22:00:17 <adrian_otto> #startmeeting containers
22:00:17 <openstack> Meeting started Tue Dec  2 22:00:17 2014 UTC and is due to finish in 60 minutes.  The chair is adrian_otto. Information about MeetBot at http://wiki.debian.org/MeetBot.
22:00:18 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
22:00:21 <openstack> The meeting name has been set to 'containers'
22:00:56 <adrian_otto> #link https://wiki.openstack.org/wiki/Meetings/Containers Our Agenda
22:01:03 <adrian_otto> #topic Roll Call
22:01:07 <adrian_otto> Adrian Otto
22:02:06 <yuanying_> Yuanying Otsuka
22:02:15 <dguryanov|2> Dmitry Guryanov
22:03:55 <adrian_otto> welcome
22:04:59 <adrian_otto> based on attendance, it seems this meeting time is less popular than 1600 UTC
22:05:00 <sdake> yo
22:05:19 <adrian_otto> hi sdake. Thanks for leading us last week.
22:05:24 <sdake> adrian_otto no problem
22:05:32 <sdake> adiran_otto ya this is so late in the day for the east and some west coasters
22:05:39 <daneyon> hey
22:06:11 <sdake> adrian_otto not sure if you had a chance to read the logs from last week, but we set a deadline of the19th for milestone #1
22:06:15 <sdake> whatever that may be ;-)
22:06:31 <adrian_otto> how many of us are joining from India, or East Asia, or Australia?
22:07:01 <adrian_otto> sdake: excllent
22:07:11 <yuanying_> I'm Japanease
22:07:54 <yuanying_> So, good morning.
22:07:56 <adrian_otto> yuanying_:  so this is like 7am your time
22:08:21 <adrian_otto> ok, let's get going
22:08:32 * dims made it back
22:08:44 <adrian_otto> were there action items to cover from last week?
22:08:51 <sdake> no
22:09:07 <adrian_otto> ok, I see that
22:09:25 <adrian_otto> so I carried forward our previous agenda
22:09:31 <adrian_otto> there we had :
22:09:34 <adrian_otto> #topic Open discussion on initial architecture
22:09:43 <adrian_otto> do we need more discussion on this?
22:09:53 <sdake> I think we covered that pretty well last meeting
22:10:00 <adrian_otto> #topic Blueprints
22:10:03 <sdake> although I have started on versioned objects support
22:10:13 <sdake> which we hadn't originally thought of using
22:10:13 <adrian_otto> are there blueprints to review / update as a group?
22:10:31 <adrian_otto> sdake: why did we think we might not need it?
22:10:44 <sdake> asalklked told me about it last night
22:10:49 <sdake> so I ported our code to it
22:10:55 <adrian_otto> ok, sweet
22:11:00 <sdake> (its from nova/ironic, going to oslo-incubator)
22:11:00 <adrian_otto> hopefully that was not a hard port.
22:11:06 <sdake> 15 patches
22:11:11 <sdake> not passing gate yet
22:11:14 <adrian_otto> sounds like a lot ;-)
22:11:16 <sdake> but works in local environment
22:11:57 <adrian_otto> ok, so let's have a look in LP for milestones and associated BPs
22:12:30 <adrian_otto> #link https://blueprints.launchpad.net/magnum Blueprint List
22:12:49 <adrian_otto> is Slower here today?
22:13:22 <adrian_otto> it looks like all the milestone 1 BP's are assigned to owners
22:13:31 <adrian_otto> https://blueprints.launchpad.net/magnum/milestone-1
22:13:39 <sdake> I think we need to move https://blueprints.launchpad.net/magnum/+spec/implement-magnum-bays into milestone 1
22:13:41 <adrian_otto> #link https://blueprints.launchpad.net/magnum/milestone-1 Milestone 1 Blueprints
22:13:44 <sdake> since that is the first feature i'll be working on
22:13:46 <sdake> and i'll own it
22:13:55 <adrian_otto> ok, proceed
22:14:26 <adrian_otto> should we look for status on the Essential ones?
22:14:59 <adrian_otto> #link https://blueprints.launchpad.net/magnum/+spec/magnum-api-service-containers
22:15:04 <adrian_otto> that is marked as not started
22:15:41 <adrian_otto> #link https://blueprints.launchpad.net/magnum/+spec/magnum-api-service
22:15:52 <adrian_otto> seem to overlap?
22:15:59 <sdake> same blueprint i think
22:16:08 <sdake> a bunch of people are working on it at once
22:16:21 <sdake> it doesn't have one owner
22:16:22 <adrian_otto> ok, so any need to merge these?
