16:00:27 <noonedeadpunk> #startmeeting openstack_ansible_meeting
16:00:28 <openstack> Meeting started Tue Sep 22 16:00:27 2020 UTC and is due to finish in 60 minutes.  The chair is noonedeadpunk. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:29 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:00:31 <openstack> The meeting name has been set to 'openstack_ansible_meeting'
16:00:36 <noonedeadpunk> #topic office hours
16:00:43 <noonedeadpunk> \o/
16:00:48 <jrosser> o/
16:02:34 <noonedeadpunk> so, regarding what we want to end up with for V. I think usage of ansible 2.10, mariadb 10.5 and cleaning up centos and suse might be really enough
16:02:52 <noonedeadpunk> (maybe add new ceph release as well)
16:03:11 <noonedeadpunk> galera is painfull and broken for centos 8 right now...
16:03:26 <openstackgerrit> Arx Cruz proposed openstack/openstack-ansible-os_tempest master: Fix tempest init logic  https://review.opendev.org/753393
16:03:26 <noonedeadpunk> (eventually it's broken even on 10.4 so)
16:04:00 <noonedeadpunk> I hope I will have time for fixing this up during this week
16:04:16 <noonedeadpunk> and jrosserdid really awesome job for ansible 2.10
16:05:16 <noonedeadpunk> another thing to raise is PTG
16:05:42 <openstackgerrit> Arx Cruz proposed openstack/openstack-ansible-os_tempest master: Fix tempest init logic  https://review.opendev.org/753393
16:05:47 <arxcruz> noonedeadpunk: ^
16:05:53 <arxcruz> fully idempotent
16:05:59 <arxcruz> and working properly
16:06:07 <arxcruz> jrosser: ^
16:06:25 <arxcruz> oh, sorry, didn't new it was meeting time
16:06:44 <jrosser> noonedeadpunk: i nearly got /etc/openstack_deploy/user-collection-requirements.yml done now
16:07:11 <jrosser> just made small mistake where ....... json_query ....... is in a ....... collection :( DOH
16:07:17 <noonedeadpunk> That is so awesome! I actually also start looking into that, so thanks for letting me know)
16:07:49 <jrosser> need to find an alterntive with ansible built ins, should be OK
16:07:51 <noonedeadpunk> yeah, I think we will be missing some collections for a while tbh
16:08:10 <noonedeadpunk> as CI doesn;t cover everything we have
16:08:24 <noonedeadpunk> and the most interesting things are not default actually
16:08:33 <jrosser> here is my WIP http://paste.openstack.org/show/798226/
16:08:35 <noonedeadpunk> (like lets encrypt)
16:10:30 <noonedeadpunk> jrosser: doesn't from_yaml load to native list of dict?
16:10:34 <jrosser> then i would also like to look at pre/post playbok hooks
16:10:53 <noonedeadpunk> an we can use map?
16:11:49 <jrosser> yes, i think that json_query is just more readable (came from the user roles stuff) but map will do the same in this case
16:12:08 <noonedeadpunk> user_collections | map(attribute=name) | list ?
16:13:48 <noonedeadpunk> can't find json_query collection...
16:14:19 <noonedeadpunk> but yeah it's neat and in some cases we can hardly find an alternative
16:14:33 <jrosser> it's in community.general
16:14:51 <noonedeadpunk> ah, lol, I got the issue
16:14:56 <noonedeadpunk> haha
16:15:03 <noonedeadpunk> yeah, that's nice one
16:15:23 <noonedeadpunk> yeah, cool.
16:15:35 <jrosser> fg
16:15:38 <jrosser> argh
16:16:25 <jrosser> did you have any thoughts on pre/post playbook hooks?
16:17:19 <admin0> jrosser,  .. even with -vvvv i can't seem to figure out what its exactly trying to do .. https://gist.github.com/a1git/6edc27b686550bf2701f381746a2c944  ..
16:17:43 <noonedeadpunk> I'm not eactly sure how it would behave if there would be no files - I think we can workaround that
16:17:57 <noonedeadpunk> but we can do that if needed
16:17:58 <admin0> haproxy.stat is present in all 3 controllers with root:root and srw  permission
16:18:36 <noonedeadpunk> I would also probably add some mine customizations there then
16:19:25 <jrosser> i've not tried anything yet, but was going to look at adding a common_tasks file
16:20:25 <jrosser> so we can include_tasks that in pre_tasks / post_tasks of the existing playbooks, and pass some var for /etc/openstack_deploy/pre_hooks/<blah>/
16:20:27 <noonedeadpunk> I don't have much time this week unfortunatelly, so won't be able to try out
16:20:54 <jrosser> then use with_fileglob | sort to include playbooks in alphanumeric order
16:20:57 <jrosser> or something
16:20:59 <jrosser> :)
16:21:15 <admin0> i am going to # all lines there and see what it does
16:21:29 <noonedeadpunk> I mean would be great to separate them per specific role
16:21:36 <admin0> since  the services are already setup, its already enabled
16:21:57 <noonedeadpunk> and eventually so that they were placed on custom location as well....
