14:02:35 <nikhil_k> #startmeeting glance_artifacts_sub_team
14:02:36 <openstack> Meeting started Mon Nov  9 14:02:35 2015 UTC and is due to finish in 60 minutes.  The chair is nikhil_k. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:02:37 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:02:39 <openstack> The meeting name has been set to 'glance_artifacts_sub_team'
14:02:43 <kairat> o/
14:02:52 <nikhil_k> #topic agenda
14:02:59 <nikhil_k> #link https://etherpad.openstack.org/p/glance-artifacts-sub-team-meeting-agenda
14:03:02 <nikhil_k> hi kairat
14:03:12 <nikhil_k> should we wait for others?
14:03:17 <kairat> hi nikhil_k
14:03:30 <mfedosin> o/
14:03:32 <kairat> yep, mfedosin will be here soon
14:04:06 <dshakhray> o/
14:04:40 <nikhil_k> cool, nice to see you all here
14:04:46 <nikhil_k> should we wait for alex?
14:05:03 <kairat> nikhil_k, AFAIK he is on vacation this week
14:05:05 <mfedosin> He's on vacation
14:05:16 <mfedosin> he will be here next week
14:05:22 <Jokke_> that lazy git ;)
14:05:25 <kairat> and that's the main problem right now:)
14:05:36 <nikhil_k> heh
14:05:48 <mfedosin> kairat proposed code to remove current v3 implementation out from glance
14:05:58 <nikhil_k> Well we just have one item on the agenda for today
14:06:15 <nikhil_k> That I think we need to discuss again with Alex
14:06:25 <kairat> +1 to nikhil_k
14:06:51 <mfedosin> https://review.openstack.org/#/c/241265/
14:07:05 <kairat> mfedosin, it needs to be improved to pass the tests
14:07:09 <kairat> I will do it soon
14:07:15 <kairat> but need to discuss with Alex
14:07:21 <mfedosin> sure, it's not urgent
14:07:31 <nikhil_k> #topic review the current v3 impl proposal
14:07:50 <nikhil_k> #link https://review.openstack.org/#/c/241265/
14:08:00 <nikhil_k> Can I ask what's the motivation behind this?
14:08:28 <mfedosin> nikhil_k: it was decided on the summit
14:08:36 <nikhil_k> Why can't we roll over to the new version rather than explicitly doing this?
14:08:50 <mfedosin> to make artifacts the whole new service
14:09:04 <Jokke_> mfedosin: it's not gonna go anywhere from the repo
14:09:09 <kairat> mfedosin, why can't we adopt the new code for this
14:09:17 <kairat> *old code sorry
14:09:50 <Jokke_> that's like 20min of work to create the paste.inis and the hooks needed and decouple it from the current api service
14:10:04 <mfedosin> two artifact implementations may confuse users
14:10:08 <nikhil_k> Just some context
14:10:15 <Jokke_> mfedosin: same repo, same codebase just its own endpoint
14:10:26 <nikhil_k> I have been asked by a few folks about this experimental version of the API
14:10:38 <Jokke_> mfedosin: what do you mean about two implementations?
14:10:40 <nikhil_k> and I think this will help us get some BETA testing done
14:11:03 <mfedosin> Jokke_: this code is considered as old and unstable
14:11:10 <nikhil_k> so, keeping it in the repo would increase our chances of finding all the wrong things and importantly blunders
14:11:19 <mfedosin> I mean the existing v3 code
14:11:31 <nikhil_k> sure, but it's off by default?
14:11:33 <mfedosin> so anyway we have to rewrite it from scratch
14:11:35 <nikhil_k> mfedosin: ^
14:11:55 <mfedosin> yes, v3 is off
14:12:00 <Jokke_> mfedosin: ahh ... I din't know that full rewrite was still the plan
14:12:29 <mfedosin> to adopt oslo.vo we have to :(
14:12:54 <Jokke_> ok, sorry my bad ... didn't realize it involving that much
14:12:55 <mfedosin> there will be new plugins
14:13:05 <Jokke_> then I do undertand the will to clean the repo first
14:13:11 <mfedosin> and db api will be changed too
14:13:32 <mfedosin> and api as well :)
14:13:40 <mfedosin> so it's better to remove it from repo
14:13:47 <mfedosin> and rewrite from scratch
14:13:57 <nikhil_k> sure, but we are not doing the migration yet as far as I could tell
14:14:20 <Jokke_> so how we are going to keep the support for murano on the experimental one, or are going to just leave it to stable/liberty to bitrot
14:14:23 <Jokke_> ?
