:21 <kfox1111> will all the glance team be part of glare, or are you breaking up and going your seperate ways?
17:09:39 <mfedosin> https://review.openstack.org/#/c/255274/
17:10:16 <mfedosin> I think it will be like searchlight separation
17:10:35 <kfox1111> searchlight's definatly a different beast then glance though. it crosses lots of projects.
17:10:47 <kfox1111> glare though, should replace glance when fully completed.
17:11:03 <mfedosin> so, all Glance core members, who want to participate in Glare will be promoted to cores automatically
17:11:21 <kfox1111> so if there is not buyin for that path, glance and glare will go down different paths and will eventually compete. :/
17:12:12 <nikhil> and that is my main worry
17:12:14 <kfox1111> I fear this is the glance team not wanting to deal with glare, so they are sluffing it off to its own project.
17:12:14 <mfedosin> I think we can implement current v2 Image API in glare
17:12:27 <nikhil> openstack services competing with each other with no definite goal overall
17:12:37 <nikhil> that's a really bad idea
17:12:56 <nikhil> it will divide the vision
17:13:04 <nikhil> vision, community and ideas
17:13:04 <kfox1111> exatly.
17:13:39 <kfox1111> I was worried back when there was more and more pushback to making images artefacts. IMHO, the images should have been the first thing made artefacts.
17:13:50 <mfedosin> okay, let's assume we stay in Glance
17:13:51 <nikhil> kfox1111: +1M
17:14:05 <kfox1111> but now, it seems like the image team doesn't want to do them ever.
17:14:18 <mfedosin> who will develop a standalone service?
17:14:36 <kfox1111> as an op, why deploy another service too?
17:14:38 <nikhil> I think currently operators have a lot of influence for images
17:14:45 <nikhil> we need to create equity in the group
17:15:09 <nikhil> the issue  is an image operator doesn't want to manage other artifacts
17:15:21 <nikhil> but that will not be true in all cases
17:15:33 <nikhil> and from a developer and architecture perspective it is a nightmare
17:15:45 <kfox1111> yeah. huge nightmare. :/
17:16:04 <mfedosin> I know about that
17:16:26 <mfedosin> it's the worst thing that may happen in short-term
17:16:37 <kfox1111> imo, splitting out of glance will just about be the death nell for glare.
17:16:51 <nikhil> totally
17:17:03 <nikhil> I think managing a separate project is not as easy
17:17:09 <nikhil> and I know that from SL experience
17:17:26 <mfedosin> what is SL?
17:17:29 <kfox1111> if all the glance developers that think artefacts are a good idea leave for a different project,
17:17:29 <nikhil> we need diversity, manage specs, worry about separate releases etc
17:17:32 <nikhil> searchlight
17:17:36 <kfox1111> that will leave glance unhealthy too. :/
17:17:36 <mfedosin> ah
17:17:38 <docaedo> I think the problem is less about just project management, but that the core value of glare is that it IS part of glance
17:17:45 <nikhil> separate project comes with a lot of overhead
17:18:23 <nikhil> docaedo: that's true. but the argument is made against short term vs long term effects
17:18:32 <nikhil> and I am saying in both cases, it's a terrible idea
17:18:47 <docaedo> nikhil: I agree in both cases too
17:19:22 <mfedosin> I think we can discuss it privately with Flavio today. I'll write an email for that
17:19:28 <kfox1111> another question: what is glance's vision if its not glare?
17:19:38 <docaedo> but this has been worked on for over a year, and if glance, after all this time is saying "you should go over there and work on this idea", it doesn't sound good to me. Maybe I'm projecting a worse idea than reality though :)
17:20:01 <nikhil> I too feel it the same way
17:20:08 <kfox1111> same here.
17:20:14 <nikhil> it was decided many times in summits that it will be in glance
17:20:30 <nikhil> and with no context we are making a radical change with no real different in reality
17:20:38 <nikhil> difference
17:21:10 <kfox1111> as a seperate project, it has the potential to do what egcs did for gcc. but forks generally don't end up being that successful.
