14:02:14 <markwash> #startmeeting glance
14:02:15 <openstack> Meeting started Thu Jan 30 14:02:14 2014 UTC and is due to finish in 60 minutes.  The chair is markwash. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:02:16 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:02:19 <openstack> The meeting name has been set to 'glance'
14:02:32 <markwash> greetings
14:02:36 <arnaud__> hi markwash
14:02:40 <rosmaita> o/
14:02:43 <iccha> hey
14:03:27 <markwash> I'm a little slow this morning
14:03:40 <zhiyan> hello
14:03:41 <markwash> I think I overcame my jetlag on the trip back almost immediately
14:04:09 <markwash> I scratched out a bit of an agenda for today here
14:04:21 <markwash> #link https://etherpad.openstack.org/p/glance-team-meeting-agenda
14:04:43 <markwash> I can't actually remember all of the items we pushed off to this meeting vs the next at the mini summit
14:05:19 <markwash> #topic summit post-processing
14:05:36 <markwash> #link https://wiki.openstack.org/wiki/Glance/IcehouseCycleMeetup
14:05:59 <rosmaita> i think today is just the drivers
14:06:03 <markwash> in case folks haven't seen that already, its got a lot of the etherpad notes from the summit, not all of the slides
14:06:09 <rosmaita> glance.store is next time
14:06:24 <zhiyan> rosmaita: +1
14:06:31 <arnaud__> ok sounds good
14:06:36 <zhiyan> arnaud__: thanks
14:06:54 <markwash> since it was first in the list, let's just do a quick reminder about what we want to have prepped for next week on glance.store
14:07:19 <arnaud__> I think it would be nice it flavio could be around to discuss it
14:07:23 <zhiyan> (a little distracted, tonight is China New Year eve)
14:07:24 <arnaud__> since he was the initiator
14:07:34 <markwash> makes sense
14:07:47 <markwash> zhiyan: happy chinese new year!
14:07:57 <zhiyan> markwash: thanks!
14:08:01 <arnaud__> +1 :)
14:08:02 <zhiyan> happy new year!
14:08:08 <zhiyan> arnaud__: thanks
14:08:23 <rosmaita> zhiyan: happy new year from virginia, too
14:08:28 <arnaud__> I will ping him this week to make sure he will be there next meeting
14:08:37 <zhiyan> rosmaita: thanks
14:09:39 <zhiyan> arnaud__: if you ok, pls involve me, thanks.
14:09:44 <markwash> Okay so I'll add glance store to the planned schedule for next meeting
14:09:53 <markwash> and here's the link to that topic so folks can prepare as well
14:09:57 <markwash> #link https://etherpad.openstack.org/p/right_place_for_glance.store_modules
14:10:53 <markwash> okay
14:10:55 <arnaud__> ok zhiyan
14:11:05 <markwash> now onto drivers
14:11:32 <markwash> I cleaned up glance-drivers in launchpad, and made the first appointment
14:11:45 <markwash> first I added rosmaita
14:11:56 <rosmaita> thank you!
14:12:15 <markwash> we'll see how well we can define the team duties and figure out the workload and then see about adding more folks
14:12:55 <markwash> so all that's left is to set up a coordination meeting
14:13:37 <markwash> rosmaita: would next tuesday sometime work for you?
14:13:39 <zhiyan> markwash: sorry, may i know what's "glance driver"? give a quick sync-up?
14:13:48 <markwash> oh sure, sorry I'm going too fast
14:14:02 <rosmaita> i will check date while you fill in zhiyan
14:14:08 <zhiyan> markwash: thanks! (streaming delay killed me, seems)
14:14:21 <markwash> in iccha 's session about blueprint triage, one of the proposals was to adopt a nova-style drivers group to perform blueprint triage
14:14:29 <markwash> we weren't sure we needed to go to the level of sponsorship
14:15:19 <zhiyan> markwash: ok. make sense.
14:15:52 <flaper87> o/
14:15:57 * flaper87 is late
14:15:58 <markwash> so the initial goals are to just figure out the workflow and to try to clean up the existing blueprints
14:16:13 <arnaud__> markwash: any idea how many people in the team?
