14:07:07 <nikhil> #startmeeting glance_artifacts_sub_team
14:07:08 <openstack> Meeting started Mon Nov  2 14:07:07 2015 UTC and is due to finish in 60 minutes.  The chair is nikhil. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:07:09 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:07:11 <openstack> The meeting name has been set to 'glance_artifacts_sub_team'
14:07:22 <nikhil> ativelkov: mfedosin: anyone around?
14:07:31 <nikhil> sigmavirus24_awa: ^
14:07:45 <mfedosin> nikhil, o/
14:07:52 <nikhil> hey
14:08:04 <mfedosin> kairat is here too
14:08:09 <nikhil> ah ha
14:08:15 <mfedosin> so, no special news
14:08:19 <nikhil> nice, hi kairat
14:08:27 <nikhil> gotcha
14:08:34 <kairat> o/
14:08:39 <kairat> hi guys
14:08:45 <nikhil> I figured it's best to log the updates or no updates so I started the meeting
14:08:47 <mfedosin> we decided to move glance artifacts to separate service
14:09:01 <mfedosin> Codename: Glare
14:09:20 <nikhil> separate (set) of processes
14:09:31 <nikhil> right?
14:09:47 <mfedosin> I'm going to prepare a spec, that describes this process
14:10:00 <nikhil> I see, cool
14:10:09 <mfedosin> it will be standalone service with own endpoint port etc
14:10:56 <mfedosin> and completely independent from glance
14:11:34 <mfedosin> here's the etherpad link https://etherpad.openstack.org/p/mitaka-glance-artifacts-review
14:12:37 <nikhil> right now we have the same database table
14:14:29 <nikhil> I will add an item for next week's discussion on this.
14:14:37 <mfedosin> I think we have to create new set of tables for Glare
14:15:06 <mfedosin> and drop the existing artifacts tables
14:16:26 <kairat> before dropping the tables we need to think if we are going to support different database api's
14:16:40 <nikhil> that is a good question
14:16:54 <nikhil> which others are you thinking?
14:17:09 <mfedosin> database api will be different
14:17:56 <mfedosin> actually I like to have glance and glare separated
14:18:13 <nikhil> are we talking about database connector api or the models and the sqlalchemy api in the code?
14:18:13 <mfedosin> because having an autonomous service will allow us to be consistent with the requirements of defcore
14:18:40 <mfedosin> nikhil, everything
14:18:50 <nikhil> I think we also need to discuss if it will need to be evaluated by defcore
14:19:11 <mfedosin> there will be new database models with different tables, new database api
14:19:34 <nikhil> hm
14:19:53 <nikhil> is there any project that supports more than sqlalchemy?
14:20:06 <mfedosin> no, defcore is okay with this - glare won't be a part of it
14:20:36 <mfedosin> no, there will be only sqlalchemy
14:20:48 <mfedosin> (and registry v2)
14:21:22 <mfedosin> but api will differ from the current
14:21:48 <nikhil> ok, looking forward to the details on the spec
14:22:09 <mfedosin> new database api won't support transitive dependencies, show level
14:23:08 <mfedosin> and generally it will be easier.
14:23:43 <mfedosin> there is an open question about client
14:23:59 <nikhil> yeah
14:24:06 <kairat> nikhil, AFAIK glance has simple_api and db_api
14:24:11 <kairat> ceilometer has mondo
14:24:24 <kairat> *mongo
14:24:45 <nikhil> kairat: yeah, someone was asking in Vancouver if we will let them add nosql support
14:25:04 <kairat> mfedosin, the question is good
14:25:11 <nikhil> that was a big refactor to the db api so we asked them to evaluate more
14:25:15 <kairat> will it be part of glance client?
14:25:25 <mfedosin> in Vancouver they wanted to use Cassandra
14:25:37 <mfedosin> which is a bad idea IMO
14:26:03 <mfedosin> yesm it was decided to use the existing glance client
14:26:42 <nikhil> I think it would be good to keep this in py-glanceclient so that we don't have to deal with releasing multiple libs and then maintain them. one gate, one team, clarity on the releases etc
14:26:45 <mfedosin> but if it becomes too huge we can separate them and create glareclient
14:27:20 <mfedosin> nikhil, yes - we will begin with glanceclient
14:28:06 <nikhil> cool
14:28:24 <mfedosin> it all artifacts updates at the moment
14:28:45 <mfedosin> you can create an action item for me to write a spec
14:28:50 <nikhil> great!
14:29:37 <kairat> mfedosin, have you got any feedback from murano and app catalog about this?
14:29:37 <nikhil> mfedosin: it might be worth listing all the action items for artifacts on the spec so that different people can pick it up
14:29:42 <kairat> Are they ok with this?
14:29:54 <kairat> or we decided this internally
14:30:07 <mfedosin> yes, app-catalog willed to use glare when it's stable
14:30:44 <mfedosin> and murano is okay with everything :)
14:30:52 <nikhil> :)
14:31:18 <kairat> so this will never be api v3)
14:31:36 <nikhil> Great work folks. we are out of time for today.
14:31:46 <mfedosin> thanks!
14:31:59 <nikhil> Let's discuss the v3 part when ativelkov is back or next or the week after.
14:32:13 <kairat> ok, thanks
14:32:14 <nikhil> I would like his feedback from the talk etc too
14:32:18 <kairat> bye
14:32:39 <nikhil> thanks!
14:32:40 <nikhil> #endmeeting