17:21:27 <mfedosin> oh... it's always a tough decision
17:21:34 <flaper87> hey folks! Joining late
17:21:39 <flaper87> I read the backlog quickly
17:21:39 <mfedosin> the problem is in glance community
17:21:41 <mfedosin> too
17:21:42 <kfox1111> has glare really had that much pushback in the glance team?
17:21:44 <flaper87> (also I'm on a call)
17:22:02 <mfedosin> hi Flavio
17:22:32 <flaper87> so, I just wanted to clarify I'm not saying we should split the project out. I just said that was my preference back then and to some extent it still is but that it doesn't necessarily needs to happen.
17:22:47 <flaper87> I believe, as far as glare goes, the priority should be getting the API right
17:22:57 <flaper87> rather than focusing on whether it should be in glance's repo or not
17:23:10 <flaper87> I was very strong on the opinion that it should be its own process/service
17:23:13 <flaper87> and that's happening
17:23:19 <flaper87> the community agreed with that
17:23:31 <flaper87> Hope that clarifies my view
17:23:35 <kfox1111> agreed. getting the api right is very important. but so is team buy in, and minimizing the number of services an op has to deal with, and developers have to target.
17:23:44 <flaper87> I think we should get the Glance's community on glare
17:23:57 <kfox1111> so starting it out as a seperate project might be ok, so long as the team supports its eventual merge, and that it becomes part of glance eventually.
17:23:58 <flaper87> kfox1111: fwiw, I've been bringing this up to the Glance's meeting for two weeks
17:24:11 <flaper87> we have a fasttrack in place for artifacts that I think we should remove
17:24:24 <flaper87> because I think that's also preventing the community from jumping in
17:24:59 <kfox1111> chicken and egg problem sort of. if its not part of glance, people might not contribute to it.
17:25:16 <kfox1111> if its part of glance, glance might be way to slow to review?
17:25:25 <mfedosin> people don't now :)
17:25:43 <kfox1111> less likely if its not part of glance though...
17:25:44 <mfedosin> We have a good team for artifacts
17:26:00 <mfedosin> 3-4 ppl from Mirantis and 2-3 from IBM
17:26:08 <kfox1111> nice.
17:26:09 <flaper87> kinda! It's a bit late for mitaka anyway. WE need to focus on what was discussed and how we can help moving the service forward
17:26:15 <mfedosin> it's enough for the first time
17:26:29 <nikhil> we need to take baby steps
17:26:41 <flaper87> that's what I'm saying
17:26:43 <kfox1111> is the goal to move all of glance to being glare backed once glare is stable?
17:26:50 <nikhil> I think we should all think about the API and not the project atm
17:27:01 <flaper87> Again, as far as glare goes, we need to make the API stable
17:27:09 <docaedo> +1 on focusing on API, and thanks flaper87 for clarifying things
17:27:37 <flaper87> docaedo: no worries, hope you're all still alive and no one had a heart attack
17:27:45 <flaper87> :D
17:27:45 <kfox1111> +1 for a stable api. -1 for lack of vision. lack of vision kills projects. :/
17:28:35 <flaper87> kfox1111: I think the vision has always been that we need to make some sense of glare's API and have it bake glance
17:28:35 <mfedosin> thanks Flavio
17:28:39 <flaper87> in the long run
17:29:26 <kfox1111> k. I totally get glance not wanting to commit to an unstable api.
17:29:54 <kfox1111> but its a different thing if the team phylisophically disagrees with the direction. just want to make sure that's not a thing.
17:30:47 <mfedosin> I think we have to finish the meeting
17:31:15 <flaper87> kfox1111: I don't think that's a thing. I think it's a matter of priorities, bandwidth and well $PUT_THE_NAME_OF_YOUR_MANAGER_HERE
17:31:22 * flaper87 stfu
17:31:33 <mfedosin> flaper87: will you send an email tomorrow?
17:32:28 <kfox1111> flaper87: ok. that's good to hear. thanks.
17:32:40 <nikhil> I guess we need to close the mtg due to time and discuss offline
17:33:06 <mfedosin> not offline :))) it's hard
17:33:17 <nikhil> I meant -glance
17:33:33 <nikhil> I will step down as chair and let mfedosin decide on when to close the mtg
17:33:40 <mfedosin> #endmeeting