14:16:28 <markwash> arnaud__: not sure yet
14:16:46 * flaper87 is working on the virtual-size and glance.store bps
14:16:58 <markwash> arnaud__: two feels a little small but I think is right just for bootstrapping
14:17:10 <arnaud__> flaper87: we will talk about the glance.store next week, hope that's ok for you :)
14:17:31 <rosmaita> markwash: tuesday is wide open for me right now
14:17:32 <flaper87> arnaud__: I'm already working on that
14:17:53 <flaper87> next weeks meeting is late for me. I'll try to make it anyway
14:18:10 <markwash> rosmaita: okay cool I'll follow up with you to pick a time
14:18:49 <zhiyan> flaper87: btw, i saw your initial code with your personal trunk. it's a good start, but tbh i think it need more wrap
14:19:01 <flaper87> zhiyan: it's not finished
14:19:04 <flaper87> I'm working on it
14:19:17 <flaper87> and I ned to submit patches in both glance and glance.store to make the transition easier
14:19:23 <flaper87> need*
14:19:37 <markwash> flaper87: one thing we were talking about a bit is the timeline
14:19:55 <zhiyan> flaper87: np. actually i'm interested in it
14:20:14 <flaper87> and I would target a huge / perfect wrapper to begin with, TBH. I'd prefer transitioning to glance.stores w/ backwards compatibility and few changes. Then we can improve the wrapper.
14:20:23 <flaper87> that's what we planned with olso.messaging to
14:20:25 <markwash> flaper87: I think it met with pretty broad consensus about being a good idea to break out the stores, but there were two hangups. . one we want to make sure this improves the testing story rather than hurts it, which I think is doable
14:20:29 * flaper87 doesn't want to hijack the meeting
14:20:30 <iccha> especially if we are waiting for changes on the scrubbing locations to get in?
14:20:45 <markwash> flaper87: the other concern was more about how we might use it in the client
14:20:55 <markwash> and I think that's really important but not necessarily so blocking
14:21:05 <flaper87> markwash: re testing. I fully agree! I'm actually spending more time there now
14:21:20 <flaper87> cleaning up things that shouldn't have been inside the store package
14:21:26 <flaper87> markwash: yeah, that's a good question
14:21:40 <flaper87> I think that relates a lot w/ direct_url access being enabled
14:21:51 <flaper87> but I admit I still need to think more about that particular bit
14:21:59 <arnaud__> I think what we said it that the lib could be ready by I-3 but integrated later with glance/glance-client. what do you think flaper87?
14:22:36 <flaper87> arnaud__: I'd agree w/ integrating it after I-3 in the client but I think glance would benefit from a non-breaking transition to glance.store during I-3
14:23:30 <markwash> flaper87: stuart had some concerns that raised red flags for that general plan of integrating with the client, mostly around having to have such a high privilege / trust level for clients. . but I think we just need to be sure we have a clear strategy for less trusted clients that still makes sense
14:23:42 <markwash> okay, moving on, we can cover this more next meeting I think
14:23:59 <markwash> #topic glance summit post-processing: glance mission
14:24:02 <flaper87> markwash: agreed
14:24:18 * iccha needs to step away for a bit
14:24:35 <markwash> during the summit we had a lot of fun chatting with jbernard and georgy and other mirantis folks about how glance might work for other projects
14:25:15 <markwash> and the general idea was that glance's mission is a bit broader and different than "image store"
14:25:32 * flaper87 is so happy to hear that!
14:25:40 <markwash> and randallburt also
14:26:05 <markwash> there are some steps we need to follow to build momentum for this in Atlanta
14:26:39 <zhiyan> sounds good
14:26:39 <markwash> I think we should have a statement of this mission, actually it can be the the Images Program mission in the governance repo or wherever
14:26:58 <markwash> and we need to have some proposals for how we can follow that mission
14:27:05 <jbernard> markwash: im happy to help in this regard
14:27:17 <flaper87> this is going to be interesting
14:27:35 <flaper87> because an Image program for a broarder mission doesn't seem to be a god fit, we'd have to rename the program
14:27:43 <markwash> I wanna give a little preview of that mission here
14:27:48 <flaper87> so establish a way to deprecate an existing program and create a new one
14:27:52 <flaper87> anyway, TC discussion
14:27:53 <markwash> flaper87: yeah good point, which might make things a bit awkward
14:28:23 <flaper87> s/god/good/
14:28:30 * flaper87 is still hungry and there are no gummy bears left
14:28:50 <markwash> the preview is, Glance's role is to store and track the "starting points" or templates for other openstack services
14:29:25 <markwash> and to make those "starting points" discoverable, shareable, and to provide some gating and validation to make the interoperability story work in general
14:30:03 <flaper87> I wish we could find a better term other than template
14:30:06 <markwash> yeah
14:30:14 <markwash> also better than "starting points" :-)
14:30:22 <flaper87> LOL, you said that
14:30:24 <flaper87> :P
14:30:40 <markwash> oh I fully admit this is definitely a work in progress
14:30:43 <markwash> that's why I bring it up here
14:30:52 <flaper87> but the overall idea sounds really good, IMHO
14:30:58 <rosmaita> +1
14:31:00 <markwash> I'd really like for us to have some good language around it before throwing it out as troll bait
14:31:02 <arnaud__> +1
14:31:17 <rosmaita> markwash: good idea!
