16:09:46 <noonedeadpunk> #startmeeting openstack_ansible_meeting
16:09:47 <openstack> Meeting started Tue Nov 26 16:09:46 2019 UTC and is due to finish in 60 minutes.  The chair is noonedeadpunk. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:09:48 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:09:50 <openstack> The meeting name has been set to 'openstack_ansible_meeting'
16:09:56 <noonedeadpunk> #topic office hours
16:11:06 <noonedeadpunk> jrosser: so, am I right, that all blocking upgrade patches have been already placed?
16:11:33 <jrosser> i think we have pushed up everything needed from our side
16:11:38 <noonedeadpunk> nice
16:11:39 <jrosser> but i don't think all merged yet?
16:11:48 <noonedeadpunk> No, it doesn't :(
16:11:50 <jrosser> oh there is documentation / example to fix for placement
16:12:00 <jrosser> the user_config is missing it
16:12:42 <noonedeadpunk> I'd say that we need pretty big effort to looks through our docs and updating them, as this wasn't done for several releases I guess
16:12:46 <jrosser> also these timeouts are bad
16:12:59 <noonedeadpunk> Yeah, centos and upgrade jobs.
16:13:02 <jrosser> i am wondering why we have centos7 jobs on master any more
16:13:16 <jrosser> becasue sometime soon there will be py3 only code and then we have to remove the jobs anyway
16:14:09 <jrosser> but that doesnt help with backports to train though
16:14:11 <noonedeadpunk> I guess because we still don't have centos8?
16:14:34 <noonedeadpunk> And I saw ml regarding that topic
16:14:37 <jrosser> yes, there was some chatter on the ML about that, sharing issues between the deployment projects
16:14:39 <jrosser> snap
16:15:02 <jrosser> i looked again yesterday and still no lsyncd in EPEL
16:15:08 <noonedeadpunk> And yes, dunno how we will be supporting both releases for the upgrade
16:15:53 <jrosser> i was really sad with network-manager too
16:16:42 <jrosser> this is really not cool at all https://review.opendev.org/#/c/689629/10/tests/roles/bootstrap-host/tasks/prepare_networking.yml
16:16:47 <noonedeadpunk> I don't really like the idea of building it from source for centos...
16:17:20 <noonedeadpunk> yeah, I saw that. But we probably can live with it for sime time...
16:18:06 <jrosser> we can maybe get a bit further with an AIO on metal by dropping the lsyncd requirement
16:18:16 <jrosser> obviously multnode repo server will be broken
16:19:20 <noonedeadpunk> we can hardcode to deploy repo only to first server in group in case of centos8 lol
16:20:32 <noonedeadpunk> btw, doesn't python-selinux appeared for centos 7.7?
16:20:37 <noonedeadpunk> *python3
16:20:54 <jrosser> oh maybe, i'm not sure
16:21:30 <noonedeadpunk> seems it's not ;(
16:21:49 * noonedeadpunk never liked centos much
16:23:42 <noonedeadpunk> do you have any thought what to do with upgrade stuff?
16:24:07 <guilhermesp> o/
16:24:11 <noonedeadpunk> it's problematic and helpfull at the same time...
16:24:52 <guilhermesp> do we have a topic come in handy for things that still needs to be merged for upgrades ?
16:25:29 <noonedeadpunk> looks like we don't
16:26:37 <jrosser> noonedeadpunk: regarding upgrades i don't know - theres also py2/3 to throw into the mix
16:27:29 <noonedeadpunk> yeah, but I think that s->t tests this migration at least for ubuntu. But I saw some weirdness in logs...
16:27:53 <jrosser> right - i'm maybe not surprised, this all needs more work really
16:28:08 <jrosser> perhaps the best thing to do for now is to concentrate on getting further with centos-8 on master
16:28:13 <noonedeadpunk> like ansible has probably cached ansible_python_interpreter or smth like that
16:28:21 <jrosser> then we understand what the scale of any backporting is
16:28:33 <noonedeadpunk> yeah, agree
16:28:34 <jrosser> noonedeadpunk: yes i think you are right - i've seen something similar
16:28:55 <jrosser> perhaps a setup of deleting the /etc/openstack_deploy facts cache at the midpoint of the upgrade would address that
16:29:01 <noonedeadpunk> so the -main- biggest problem with centos 8 is lsycnd?
16:29:19 <jrosser> well i don't know - it was the point at which i gave up :)
16:29:36 <jrosser> several days cursing networkmanager and i'd had enough
16:29:39 <noonedeadpunk> ok, by far at least:)
16:29:57 <noonedeadpunk> I can totally understand you - I'd probably gave up on it actually
16:30:48 <jrosser> it's probably worth another look - trouble is i'm super busy for a few weeks
16:33:00 <miloa> jrosser: perhaps not usefull but about  https://review.opendev.org/#/c/689629/10/tests/roles/bootstrap-host/tasks/prepare_networking.yml in order to prepare network we are using an ansible playbook using role "michaelrigart.interfaces"
16:33:30 <jrosser> noonedeadpunk: also when i last looked there was no RDO repo at all for Centos-8, thats also a big showstopper
16:33:42 <jrosser> i don't know if thats changed at all now
16:33:46 <noonedeadpunk> I think the problem is that ansible mosdules can not work with centos 8 now
16:34:27 <jrosser> noonedeadpunk: whats broken there?
16:35:19 <jrosser> miloa: one of the troubles i had was that networkmanager doesnt seem to have a concept of a dummy interface..... did you make those work?
16:35:26 <noonedeadpunk> I think rdo is not ready as well :(((( https://trunk.rdoproject.org/
16:35:54 <noonedeadpunk> jrosser: disregard my opinion about networking
16:36:51 <noonedeadpunk> anyway, seems like we will be backporting stuff after releasing...
16:39:07 <jrosser> i made a topic for stuff we need to merge https://review.opendev.org/#/q/topic:osa-train+(status:open+OR+status:merged)
16:39:10 <jrosser> guilhermesp: ^
16:41:43 <guilhermesp> nice jrosser thanks
16:41:45 <guilhermesp> looking now
16:42:00 <jrosser> its all needing review, not just merge
16:43:28 <miloa> jrosser: I did not try with type dummy.
17:01:58 <noonedeadpunk> #endmeeting