16:03:38 <noonedeadpunk> #startmeeting openstack_ansible_meeting
16:03:39 <openstack> Meeting started Tue May 12 16:03:38 2020 UTC and is due to finish in 60 minutes.  The chair is noonedeadpunk. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:03:40 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:03:42 <openstack> The meeting name has been set to 'openstack_ansible_meeting'
16:03:45 <noonedeadpunk> #topic office hours
16:04:24 <openstackgerrit> Jonathan Rosser proposed openstack/ansible-role-uwsgi master: Do not use bare variables in ansible conditionals  https://review.opendev.org/727258
16:05:08 <noonedeadpunk> jrosser: so, I see you've started tough way of cleaning up to the new ansible versions?)
16:05:25 <noonedeadpunk> ok, I think we need to branch this week for sure
16:05:42 <noonedeadpunk> the only solid thing left - bump of mariadb
16:05:49 <jrosser> noonedeadpunk: yes i just have been working top to bottom on the log from an integrated repo build fixing as many warnings as i can
16:06:07 <jrosser> i think that we should try to merge the mariadb client and server roles
16:06:14 <noonedeadpunk> and eventually we want this to be merged as well
16:06:15 <jrosser> that would be a really good improvement
16:06:25 <noonedeadpunk> like a lot of effort but clean run is nice)
16:06:42 <noonedeadpunk> Yeah, ok, let's do it
16:06:55 <noonedeadpunk> I recall we've stopped discussion on how we're going to use it
16:06:59 <jrosser> i was thinking it was not so difficult - personally i would just need a good block of time to write the code nicely rather than just hack it
16:07:31 <jrosser> yeah, there was discussion about what the inputs would be to install client/server/both
16:08:58 <jrosser> i was quite liking keeping main.yml to install the server, with defaults vars install_server=true install_client=false
16:09:10 <noonedeadpunk> So if we're merging than we should drop client role I think
16:09:22 <jrosser> then a playbook specifically for the client which could be called with tasks_from: install_client.yml
16:09:23 <noonedeadpunk> Yeah
16:09:31 <jrosser> to replace places where currently we just use the galera_client role
16:09:39 <jrosser> then it would not have to skip all the server tasks
16:09:50 <noonedeadpunk> yep, really great idea
16:09:54 <noonedeadpunk> wath think about the same
16:10:00 <noonedeadpunk> *was thinking
16:10:14 <jrosser> i also though we can factor out the tasks which install the repos
16:10:35 <jrosser> and set a fact "repos_installed=true" so that we avoid doing that multiple times when installing both client and server
16:11:08 <jrosser> becasue the main.yml and install_client.yml code paths will both need to include something like install_repos.yml
16:11:22 <noonedeadpunk> actually we can set repo installation out of client/server install
16:11:44 <noonedeadpunk> like do preparation steps out of scope as they're mostly the same I guess
16:12:02 <noonedeadpunk> (or at least do this only for matching things)
16:12:11 <noonedeadpunk> and move the rest to install
16:13:14 <noonedeadpunk> I mean in main.yml do include 3 things - common_pre steps, client_install, server_install
16:13:54 <noonedeadpunk> Just why do we dublicate code
16:15:08 <jrosser> [WARNING]: No MPM module loaded! apache2 reload AND other module actions will fail if no MPM module is loaded immediately.
16:15:19 <noonedeadpunk> ah, saw that
16:15:29 <jrosser> i wonder what it really means
16:15:31 <noonedeadpunk> but actually there's mpm loaded...
16:15:38 <jrosser> yeah, it's odd
16:15:55 <noonedeadpunk> maybe because it's loaded implicitly
16:16:11 <noonedeadpunk> and it expects it to be enabled explicitly
16:19:41 <openstackgerrit> Jonathan Rosser proposed openstack/openstack-ansible-os_tempest master: Fix deprecated os_project_facts ansible module usage  https://review.opendev.org/727264
16:21:20 <jrosser> phew right thats got to the end of the warnings from one job
16:21:38 <jrosser> it think there are more ansible os_<foo>_facts usages that need fixing
16:22:42 <jrosser> noonedeadpunk: did you see i replied to the github issue about the new pip resolver?