14:31:27 <markwash> because I think the next step in the process is the ML and any related TC discussions
14:31:40 <rosmaita> there will be strong but not necessarily helpful opinions on this
14:31:50 <markwash> (in parallel to the code process we can talk about, which shouldn't feel blocked by the mission imo)
14:32:27 <markwash> it might actually be good to just be ready with a mission and then wait a bit until we are adding some of the code
14:32:55 <flaper87> markwash: fully agree
14:33:03 <markwash> that can sometimes ease the confusion for folks
14:33:21 <flaper87> and it sounds like for this transition we'll need an API v3 :)
14:33:31 <markwash> flaper87: yeah we aren't quite clear on that
14:33:46 <markwash> it might be able to be bolted onto v2, which could ease the transition a bit
14:34:04 <markwash> I think that's the current code plan
14:34:15 <flaper87> I'm not sure about that. What worries me the most is that we don't have a client release with support for v2
14:34:20 <markwash> folks are just going to submit patches against v2 that sort of do their own thing
14:34:46 <flaper87> My concern is that doing that could end up in corner cases that we won't be able to see
14:35:07 <flaper87> becase we are going to change the existing API instead of working on a new API thought for this transition
14:35:08 <markwash> flaper87: I'm not sure I follow about the client release
14:36:02 <markwash> I've been using v2 with the current glanceclient
14:36:16 <markwash> at least, I thought so! :-)
14:36:46 <flaper87> markwash: yeah but the API is not fully supported, right? I mean, we don't have support for multi-locaitons in the client, for example
14:36:48 <flaper87> AFAIK
14:37:14 <arnaud__> flaper87: that's correct
14:37:24 <zhiyan> flaper87: are you talking PATCH api?
14:37:49 <flaper87> zhiyan: I'm talking about multi-location in general.
14:37:53 <flaper87> but lets stay in-topic
14:38:07 <markwash> I guess, are you concerned that the v2.x series is continuing to advance, ahead of the client, while we would be doing yet more stuff?
14:38:24 <flaper87> what I wanted to say is that implementing a v3 API w/o having full support for v2 would be weird
14:38:31 <flaper87> markwash: yeah, pretty much that
14:38:31 <markwash> oh
14:38:37 <markwash> yeah
14:38:41 <markwash> that's my concern about v3 as well
14:38:46 <rosmaita> flaper87: what multi-location support is missing in client?
14:38:53 <markwash> revving major version is really disruptive for downstream
14:38:56 <flaper87> so, I'm there are som good points about working on a v3 but also things that make me think that v2.X would be better
14:38:59 <markwash> so I think we should only do it if we have to
14:39:21 <flaper87> rosmaita: adding new locations for an image. AFAIK, that hasn't been implemented. I could be wrong, though
14:39:30 <flaper87> I admit I've been heads down in the server myself
14:39:54 <markwash> my current thinking is that we have a broad strategy of adding new, more generic resource types to v2 to deal with all the
14:39:57 <markwash> darnit
14:39:59 <markwash> formatting
14:40:05 <markwash> broad strategy of:
14:40:10 <zhiyan> rosmaita: imo, except receive multi-locations, others are all miss currently
14:40:35 <markwash> 1) add new, more generic-ish resource types to v2 to deal with instances, heat templates, murano packages, etc
14:40:48 <arnaud__> markwash: artifacts? :)
14:40:51 <markwash> 2) try to shrink v2/images to *just block devices* none of this multi-disk stuff
14:40:55 <markwash> arnaud__: yeah artifacts
14:41:02 <markwash> thats what I think they were calling it, right?
14:41:07 <arnaud__> yep
14:41:10 <markwash> okay cool
14:41:37 <markwash> and 3) have the new artifacts stuff maintain references to v2/images where appropriate in their hierarchy
14:41:58 <markwash> arnaud__: artifacts might be the word to solve our previous mission terminology problem as well
14:42:12 <arnaud__> yes, that's what I was thinking too
14:42:21 <rosmaita> zhiyan: you can also set the multi-locations, the json pointer code was enhanced so PATCH would work for that
14:42:28 <arnaud__> btw, markwash: I would be interested to work on the instance templates
14:43:19 <zhiyan> rosmaita: from glanceclient?
14:43:54 <markwash> arnaud__: cool, I think jbernard is sort of "carrying the torch" on instance templates for now but I know we're going to need a broad front in order to get buyin across the projects, there is probably a place you can fit into those efforts
14:44:09 <zhiyan> rosmaita: i added location patch api support on server side.
