20:01:34 <sdake> #startmeeting kolla 20:01:35 <openstack> Meeting started Mon Mar 9 20:01:34 2015 UTC and is due to finish in 60 minutes. The chair is sdake. Information about MeetBot at http://wiki.debian.org/MeetBot. 20:01:36 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 20:01:39 <openstack> The meeting name has been set to 'kolla' 20:01:41 <sdake> #topic agenda 20:01:45 <sdake> rather 20:01:46 <sdake> #undo 20:01:46 <openstack> Removing item from minutes: <ircmeeting.items.Topic object at 0x96cdbd0> 20:01:49 <sdake> #topic rollcall 20:01:56 <rhallisey> hi 20:01:57 <sdake> \o/ :) 20:02:01 <daneyon_> hey 20:02:01 <sdake> hey ryan 20:02:12 <jpeeler> o/ 20:02:16 <sdake> hey jeff 20:02:46 <sdake> #topic CI 20:03:02 <sdake> so nikolay got our integrated functional test cases running 20:03:15 <rhallisey> great 20:04:11 <sdake> for example: 20:04:15 <sdake> #link https://review.openstack.org/#/c/162034/ 20:04:24 <sdake> has a 20:04:27 <sdake> #link http://logs.openstack.org/34/162034/1/check/check-kolla-functional-f21/7f365e3/ 20:04:32 <sdake> a check kolla functional 20:04:45 <sdake> I sent him an email to make the gate voting 20:04:55 <sdake> so that review should be going through shortly 20:05:09 <sdake> presently we run the precommit script in the tools directory 20:05:20 <sdake> jpeeler has also been working on our first functional test 20:05:25 <sdake> jpeeler mind giving us an update on that 20:06:04 <jpeeler> i posted some code last week, but i'm going to have to change the approach 20:06:19 <sdake> I noticed you are launching inside a heat vm 20:06:23 <jpeeler> i thought it would be nice to have the test code also run in a container, but if that's not worth the trouble 20:06:26 <sdake> have you thought szabout just running on the baremetal we get? 20:07:22 <jpeeler> sdake: i figured the test code would run the same place the containers did 20:07:33 <sdake> right, but aren't you running it in a vm now? 20:08:27 <jpeeler> yes, i thought the devenv in the source was the expected environment 20:08:36 <sdake> that is jsut for developers to test with 20:08:38 <sdake> not for the gate 20:08:55 <sdake> vm on vm will be too slow for us to gate with is the reason I bring it up 20:09:00 <sdake> according to #openstack-iunfra core guys 20:09:04 <sdake> they said our approach wont work well 20:09:13 <sdake> but running container on the virt machine we get would be fine 20:09:18 <jpeeler> well it's really container on vm :) 20:09:30 <sdake> container on vm, on the vm that infra gets 20:09:36 <sdake> infnra doesn't actually allocate bare metal machines 20:09:40 <jpeeler> ah i see 20:09:49 <sdake> and my understanding from #openstack-iunfra is virt on virt in their environments is terrible :( 20:10:03 <sdake> so I think a simple approach would be build container, run test case, output result 20:10:23 <sdake> anyway nice work there - I like what I saw this morning when i reviewed the patch 20:10:51 <sdake> anyone else on this topic? 20:11:20 <jpeeler> just make sure everybody knows links are not available for us 20:11:36 <sdake> because of the --pid=host problem you mean? 20:11:52 <jpeeler> i think it's the --net option actually 20:11:57 <sdake> yup 20:12:00 <sdake> that is what I meant :) 20:12:07 * sdake brain malfunctions increasing :( 20:12:22 <sdake> #topic milestone #3 progress 20:13:24 <sdake> #link https://blueprints.launchpad.net/kolla 20:13:41 <sdake> we have ea slew of completed blueprints alrready this cycle 20:13:47 <sdake> unfortunately the link above doesn't show what we have done 20:13:53 <sdake> so it could look kind of depressing 20:14:02 <sdake> but we have done about as many as are currently outstanding 20:14:20 <sdake> lets just focus on the essentials for now and go over them 20:14:33 <sdake> #link https://blueprints.launchpad.net/kolla/+spec/add-first-functional-test 20:14:41 <sdake> #info first functional test is almost complete 20:14:50 <daneyon_> I need to implement the feedback from 159701 20:14:52 <sdake> yay nice work jpeeler and nikolay !:) 20:15:01 <daneyon_> #link https://review.openstack.org/#/c/159701/ 20:15:07 <sdake> #link https://blueprints.launchpad.net/kolla/+spec/create-dev-quickstart 20:15:19 <sdake> ryan any progress on the quickstart? 20:15:40 <daneyon_> I have to rebase #link https://review.openstack.org/#/c/159004/ 20:15:49 <rhallisey> ya I've been standing up openstack with a bunch of docker runs 20:16:01 <rhallisey> but I'm going to move to fig 20:16:09 <daneyon_> I'll get these 2 addressed later this week. 20:16:16 <sdake> daneyon_ nice :) 20:16:23 <rhallisey> sdake, so I should be done soon 20:16:34 <sdake> #info database control and api control almost complete 20:16:46 <sdake> #info ryan has openstack stood up with docker r uns - going to document it 20:17:01 <sdake> rhallisey I'd recommend just getting a real rough quickstar tguide and we can all iterate on it with a bunch of revisions 20:17:19 <rhallisey> sdake, I have my own repo 20:17:19 <sdake> so like cranking something out quickly and getting it merged today/tomorrow - then we have the week to fix it up 20:17:22 <sdake> and individually try it 20:17:23 <rhallisey> I can link it 20:17:30 <sdake> cool drop a link in 20:17:38 <rhallisey> #link https://github.com/rthallisey/atomic-osp-installer 20:17:52 <rhallisey> that has a decent guide to get up and running 20:18:10 <rhallisey> sdake, by the way I got nova fully working :P 20:18:22 <rhallisey> I wasn't sure if you have gotten there yet 20:18:23 <sdake> nice, nova in master works as well 20:18:30 <sdake> atleast it wfm 20:18:32 <rhallisey> excellent 20:18:34 <sdake> (nova-network) 20:18:37 <sdake> did you get neutron rolling? 