16:22:27 <noonedeadpunk> As for instance I have repo with roles and playbooks, which are separated with some patches per role
16:22:53 <noonedeadpunk> I ran them after all roles, but if they were included by plays, that would be awesome
16:23:08 <jrosser> oh yes, custom location, good idea
16:23:38 <noonedeadpunk> I mean maybe we can define path for pre/post hook with some variable and run them only if it's defined?
16:24:50 <noonedeadpunk> not sure, needs thinking through
16:25:18 <jrosser> sure, i'll try to spend some time on this tomorrow for POC
16:25:21 <watersj> jrosser, thanks was push i needed. haproxy failing start because cinder-api not starting, because, i changed ips. in short my fault. I need to re-ip my storage network, other than changing openstack_inventory.json file, what else should I touch 1st. looks like a lot
16:25:56 <jrosser> watersj: that is strange - haproxy should generally always start even if the services are down
16:26:00 <watersj> cinder, glance atleast
16:26:04 <noonedeadpunk> jrosser: if you're busy, I can try looking into it next week as well
16:26:20 <jrosser> noonedeadpunk: this week is OK - i have reasonable time for OSA stuff
16:26:35 <jrosser> but like you very concerned with galera :(
16:26:44 <noonedeadpunk> yeah
16:26:56 <noonedeadpunk> I just feel it may take _some_ time
16:27:09 <noonedeadpunk> ok, great
16:27:41 <jrosser> i can maybe ask andrewbonney to also look at that, it's a good match problem
16:28:12 <noonedeadpunk> so returning back to ptg, I booked slots for Monday, October 26 from 13 UTC till 17 UTC and October 27 from 21-00 UTC till 23-00 UTC
16:28:26 <noonedeadpunk> so everybody is warmly welcome
16:28:47 <noonedeadpunk> as well as input and ideas for W cycle
16:29:07 <jrosser> we should start an etherpad
16:29:16 <noonedeadpunk> Was just writing that
16:29:31 <jrosser> need to do a bit of planning for finishing V too, decide what we try to tackle
16:29:34 <noonedeadpunk> https://etherpad.opendev.org/p/osa-wallaby-ptg ?
16:29:40 <jrosser> bind-to-mgmt still so close :/
16:30:01 <noonedeadpunk> we should finish it
16:30:06 <noonedeadpunk> it's really too close
16:30:48 <noonedeadpunk> I will fill in with current status of things, so we could decide what should be moved to another cycle
16:31:22 <noonedeadpunk> #link https://etherpad.opendev.org/p/osa-wallaby-ptg
16:31:36 <noonedeadpunk> btw, PTG schedule is available here
16:31:39 <noonedeadpunk> #link https://ethercalc.openstack.org/7xp2pcbh1ncb
16:34:51 <admin0> oh well .. now my vip is not coming up :D
16:38:05 <noonedeadpunk> I guess that's it for me:)
16:38:55 <noonedeadpunk> There's also a bug that I'd probably want to discuss, but have pretty low energy for that...
16:39:19 <noonedeadpunk> https://bugs.launchpad.net/bugs/1896589
16:39:20 <openstack> Launchpad bug 1896589 in openstack-ansible "main interface renamed from ens160 to eth0 after AIO VM restart" [Undecided,New]
16:41:00 <jrosser> hmm yes it does rename it in the AIO host prep i think?
16:41:09 <noonedeadpunk> yep it did
16:41:25 <noonedeadpunk> and yeah, it's not working after reboots which is annoying
16:42:50 <noonedeadpunk> so I guess it's confirmed, but don't have aio on hands I can reboot
16:43:28 <jrosser> i think that mine might boot as eth0 in the first place?
16:43:57 <noonedeadpunk> um, so mine booted with ens3 and was not changed instantly
16:44:09 <noonedeadpunk> but yeah, I suggest that after reboot it might change naming
16:44:21 <jrosser> actually you are right, mine here is ens3 too
16:44:37 <jrosser> but then again i kind of never reboot them
16:44:59 <noonedeadpunk> yeah mee too.
16:45:16 <noonedeadpunk> I got used to that we can't rely on aio, but that's probably not really right
16:45:38 <noonedeadpunk> and yeah http://paste.openstack.org/show/798227/
16:45:44 <noonedeadpunk> tons of configuration errors
16:46:15 <noonedeadpunk> so ansible-role-systemd_networkd  worth having a look
16:46:25 <jrosser> oh my, yes
16:46:47 <noonedeadpunk> I guess that's it https://opendev.org/openstack/ansible-role-systemd_networkd/src/branch/master/templates/systemd-link.j2
16:47:00 <noonedeadpunk> but again that might depend on systmed version...
16:48:11 <noonedeadpunk> will also try to pickup that then this week
16:48:49 <noonedeadpunk> but actually reboot will still be impossible I guess due to the way we mount loop devices...
16:49:12 <jrosser> i have a feeling we do that with systemd units, and this has come up before
16:49:29 <jrosser> there was some patches to change the dependancy of those to try to get the ordering correct at a reboot
16:50:10 <noonedeadpunk> we're doing it here https://opendev.org/openstack/openstack-ansible/src/branch/master/tests/roles/bootstrap-host/tasks/prepare_networking.yml#L37
16:57:22 <noonedeadpunk> #endmeeting