16:22:59 <noonedeadpunk> not really. let me check
16:23:43 <jrosser> its a bit, errm, 8-O
16:24:45 <noonedeadpunk> yeah
16:24:59 <noonedeadpunk> So as I said we probably can just use requirements isntead...
16:25:07 <noonedeadpunk> but not 100% sure
16:25:34 <jrosser> i guess that warrants some kind of POC before the pip stuff is cast in stone
16:26:07 <noonedeadpunk> yep and it's nice that we bump pip version :p
16:28:32 <noonedeadpunk> actually that will even simplify things as we'll have less stuff
16:28:48 <noonedeadpunk> as all we have in source-contraints would be in https://opendev.org/openstack/ansible-role-python_venv_build/src/branch/master/tasks/python_venv_wheel_build.yml#L79-L86
16:29:00 <noonedeadpunk> But I'd rather leave this to handle for V?
16:29:15 <jrosser> absolutely :)
16:29:39 <jrosser> i was a bit surprised to see that our source constraints file it acutally includes "--constraint <u-c url>"
16:29:46 <jrosser> inside the constraints file
16:30:08 <noonedeadpunk> huh...
16:30:11 <noonedeadpunk> yeah...
16:30:25 <noonedeadpunk> we defenitely need to look through what we do there
16:31:16 <jrosser> the other thing i wanted to bring up was this patch https://review.opendev.org/#/c/720705/
16:31:23 <jrosser> chandankumar: are you around?
16:31:48 <noonedeadpunk> was just looking at it right before meeting
16:32:02 <jrosser> i made a depends-on just to prove beyond doubt that it does break everything for OSA https://review.opendev.org/#/c/727067/
16:32:07 <jrosser> and indeed it does
16:32:48 <chandankumar> jrosser, Hello
16:32:54 <chandankumar> jrosser, anything I can help
16:32:58 <jrosser> o/ hello there
16:33:16 <jrosser> just wanted to ask if you were across what was happening with this patch https://review.opendev.org/#/c/720705/
16:33:53 <openstackgerrit> Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_glance master: Cleanup after repo_build and pip_install retirement  https://review.opendev.org/727269
16:34:08 <jrosser> the issue i see there is that the patch brings a bunch of assumptions that it's being used in tripleo, like OVS, and the switch names and so on
16:34:26 <chandankumar> jrosser, yes, I am aware of that, the goal was to enable and run ironic test using OSA
16:34:48 <chandankumar> using os_tempest
16:35:12 <chandankumar> jrosser, May be I can revisit the patch and make it more generic for osa and tripleo
16:35:31 <jrosser> i made a depends-on job for the OSA ironic role https://review.opendev.org/#/c/727067/
16:36:03 <jrosser> and i think that highlights what needs working on, like the OSA test is using linuxbridge rather than OVS principally
16:36:28 <jrosser> so i wondered really where the line should be for stuff that should be in os_tempest, and stuff that should not
16:38:51 <chandankumar> jrosser, yes, you are correct, May be we can start by adding to job on osa side with ironic test enabled and see what is required for running that then we can decide what we can keep in os_tempest side
16:39:25 <jrosser> yeah, becasue it would be really great to also get the OSA ironic test actually doing something real, i'm not sure its actually testing anything at the moment
16:40:08 <chandankumar> akahat, hello ^^ can you start adding a job on https://opendev.org/openstack/openstack-ansible-os_ironic
16:40:23 <jrosser> chandankumar: anyway we should not try to fix it all here and now i just wanted to highlight that we need to find a middle ground somewhere
16:40:45 <chandankumar> to enable ironic test
16:41:25 <jrosser> if nothing else to make sure that any code we commit from each side does not break the other, so adding the OSA ironic test to os_tempest would be ++
16:41:51 <jrosser> jamesdenton: bjoernt ^ this is of interest to you guys i think
16:42:21 <chandankumar> jrosser, yup on tripleo side we started with this test https://opendev.org/openstack/tripleo-quickstart/src/branch/master/config/general_config/featureset060.yml#L43 as a first goal
16:42:30 <jamesdenton> thanks jroll
16:42:36 <jamesdenton> thanks jrosser
16:43:05 <chandankumar> may be enabling the same on OSA side and re-iterating over that to get this test running ironic_tempest_plugin.tests.scenario.test_baremetal_multitenancy.BaremetalMultitenancy.test_baremetal_multitenancy will help
16:43:30 <jroll> any time :P
16:43:35 <jamesdenton> :)
16:44:38 <chandankumar> jrosser, akahat will be working on ironic tests, let's see how it goes.
