16:02:06 <noonedeadpunk> #startmeeting openstack_ansible_meeting
16:02:07 <openstack> Meeting started Tue Jan 28 16:02:06 2020 UTC and is due to finish in 60 minutes.  The chair is noonedeadpunk. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:02:08 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:02:10 <openstack> The meeting name has been set to 'openstack_ansible_meeting'
16:02:16 <noonedeadpunk> #topic office hours
16:02:27 <jrosser> o/
16:02:37 <noonedeadpunk> o/
16:04:01 <noonedeadpunk> sorry I wasn't really helpful with tempest issue we've faced with - was really concentrated on manila stuff. Regarding it's testing - centos and ubuntu metal jobs are passing. LXC is still in progress....
16:04:22 <noonedeadpunk> Debian ceph deployment is pretty much broken as well - so it's also pending
16:04:55 <jrosser> ooi how do the VM talk to the nfs gateways there
16:05:10 <jrosser> i'm interested in how the multitenancy works
16:05:47 <noonedeadpunk> in metal is not an issue, but in prod it will ssh to ganesha host
16:05:59 <noonedeadpunk> (this is tbd as well actually)
16:06:48 <noonedeadpunk> so manila-share will need it's ssh keypair to access nfs hosts
16:07:35 <noonedeadpunk> I think this part is also required for lxc and just started to work on it...
16:07:57 <noonedeadpunk> (but tempest fails with completely another issue)
16:08:52 <noonedeadpunk> as I can see temest stuff is pretty much done? except ubuntu mess?
16:11:05 <jrosser> i think done might be anoverstatement :(
16:11:30 <jrosser> rocky seems unaffected
16:11:45 <jrosser> stein looks fixed if my patches will merge
16:11:51 <noonedeadpunk> repo server saved us at least once? :p
16:12:10 <jrosser> train is waiting for whatever works out on master regarding ubuntu/distro
16:13:10 <jrosser> i have a patch here for ubuntu/distro to roll back all the "mixed" tempest installs any always use a virtualenv https://review.opendev.org/#/c/704567/
16:13:30 <noonedeadpunk> just commented it
16:13:44 <noonedeadpunk> I like the approach - just small nit there
16:13:49 <jrosser> oops
16:14:29 <openstackgerrit> Jonathan Rosser proposed openstack/openstack-ansible-os_tempest master: [WIP] Always use virtualenv and pip to install tempest on Ubuntu  https://review.opendev.org/704567
16:18:46 <openstackgerrit> Merged openstack/openstack-ansible-openstack_hosts master: Switch to Ussuri UCA repo  https://review.opendev.org/704317
16:19:55 <noonedeadpunk> jrosser: hm, how master bump has passed ci if tempest is broken?
16:20:06 <jrosser> oh thats fun
16:20:18 <jrosser> i think it all starts to go sideways if you apply u-c
16:20:53 <jrosser> which we were not doing, and was the ultimate cause of neutron-lib being too new
16:21:14 <jrosser> but once you start doing that all sorts of other stuff breaks
16:21:36 <noonedeadpunk> I see...
16:21:41 <jrosser> that brings out the unsitability of the distro python module versions which previously we were just getting away with
16:22:17 <noonedeadpunk> maybe we should break u-c for mixed setup? :p
16:23:03 <jrosser> i think thats what we agreed was the "bad idea", becasue mixed setup still put the tempest plugins on the host with pip
16:23:25 <jrosser> well yes, it's all horrible whichever way :)
16:23:31 <noonedeadpunk> yeah, that;s not good idea
16:23:41 <noonedeadpunk> ++
16:24:05 <jrosser> i think just "tempest is in a venv" is going to make this all simpler, and we could extend that to all the distro installs if we wanted
16:24:26 <jrosser> it's just acute on ubuntu because of the lack of packages in UCA for tempest plugins
16:24:36 <noonedeadpunk> yep, sure - we will just 2 ways of deploying tempest instead of 3
16:25:11 <noonedeadpunk> so there will be defenitely less mess there
16:28:20 <noonedeadpunk> oh, btw, once bump will be placed I'm going to release OSA on these bumps, without waiting 2 weeks - ike we have lots of things really fixed with them (like tempest for S, neutron for T) Maybe only rocky will be released as usual
16:28:32 <noonedeadpunk> Is anyone have any thought n this?
16:29:29 <noonedeadpunk> also - do we want to backport py3 centos to train?
16:29:39 <noonedeadpunk> it's a bit messy but seems to work in master
16:30:27 <noonedeadpunk> and we have broken update of docs - probably our periodic jobs are not working? Not sure about this, but docs are not updated...
16:31:38 <noonedeadpunk> Like https://docs.openstack.org/project-deploy-guide/openstack-ansible/rocky/deploymenthost.html#install-the-source-and-dependencies was last updated on 2019-08-12 but I think it's supposed to update tags at least?
16:33:58 <jrosser> py3 centos to train might be a good thing(tm) in order to ease a centos7>8 transition release
16:34:11 <jrosser> but i've got no idea how that will work at all right now
16:34:59 <noonedeadpunk> I can recall that rdo packages for 8 are going to be ready for U only? So probably transition release will be U anyway...
16:35:24 <noonedeadpunk> Right now ansible works under py2 while it deploys py3 venvs
16:35:44 <noonedeadpunk> since libeselinux is still not ready for py3 in centos 7
16:36:05 <noonedeadpunk> so kind of "mixed" thing again....
16:37:33 <noonedeadpunk> I think it;s full list of patches for master for this https://review.opendev.org/#/q/topic:osa_py3_centos+(status:open+OR+status:merged)
16:38:15 <noonedeadpunk> patches for nova and masakari are really nasty....
16:38:26 <noonedeadpunk> (especially for nova)
16:39:51 <noonedeadpunk> and maybe I've missed some role, which requires either libvirt-python or libguestfs...
17:16:01 <noonedeadpunk> #endmeeting