20:18:42 <rhallisey> were you running it with glance? 20:18:46 <sdake> if you guys have code sitting around get reviews in 20:18:47 <rhallisey> no just nova-network 20:18:49 <sdake> I used glance from devstack 20:18:58 <sdake> only ran nova-compute on from containers 20:19:06 <rhallisey> ok. I have glance + keystone + nova working together 20:19:15 <sdake> cool can you get reviews up if the current master is bust? 20:19:21 <rhallisey> ya sure 20:19:30 <sdake> that way we are wall working from the same code base rather then divergent code bases :) 20:19:31 <rhallisey> ccrouch's patch is needed 20:19:34 <rhallisey> for glance 20:19:41 <sdake> I haven't seen ccrouch's patch 20:19:54 <sdake> ok well looks like we are in good shape for the next week of tasks, everyone has work to do 20:19:54 <rhallisey> we are going to run into a lot of race conditions 20:20:03 <sdake> #topic lxc vs docker 20:20:14 <sdake> daneyon have any thoughts on leading this agenda item 20:21:08 <sdake> daneyon must havestepped away :) 20:21:27 <sdake> essentially I think tripleo may not be too interested atm in integrating container work into their code base 20:21:30 <sdake> in the short term 20:21:42 <sdake> so we had a look around other projects and came across os-ansible-deployment 20:21:50 <sdake> that project uses lxc for containers 20:21:57 <sdake> and does container deployment 20:22:07 <sdake> using ansible rather then tripelo 20:22:23 <sdake> at some point it may (or may not) make sense totry to integrate wtih that project 20:22:40 <sdake> since tripleo's future is unclear 20:23:00 <sdake> there are two options - get them to use docker, or get us to use lxc 20:23:09 <sdake> for the moment we are sticking with docker 20:23:21 <sdake> but I wsanted to open it up to the broader community, folks have any thoughts on this integration path? 20:24:25 <rhallisey> why do you think tripleo isn't interested? 20:24:32 <jpeeler> would have to look closer at os-ansible-deployment - does it support knowing when a container is up rather than just started? 20:24:51 <sdake> jpeeler I dont know, daneyon may know though 20:25:09 <sdake> rhallisey kolla uses shell scripts, I think tripleo is more interested in a puppet deployment model 20:25:17 <sdake> persoanlly I think puppet is overkill 20:25:34 <sdake> these are just things to think about for our longer term roadmap 20:26:50 <sdake> I can't see us switching entirely to a puppet model 20:26:53 <sdake> folks agree or disagree? 20:28:02 <jpeeler> if we're considering ansible, puppet doesn't sound crazy 20:28:16 <jpeeler> the real question is long-term do we want to merge with tripleO I guess 20:28:19 <sdake> we aren't using ansible interlaly 20:28:24 <sdake> ansible would deploy our containers 20:28:44 <sdake> so instead of merging with tripleo, it would be merge with os-ansible-deployment 20:30:36 <sdake> well something to think about - seems like people are thinking rather then typing :) 20:30:49 <jpeeler> forgive me for asking, but what is kolla's goal again? 20:30:50 <sdake> #topic open items 20:30:59 <sdake> #undo 20:31:00 <openstack> Removing item from minutes: <ircmeeting.items.Topic object at 0x9384990> 20:31:03 <sdake> jpeeler that is a great question 20:31:12 <sdake> the answer is "bring container deployment to openstack" 20:31:45 <sdake> we initially chose tripleo as our upstream to bring that to 20:32:15 <sdake> another answer woudl be "deploy openstack in containers" 20:32:28 <sdake> we have drawn a line at the "deploy" part - as in that is someone elses business 20:32:34 <sdake> I don't think w e want to do that step ourselves 20:32:41 <sdake> but we have to chose the upstream we want to integrate with 20:32:51 <sdake> I just can't see an easy way to integrate with tripleo 20:33:02 <sdake> I can see alot of easy ways to integrate with os-ansible-deployment 20:33:20 <sdake> path of least resistance and all that ;) 20:33:21 <sdake> make sense? 20:33:44 <jpeeler> it does - i'm just not sure we'll have the adoption we were looking for without tripleO 20:34:14 <sdake> yup so your question is - if we integrated with os-ansible-deployment would anyone actually use our stuff 20:34:25 <sdake> my answer to that is, that community is going to start seeing some serious growth 20:34:49 <sdake> which implies yes 20:35:11 <jpeeler> okay then - but that also means no more fig then right? 20:35:17 <sdake> who knows ;-) 20:35:29 <sdake> lets wait until os-ansible-deployment grows as I think it will 20:35:34 <sdake> before we make a commitment 20:35:40 <sdake> but keep it on our minds 20:36:16 <sdake> I do think there is room for fig in that model, but again they use lxc currently 20:36:18 <sdake> rather then docker 20:36:25 <sdake> #topic open items 20:36:39 <sdake> any open discussion folks kwould like to have? 20:36:58 <rhallisey> sdake, where do you want me to put the dev quickstart? 20:37:14 <sdake> docs directory 20:41:04 <sdake> ok well thanks folks 20:41:13 <sdake> see you all next week - lets keep cranking on those blueprints and get em knocked out :) 20:41:15 <sdake> #endmeeting