16:00:03 #startmeeting openstack_ansible_meeting 16:00:04 Meeting started Tue May 7 16:00:03 2019 UTC and is due to finish in 60 minutes. The chair is mnaser. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:05 #topic office hours 16:00:06 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:00:07 o/ 16:00:08 The meeting name has been set to 'openstack_ansible_meeting' 16:00:13 o/ 16:00:29 hey noonedeadpunk how's it goin 16:00:32 you were missed at the summit :)_ 16:01:17 Yeah, I was reading yours twitter and was really kinda jealous about missing it... 16:01:56 * noonedeadpunk looking for flights to Shanghai 16:03:04 btw extra congrats with superuser awards:) 16:05:34 so, what I've missed regarding train plans and new collaborations?:) 16:08:23 thanks noonedeadpunk :D 16:08:27 did you check out the ether pad? 16:10:11 oh, it was updated. Will look trough it today 16:10:19 yep! 16:10:25 tons of good discussion 16:11:29 Yeah, I was pretty excited about a letter from kolla-ansible team, regarding collaboration. 16:16:07 personally I feel, that containers are good thing, and we probably shouldn't fully drop their support 16:17:43 o/ 16:22:25 yep, agreed on not dropping machine container support. the migration path would be painful for a lot of deployers. switching the default deployment to metal seems like a reasonable thing, but it'll be interesting to review a spec about that 16:26:01 yeah 16:26:27 logan-: noonedeadpunk I wonder if there would be a way to handle machine container management outside OSA 16:27:01 the first step to that I think is to define the group structure in a yaml inventory and shift it out of the dynamic inventory 16:28:09 ansible-inventory can be useful for analyzing how we currently structure it, but it outputs pretty garbage inventory structures for complex inventories, so I don't think you'd want to just dump ansible-inventory yaml output to shim out groups and call it a day 16:28:49 but getting the group tree as a base inventory layer, and then adding hosts on top of that with additional inventories (whether metal or container) is a good direction imo 16:29:42 logan-: yeah I think we need a 'defined' inventory api so to speak 16:29:58 yep 16:33:59 we have a lot of work, but I'm not sure how much of it we can get done this cycle 16:34:03 most folks seem busy 16:35:16 yep, etherpad seems really filled 16:35:31 yeah. and this inventory thing has been a topic for idk 3-4 ptgs in a row now. it seems no one has time to go fishing around that mess of dynamic inventory we have. 16:39:33 the problem of baremetal is that distro packages of different services may actually have conflicts, which are hard to resolve... 16:48:37 would this be a good time to see if https://docs.openstack.org/openstack-ansible/latest/admin/scale-environment.html could be reviewed, I've been attempting to add a new shared infra node node and encountered several issues 16:55:08 noonedeadpunk: how so? 16:57:25 So I have one non-OSA deployment, installed from packages on Ubuntu. And all services are without containers on bare metal. And touching some python package usually results in upgrading everything. 16:58:24 So in if you would like just to add some new service to the node? or just update single python package, you should be ready for the full upgrade 17:00:58 so I don't feel that bare metal distro installs are really good idea, as they are pretty hard to maintain, imo 17:14:11 Yeah bare metal distro is an exploding upgrade 17:14:16 That’s just part of that decision 17:14:28 #endmeeting