16:00:24 <strigazi> #startmeeting containers 16:00:24 <openstack> Meeting started Tue Apr 11 16:00:24 2017 UTC and is due to finish in 60 minutes. The chair is strigazi. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:25 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:00:28 <openstack> The meeting name has been set to 'containers' 16:00:34 <strigazi> #topic Roll Call 16:01:16 <Drago1> o/ Hi strigazi! 16:01:25 <tonanhngo> Ton Ngo 16:01:28 <jasond> o/ 16:01:30 <strigazi> hi Drago1 16:01:44 <coreyob> o/ 16:02:20 <juggler> o/ 16:02:30 <strigazi> Thanks for joining the meeting Drago1 tonanhngo jasond strigazi coreyob juggler 16:02:33 <juggler> hello strigazi! 16:03:05 <strigazi> hi juggler 16:03:06 <strigazi> #topic Announcements 16:03:46 <strigazi> None 16:03:55 <strigazi> Team, do you have any? 16:04:18 <juggler> currently, none here 16:04:31 <strigazi> #topic Essential Blueprints Review 16:05:12 <strigazi> 1. Support baremetal container clusters (strigazi) 16:05:15 <strigazi> #link https://review.openstack.org/#/c/395012/ 16:05:26 <strigazi> No, wrong url 16:05:41 <strigazi> https://blueprints.launchpad.net/magnum/+spec/flatten-attributes this is ti 16:06:12 <strigazi> please have look in the review, I'm a bit stuck in the update operation 16:07:15 <strigazi> Do you have any question about it? 16:08:02 <tonanhngo> Which review do you mean? 16:08:19 <juggler> stuck in what respect? 16:08:21 <strigazi> I'm thinking in doing it in two steps like the contruct and expand now to move forward faster. 16:08:36 <strigazi> This review https://review.openstack.org/#/c/395012/ 16:09:39 <strigazi> I can't get the update right since the API object is different than the rpc object 16:10:45 <tonanhngo> ok, will take a close look 16:11:10 <juggler> thanks tonanhngo for assisting s 16:11:32 <strigazi> ok, I was thinking expanding the tables with all attributes in a patch and the reduce them again in a followup 16:12:44 <strigazi> We can follow it up later 16:12:50 <strigazi> next is: 16:13:19 <strigazi> #link https://blueprints.launchpad.net/magnum/+spec/nodegroups Nodegroups Drago1 16:13:23 <Drago1> I don't have any update. Jaycen didn't send me any update either. 16:13:32 <strigazi> ok 16:13:56 <strigazi> Finally is #link https://blueprints.launchpad.net/magnum/+spec/cluster-upgrades Cluster 16:15:13 <strigazi> I did a prototype with node replacement but without finishing the first bp I can't move forward. 16:16:06 <strigazi> doing rebuild is very fast. I'm aiming doing a demo before the summit 16:16:47 <strigazi> That's it. 16:17:21 <strigazi> #topic Other blueprints/Bugs/Reviews/Ideas 16:17:33 <strigazi> I have a serious one 16:17:48 <strigazi> https://bugs.launchpad.net/magnum/+bug/1680900 16:17:49 <openstack> Launchpad bug 1680900 in Magnum "Reduce size of software-configs in user-data" [High,New] 16:18:13 <strigazi> Currently, we pass all our scripts in user-data 16:18:36 <strigazi> that field in a mariadb or mysql is 64kb 16:19:07 <strigazi> we are near 63kb 16:19:13 <Drago1> We started using software deployments 16:19:22 <Drago1> In our downstream driver 16:19:42 <strigazi> with the heat-agents? 16:20:10 <Drago1> yes 16:20:28 <Drago1> works well 16:21:49 <strigazi> ok, so if we need software deployments we need them as well. 16:22:40 <strigazi> OS allows golang now, we can write them in golang :) 16:23:06 <tonanhngo> Yay 16:23:16 <strigazi> Drago1 You use a custom image I suppose? 16:23:21 <randallburt> uh, those two things aren't quite related but yay 16:23:26 <Drago1> Yes, we're using fedora cloud 24 16:23:39 <Drago1> non-atomic, with the agents baked in 16:23:51 <juggler> golang, excellent! 16:24:09 <Drago1> randallburt: they're related in the sense that non-python docker images are way smaller 16:24:28 <strigazi> I asked what is the status and the reply was feel free to contribute :) 16:24:33 <randallburt> so you want to rewrite all the agents on top of including software deployments? 16:25:55 <strigazi> I would be nice, but it is a massive task. 16:26:55 <strigazi> Let's go to open discussion 16:26:56 <strigazi> #topic Open Discussion 16:27:49 <Drago1> Possibly naive, but to get what we want out of software deployments, is it that hard? I mean, could it be written in bash in the meantime? 16:27:52 <Drago1> randallburt: ^ 16:28:03 <strigazi> Team, is anyone using swarm-mode? I don't have time for the functional tests, I need some help 16:28:23 <Drago1> (not signing up for it, by the way) 16:29:42 <randallburt> Drago1: but to get "what we want" out of software deployments we don't have to rewrite anything but our templates 16:29:53 <randallburt> Drago1: so I don't see the point (at least at this stage) 16:30:28 <randallburt> strigazi: sorry, but I don't know anyone using swarm atm 16:30:33 <Drago1> randallburt: If you're talking about running the agents in a container, pulling/loading is too slow in the gate 16:31:00 <Drago1> And we didn't really get anywhere with having the agents run on the host 16:31:07 <randallburt> Drago1: which we asked to have solved with pre-loaded images in atomic 16:31:35 <jasond> Drago1: i thought we were going to try using the host OS's python installation 16:31:45 <Drago1> randallburt: Unless the images are actually already loaded in docker at boot time, loading is still too slow 16:31:59 <Drago1> jasond: well we never figured out how to make that work 16:32:27 <jasond> Drago1: https://etherpad.openstack.org/p/magnum-ptg-pike-upstream-linux-os 16:32:35 <randallburt> Drago1 jasond well Adrian was going to play with it, I don't remember hearing how that went 16:33:11 <Drago1> Oh okay, so containerized, but just binding all the things into the container so we don't have to include python in the image 16:35:42 <strigazi> One more thing, have you tried cinder with k8s again? It works for us with k8s 1.5.2 16:35:53 <Drago1> randallburt: ^ 16:36:00 <strigazi> even dynamic claim works 16:36:27 <strigazi> with patch: https://review.openstack.org/#/c/445404/ 16:37:08 <Drago1> We're on 1.5.4 and from what I remember randallburt saying, it didn't work for us, so we were going to test 1.6 16:37:09 <strigazi> and claim like: http://clouddocs.web.cern.ch/clouddocs/containers/tutorials/cinder.html 16:37:37 <strigazi> the above examples work with k8s 1.5.2 ^^ 16:38:45 <strigazi> I guess 1.5.4 as well 16:39:45 <Drago1> I haven't personally done any testing so I can't say 16:40:20 <strigazi> ok 16:40:26 <randallburt> I haven't tried recently but will today 16:40:51 <strigazi> randallburt be sure to use the same kubectl client version 16:41:02 <strigazi> server 1.5 and client 1.6 won't work 16:42:03 <randallburt> oh I bet the udev ln fixes it! 16:42:21 <randallburt> cool, thanks strigazi 16:42:39 <strigazi> Team, we wrap it early. Is there anything else? 16:42:48 <strigazi> You are welcome randallburt 16:43:03 <Drago1> Nice 16:43:13 <juggler> good here. thx strigazi for presiding 16:43:54 <strigazi> see you all next week! 16:43:56 <strigazi> #endmeeting