14:44:12 <arnaud__> ok sounds good
14:44:30 <rosmaita> zhiyan: probably not
14:44:35 <jbernard> markwash, arnaud__: yes, all help is appreciated
14:44:59 <markwash> jbernard: great, thanks
14:45:48 <markwash> I know a lot of folks here will be interested in helping, there's always the additional challenge of keeping focus if we end up with too many cooks :-)
14:46:11 <markwash> not saying we'll end up there but its a good thing to be aware of
14:46:27 <markwash> okay, I think I sidetracked myself
14:46:51 <markwash> just a quick announcement, I'll be providing some more summary material from the summit soon
14:47:04 <markwash> and ashwini can post videos as appropriate (?)
14:47:14 <markwash> no promises on the quality, we weren't checking or talking directly into the mic or anything
14:47:24 <rosmaita> she will not post inappropriate videos
14:47:29 <markwash> haha
14:47:51 <markwash> moving on
14:47:54 <markwash> #topic icehouse 3
14:48:13 <ashwini> markwash: yes i will look into sorting those soon
14:48:14 <markwash> first thing to note, we're trying to coordinate a cross-project code proposal deadline
14:48:37 <markwash> I'm not sure if the date has been announced yet, but it was looking like february 18 which is quite soon
14:48:57 <markwash> code proposal meaning the review needs to be up in gerrit by then
14:49:17 <rosmaita> so, would glanceclient stuff be excluded from the freeze?
14:49:20 <markwash> look to the ML for more details, either already there or coming soon
14:49:41 <markwash> glanceclient I think is generally excluded from the feature freeze, so I expect this wouldn't really affect it either
14:50:08 <rosmaita> ok, could give us time to get the client up to icehouse API before icehouse release
14:50:29 <markwash> in terms of review priorities, I think folks should be focusing on stuff in https://launchpad.net/glance/+milestone/icehouse-3
14:50:49 <markwash> there are a bunch of things there that are already proposed and done essentially
14:50:57 <markwash> carryover from icehouse 2
14:51:02 <markwash> the more of those we can land early the better
14:51:29 <markwash> the gate is a real blocker and I'm worried we'll be in for sad times as we approach the Feature Freeze window
14:52:11 <markwash> okay, that's all I have, I'm going to kick it over to open discussion but feel free to discuss this more if you have questions
14:52:14 <markwash> #topic open discussion
14:52:52 <zhiyan> markwash: actually i have three review request for the testing
14:53:26 <markwash> zhiyan: cool, links?
14:53:32 <zhiyan> one is testrepository migration, flaper87 just mentioned in team room, i think it can help gate
14:53:33 <zhiyan> https://review.openstack.org/#/c/59699/
14:53:42 <zhiyan> second is https://bugs.launchpad.net/glance/+bug/1274365
14:53:57 <flaper87> please, lets merge the testr one
14:54:08 <markwash> sounds like the testrepository one can fix my virtualenv/pip problems too?
14:54:12 <zhiyan> third is https://bugs.launchpad.net/glance/+bug/1272136
14:54:31 <zhiyan> testr can't, but #59699 could :)
14:54:44 <zhiyan> actually it can fix two issues
14:54:58 <zhiyan> at last, it can make "run_tests.sh" real work
14:56:18 <markwash> I'm running the testrepository one now locally btw
14:56:47 <zhiyan> markwash: yes, doublecheck will be better
14:57:03 <zhiyan> that's all for me.
14:57:40 <hdd_> hello, I have been working on cross-service request ID, which will log the request ID mapping between services
14:57:52 <hdd_> this would require updates to all the python clients
14:58:08 <hdd_> I was wonder what the procedure is for cutting a new release of glanceclient, once I have the code done
14:58:29 <zhiyan> hdd_: it can be used to help log analysis?
14:58:30 <markwash> hdd_: I just cut the release
14:58:35 <markwash> I mean
14:58:41 <markwash> the process is "markwash does it"
14:58:43 <hdd_> zhiyan: yes, that's the idea
14:58:45 <hdd_> (-:
14:59:22 <markwash> hdd_ I push a git tag to openstack or gerrit or whatever, a tarball is built by jenkins, and then jenkins uploads stuff to pypi
14:59:51 <hdd_> ok, so I'll just contact you when the change is merged?
14:59:57 <markwash> yeah
15:00:02 <hdd_> cool
15:00:05 <markwash> it never hurts to pester me about releasing the client
15:00:12 <markwash> sad but true
15:00:27 <markwash> okay I guess we'll make way, in case there is another meeting right after
15:00:33 <markwash> thanks everybody!
15:00:36 <rosmaita> bye!
15:00:42 <arnaud__> bye
15:00:44 <markwash> #endmeeting