16:44:47 <noonedeadpunk> btw will be there U packages in rdo?
16:44:56 <noonedeadpunk> for Centos 7 I mean
16:44:57 <chandankumar> *ironic tests
16:45:17 <chandankumar> noonedeadpunk, https://trunk.rdoproject.org/centos8-ussuri/
16:45:39 <noonedeadpunk> ok so only Centos*
16:45:43 <noonedeadpunk> ok so only Centos8
16:45:46 <chandankumar> noonedeadpunk, it is now available
16:45:55 <chandankumar> noonedeadpunk, for train only we have c7 and c8
16:46:19 <chandankumar> noonedeadpunk, https://trunk.rdoproject.org/centos8-train/
16:46:42 <noonedeadpunk> chandankumar: yeah we just still didn't manage to add it's support
16:47:07 <noonedeadpunk> mgariepy: have you any progress while playing with centos8?
16:47:43 <chandankumar> noonedeadpunk, if you hit any c8 package related issue on rdo side, please poke us or #tripleo people
16:48:05 <noonedeadpunk> chandankumar: I think we stuck a bit before that:(
16:48:10 <akahat> chandankumar, jrosser o/ noted.
16:50:42 <noonedeadpunk> maybe we'll have to drop centos distro deploys for U for a while.... until we implement and backport centos 8...
16:51:09 <chandankumar> noonedeadpunk, any specific issue we can look, may be yatin can help
16:51:11 <jrosser> that could work - i'm a bit surprised that the centos-7 source installs are still working
16:51:30 <jrosser> i think i'm expecting some too-old-libvirt or something to happen at some point for centos7
16:51:50 <noonedeadpunk> but maybe it will be not for U....
16:52:18 <jrosser> chandankumar: i think primarily we had issues with translating stuff that had worked prevously with yum to dnf
16:52:28 <jrosser> like repo priorities
16:53:01 <jrosser> the documentation did not seem to match what actually happened when i tried it
16:53:12 <chandankumar> jrosser, https://fedoraproject.org/wiki/Yum_to_DNF_Cheatsheet is pretty much good
16:56:51 <openstackgerrit> Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible master: Drop proprietary repo usage for libselinux-python3  https://review.opendev.org/727273
16:58:26 <akahat> q
16:59:47 <noonedeadpunk> I can try to see what I can do about it. As I think jrosser is more valuable at galera stuff at the moment?
17:01:47 <jrosser> there is quite a lot of code review to do atm mnaser guilhermesp if you have some time
17:02:15 <jrosser> we are actually merging / fixing things for a change which is great!
17:02:46 <jrosser> spotz: also if you have some time for some code review....
17:03:30 <guilhermesp> sure jrosser i will find out a slot asap
17:05:34 <jrosser> also other good news is that we are almost complete on the py2.7 goal https://review.opendev.org/#/q/topic:drop-py27-support+(status:open)
17:06:19 <jrosser> just rsyslog_server remains there from OSA side
17:12:04 <noonedeadpunk> that's great!
17:12:09 <noonedeadpunk> #endmeeting