22:16:31 <sdake> yup makes sense
22:16:57 <sdake> oh wait
22:17:06 <sdake> service-containers is a breakout of the top level container
22:17:09 <sdake> so should be a dpeendency
22:17:15 <adrian_otto> we also have another pretty important one that is also marked as not started:
22:17:17 <adrian_otto> #Link https://blueprints.launchpad.net/magnum/+spec/magnum-backend-docker-api
22:17:28 <adrian_otto> #link https://blueprints.launchpad.net/magnum/+spec/magnum-backend-docker-api
22:17:38 <sdake> adrian_otto to do any of the backends, we have been blocked on RPC
22:17:39 <dims> adrian_otto: waiting for the current series from sdake to merge
22:17:44 <sdake> that was unblocked Monday
22:17:58 <sdake> then I reblocked it with the versioned objects patch stream
22:18:09 <adrian_otto> ok, fair enough.
22:18:38 <adrian_otto> so sdake, when that's working, you'll ping any devs who are queued behind that work?
22:18:51 <sdake> well I guess they can ping me if they need
22:18:57 <sdake> what we need are reviews on the open patch stream
22:19:00 <sdake> for the patches that are passing
22:19:04 <sdake> all the patches are "done"
22:19:10 <sdake> they just have minor nitpicks with the gate
22:19:16 <sdake> so a full review by a couple folks would be helpful
22:19:27 <adrian_otto> ok, I'll plan to chip in and help with that
22:19:32 <sdake> cool
22:20:11 <dims> sdake: you ok with say me fixing some of the gate stuff?
22:20:14 <adrian_otto> any trouble with things outside our code, such as help needed from members of other projects?
22:20:21 <sdake> dims i'm going as fast as the gate can go ;-)
22:20:29 <dims> sdake: ack :)
22:20:41 <sdake> unfortunately my pep8 produces different results then the gate
22:20:56 <sdake> i could use help on the test cases in the last few patches
22:21:06 <sdake> I am likley going to comment them out just to unblock us
22:21:06 <dims> sdake: aha. i can help with that
22:21:08 <sdake> and then fix later
22:21:21 <adrian_otto> #topic Open Discussion
22:21:55 <sdake> after the versioned objects patch is in
22:22:02 <sdake> we need the backend rpc server written
22:22:05 <sdake> anyone interested in tackling that?
22:22:15 <sdake> should probably be a new blueprint
22:22:20 <adrian_otto> +1
22:22:35 <sdake> i'll file a blueprint after the meeting
22:22:45 <adrian_otto> tx sdake
22:24:02 <sdake> cool four more patches passing the gate
22:24:11 <sdake> I think atm our ciritcal need is to get versioned objects merged
22:24:38 <sdake> please submit a pass of reviews today if possible, and I'll address comments when I return from dinner
22:24:49 <sdake> since version objects is blocking everything
22:24:58 <dims> ack
22:25:24 <adrian_otto> how have our reviewers been working together?
22:25:38 <adrian_otto> are the -1 votes producing meaningful improvements in the code?
22:25:51 <sdake> -1 seems to be improving the code
22:25:58 <sdake> seems like folks aren't requireing test cases yet
22:26:00 <sdake> which is good and bad
22:26:06 <sdake> I'd suggest we start enforcing test cases sshortly
22:26:11 <sdake> probably once we get the rpc backend w ritten
22:26:16 <adrian_otto> ok, so I'm happy with allowing that at this early stage under one condition
22:26:23 <adrian_otto> that we file a bug marked as tech debt
22:26:41 <adrian_otto> so that we have a full accounting of where to come back and solve it later
22:27:07 <sdake> wfm
22:27:26 <adrian_otto> ok, everyone here knows how to file bugs in LP, right? If not, see me, I will help.
22:28:30 <adrian_otto> ok, sounds like our remaining time would be best spent reviewing:
22:28:36 <adrian_otto> #link https://review.openstack.org/#/q/status:open+magnum,n,z Magnum Reviews
22:29:44 <adrian_otto> anything else to cover today? If not let's close early and focus on reviewing.
22:30:38 <adrian_otto> thanks all for attending.