14:14:26 <mfedosin> about db tables - they will be different
14:14:54 <mfedosin> so we will remove the existing and create new
14:15:11 <nikhil_k> yeah, the murano client /glance feature branch would be rendered useless
14:15:23 <mfedosin> Murano uses artifacts as experimental feature
14:15:39 <mfedosin> and they are okay with completely removing it
14:16:28 <nikhil_k> ok, it might help to have this discussion on the spec itself so that everyone is on the same page for that.
14:16:58 <mfedosin> I wish Alex to be here to confirm my words
14:17:05 <nikhil_k> well, at least I think we should document this downgrade and rewrite
14:17:15 <mfedosin> yes - I will prepare a light spec
14:17:31 <mfedosin> that describes this transition
14:17:44 <mfedosin> glance-v3 -> glare
14:18:11 <nikhil_k> sounds good, let's start with discussion there so that all the underlying points/concerns are clear.
14:18:53 <mfedosin> I'm going to start working on it on Wednesday, so next week we can discuss it
14:19:10 <nikhil_k> #info all: comment on the to be proposed {lite}-spec by mfedosin about glare
14:19:23 <mfedosin> thanks ;)
14:19:39 <nikhil_k> thank you!
14:20:33 <nikhil_k> #topic start review questions from the summit
14:20:36 <nikhil_k> #link https://etherpad.openstack.org/p/mitaka-glance-artifacts-review
14:21:00 <mfedosin> "Agreed to run artifacts in a separate port/process! YAAAAAAAAAAAAAAAAAYYYYYYYYYYYYYYYY!!!!!!!!!!!!!!!!!!!:)"
14:21:06 <nikhil_k> We have agreed to move it to a separate process and we have agreed to call it Glare.
14:21:08 <nikhil_k> :)
14:21:28 <mfedosin> Glare is awesome name
14:21:48 <nikhil_k> Would this process have similar semantics as g-api?
14:22:06 <kairat> Once app catalog use it, will it stay forever as separate service?
14:22:27 <mfedosin> frankly speaking my vision may differ from the Alex's
14:22:46 <mfedosin> but yes, it will be similar to g-api
14:23:14 <nikhil_k> so, we are going to use the same libraries wsgi, eventlet etc?
14:23:16 <mfedosin> and it will be absolutely separate service
14:23:32 <nikhil_k> should we consider some research around webob, twisted?
14:23:45 <mfedosin> but Glare api will be called 'Artifact API'
14:24:05 <mfedosin> nikhil_k: for sure we can discuss it
14:24:17 <mfedosin> twisted and other stuff
14:24:26 <nikhil_k> I think the branding might just have to be done during adoption as we are a bit unsure on what it completely looks  like.
14:25:10 <mfedosin> I believe next week Alex will tell you more about Glare
14:25:31 <nikhil_k> I think we need to consider some performance constraints for the regular upload download workflows and the lessons learnt from Glance on that.
14:25:46 <mfedosin> one thing he asked me to do before his vacation - to remove the existing code from the repo
14:26:08 <nikhil_k> It might be best to design the API in a way that we  don't have to improvise too much during optimizations/
14:26:55 <mfedosin> nikhil_k: agree
14:27:02 <kairat> +100
14:27:03 <nikhil_k> mfedosin: I see, in that case hopefully he can simply comment "I agree" on the {lite-}spec when that's up and ready?
14:27:43 <mfedosin> I won't work on it before I have a clear plan
14:28:23 <mfedosin> and implementation description
14:28:45 <nikhil_k> No worries, I think we have 2-3 people who may help with establishing the plan, etc.
14:29:16 <mfedosin> yes
14:29:45 <nikhil_k> so, the take away from this question was that 1) yes, let's consider twisted, webob etc 2) API should be better performant
14:29:48 <mfedosin> and I like that AppCatalog agreed to use Glare as backend
14:30:23 <nikhil_k> I see
14:30:37 <nikhil_k> Oops, we are out of time today.
14:30:49 <nikhil_k> Let's continue the review items next week.
14:30:55 <mfedosin> thanks folks
14:31:00 <kairat> thanks
14:31:02 <nikhil_k> Looks like a solid start.
14:31:06 <nikhil_k> Thanks all!
14:31:08 <mfedosin> yup!
14:31:14 <nikhil_k> #endmeeting