22:30:50 <sdake> I had a q
22:30:52 <sdake> about the spec
22:30:55 <adrian_otto> ok
22:31:25 <sdake> #link https://review.openstack.org/136103
22:31:38 <sdake> last time about 10 folks voted to make magnum a wrapper ove rdocker+k8s
22:31:50 <sdake> but that spec isn't in alignment
22:32:05 <sdake> since this is a different group, wanted to make sure folks were aware of that
22:32:35 <sdake> #link https://github.com/stackforge/magnum/blob/master/README.rst
22:33:20 <adrian_otto> ok
22:34:10 <adrian_otto> one consideration is if this project wants to become integrated, will there be any bias against k8s as a dependency
22:34:27 <sdake> ya unclear to me - nova depends on libvirt, neutron depends on openvwitch
22:34:37 <sdake> seems like big major dependencies are ok
22:34:48 <adrian_otto> good point
22:35:16 <sdake> one cool thing is we can automate the launching of k8s in bays
22:35:19 <sdake> via virt or bare metal
22:35:21 <dguryanov|2> What about docker backend?
22:35:30 <dguryanov|2> If someone wants to use only docker
22:35:32 <adrian_otto> so the part of the spec that conflicts with the k8s approach would be the intent to use gantt as a scheduler?
22:35:46 <sdake> I'm not sure there is conflict
22:35:52 <sdake> Just that it may need to be updated
22:36:12 <sdake> duryanov|2 then yo uwouldn't be able to use the services/pods abstraction
22:36:14 <adrian_otto> ok, we can revise the spec to confirm with the adjusted intent
22:36:56 <dguryanov|2> I thought pods will be implemented in magnum
22:37:05 <adrian_otto> sdake: there are new features coming in Docker that we should be aware of… getting you a few links to PRs…
22:37:12 <dguryanov|2> Are there any documents, describing architecture?
22:37:23 <sdake> docker may implement pods natively
22:37:38 <sdake> then there is rocket to deal with as well ;-)
22:38:06 <adrian_otto> yes, which might actually be a better fit for Magnum
22:38:14 <dguryanov|2> Is it possible to implement pods in magnum-backend-docker?
22:38:26 <adrian_otto> #link https://github.com/docker/docker/issues/8681 (Docker) Proposal: Host management #8681
22:38:34 <sdake> dgur possible but difficult
22:38:42 <sdake> essentially you would have to reimplement k8s
22:38:45 <adrian_otto> #link https://github.com/docker/docker/pull/8859 Docker Clustering: Design proposal. #8859
22:39:34 <adrian_otto> #link https://github.com/docker/docker/issues/8637 (Docker) Proposal: Container groups #8637
22:39:50 <adrian_otto> ok, those are the three that I wanted to point out that may be informative for us
22:40:05 <dguryanov|2> Thanks!
22:40:20 <adrian_otto> the Host Management feature could be paired with an openstack driver to interact with Nova
22:40:34 <adrian_otto> which would give API/CLI access to add instances
22:41:06 <adrian_otto> the Clustering Design is a partial re-implementation of k8s, with existing POC code that (although requested) I don't think has been released
22:42:53 <dguryanov|2> How will magnum interact with other openstack services, glance, cinder, neutron, heat?
22:43:43 <adrian_otto> but there is a video of it…
22:43:57 <adrian_otto> #link https://docker.com/community/globalhackday October 31st video of demos from Docker Global Hackday
22:44:12 <adrian_otto> it's the last demo in a set of three I think
22:45:19 <adrian_otto> humm that link is not the same as when I first saved it
22:46:41 <dims> cluster demo? - https://www.youtube.com/watch?v=vtnSL79rZ6o
22:47:14 <adrian_otto> #link https://www.youtube.com/watch?v=lZGmvGw-mWc&feature=youtu.be Host Management Demo Video (edited) from Oct 31st Docker Global Hackday
22:47:28 <adrian_otto> dims, bingo.
22:47:49 <adrian_otto> #link https://www.youtube.com/watch?v=vtnSL79rZ6o Docker Cluster demo from Oct 31st
22:48:03 <adrian_otto> ok, so that's some food for thought
22:48:57 <adrian_otto> so should I take an action to revise the spec, sdake?
22:49:16 <adrian_otto> or should we consider additional alternatives first?
22:49:39 <sdake> not sure
22:49:47 <sdake> pretty early days, I hate to make it permanent
22:49:48 <sdake> ;-)
22:50:02 <adrian_otto> my gut says let's experiment with k8s now
22:50:16 <sdake> agree
22:50:17 <adrian_otto> and let's iterate from there
22:50:34 <adrian_otto> and keep an eye on the other options as they progress
22:51:11 <adrian_otto> the composable nature of Rocket is really attractive
22:51:45 <adrian_otto> so I'm interested in having a much closer look at that too
22:54:49 <adrian_otto> ok, let's wrap up
22:55:00 <adrian_otto> thanks for attending everyone.
22:55:05 <adrian_otto> #endmeeting