15:04:57 <sshnaidm> #startmeeting ansible-sig
15:04:58 <openstack> Meeting started Tue Mar 24 15:04:57 2020 UTC and is due to finish in 60 minutes.  The chair is sshnaidm. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:05:00 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:05:02 <openstack> The meeting name has been set to 'ansible_sig'
15:05:06 <sshnaidm> sorry, my bad
15:05:28 <sshnaidm> ok, so who is still available for the meeting? :)
15:05:39 <gtema_> me for 10%
15:06:10 <sshnaidm> #link https://etherpad.openstack.org/p/openstack-ansible-modules
15:06:14 <sshnaidm> agenda today ^
15:06:45 <sshnaidm> dtantsur, mordred odyssey4me and all
15:06:55 <dtantsur> o/
15:08:05 <sshnaidm> firstly I'd like to ask for reviews where it's possible: https://review.opendev.org/#/q/topic:collection-jobs+(status:open+OR+status:merged)
15:08:35 <sshnaidm> seems like train and rocky on openstacksdk are broken
15:08:46 <sshnaidm> pep8 failures and neutron tests
15:09:03 <dtantsur> there have been a lot of transient issues today
15:09:06 <dtantsur> make sure it's not them
15:09:19 <sshnaidm> dtantsur, it's from a few days earlier
15:09:36 <sshnaidm> but I'll rebase patches, just in case it helps..
15:10:15 <sshnaidm> when this is merged, I think we're good about coverage in SDK and Ansible branches
15:10:57 <sshnaidm> next jobs it's testing collections with Ansible 2.8, which will require a little trick, cause you can't build collection in 2.8, only can use it :)
15:11:27 <sshnaidm> and job to run unittests that currently exist in collection, I think we ignored them so far
15:11:46 <sshnaidm> hopefully it will encourage people to write them more..
15:12:22 <sshnaidm> any comments so far?
15:12:47 <sshnaidm> moving on
15:12:52 <sshnaidm> * Keep extracting common code to library and standartization of modules
15:13:24 <sshnaidm> like mordred has started with a general class for Openstack Ansible module
15:13:26 <odyssey4me> o/
15:13:31 <sshnaidm> odyssey4me, hey
15:14:03 <sshnaidm> it's a way we need to go imho, to try pulling out everything Ansible specific from modules to library
15:14:35 <sshnaidm> to make module writing easier for people unfamiliar with Ansible, modules, etc
15:15:06 <tremble> AnsibleAWSModule makes things much easier for AWS modules.
15:15:29 <sshnaidm> tremble, right, the same idea with OpenstackAnsible class
15:15:38 <dtantsur> I agree, it's hugely important
15:15:55 <dtantsur> esp if we see a growth of contributors with little ansible experience
15:15:56 <dtantsur> (like me)
15:16:30 <sshnaidm> dtantsur, and I'm afraid we won't see much if we don't do it :)
15:16:45 <dtantsur> exactly
15:17:28 <sshnaidm> I don't have specific steps here, so point to think and to experiment in patches till next meeting
15:17:57 <sshnaidm> maybe even a spec, not sure though if we need it in this stage
15:18:38 <sshnaidm> ok, I think we all agree on that, moving on
15:19:01 <sshnaidm> *    Pushing collection
15:19:12 <sshnaidm> for that we need mordred, so skipping for now
15:19:20 <sshnaidm> * Work on a new job from scratch for modules, trying to reduce the time.
15:19:50 <sshnaidm> back to CI, I'm not so familiar with devstack jobs, but wonder if it's possible to make it shorter somehow..
15:20:03 <sshnaidm> or maybe to set up something from scratch
15:20:20 <sshnaidm> currently it's quite long and spamming a lot in logs
15:20:59 <sshnaidm> moving on
15:21:07 <sshnaidm> * Work on a separate jobs for specific modules for more focused and optimal testing
15:21:19 <sshnaidm> just an idea to have module-specific jobs
15:21:44 <sshnaidm> and reduce load, for example if change is in neutron, maybe to skip glance modules check
15:22:29 <sshnaidm> and sub-question, how to test ironic modules
15:22:32 <sshnaidm> dtantsur, ^^
15:23:00 <sshnaidm> we'll need real bm nodes for functional tests
15:23:20 <dtantsur> nope
15:23:21 <sshnaidm> is it possible on third-party only or we can do something in openstack infra
15:23:26 <dtantsur> ironic has a virtual CI
15:23:35 <dtantsur> we even have a job on openstacksdk
15:23:43 <dtantsur> we can either use our devstack plugin OR bifrost
15:23:47 <sshnaidm> dtantsur, nodes as virtual bms?
15:23:53 <dtantsur> yep
15:24:17 <sshnaidm> dtantsur, so we can run introspection on them?
15:24:26 <dtantsur> yes, we can
15:24:36 <dtantsur> you'll need to add the ironic-inspector devstack plugin to the mix
15:24:49 <sshnaidm> interesting, that's good news
15:24:55 <dtantsur> this is without inspector: https://github.com/openstack/openstacksdk/blob/master/.zuul.yaml#L230-L277
15:25:49 <sshnaidm> dtantsur, and openstacksdk-functional-devstack-ironic runs actually with virtual nodes?
15:26:21 <dtantsur> yep, although I don't think we actually use them in the tests
15:26:36 <dtantsur> (we use the fake driver in SDK tests, I think)
15:26:59 <sshnaidm> dtantsur, ok, seems like we'll need to take it offline to some specific ironic modules talk
15:27:18 <dtantsur> yeah
15:27:27 <sshnaidm> I'm close to finish with BM modules for tripleo, so we'd better to start thinking adapting them for collection
15:27:32 <dtantsur> indeed
15:27:42 <sshnaidm> cool
15:27:56 <sshnaidm> * Adding TripleO jobs and testing
15:28:20 <sshnaidm> well, this one is too early to talk about maybe, but also good to plan ahead
15:28:50 <sshnaidm> as we use in tripleo Keystone modules, but it's taken from Ansible repo
15:29:05 <sshnaidm> all because mordred doesn't push collection :D
15:29:10 <dtantsur> that mordred!
15:29:15 <sshnaidm> yeah
15:29:47 <dtantsur> anyway, this should be solved within a few weeks
15:30:04 <dtantsur> then we can just add jobs as usual: first non-voting, then consistently green, then maybe voting
15:30:15 * dtantsur has reservations about tripleo jobs being consistently green
15:30:20 <sshnaidm> dtantsur, yeah, that's the plan
15:30:50 <sshnaidm> dtantsur, well, we have standalone jobs that should fit our goals here
15:31:05 <dtantsur> ideally, we'll need something with ironic
15:31:09 <dtantsur> but that can wait
15:31:10 <sshnaidm> and 3-party OVB jobs for ironic
15:31:14 <dtantsur> right
15:31:21 <sshnaidm> but they're loooong
15:31:45 <dtantsur> we can limit them to patches touching ironic
15:31:52 <sshnaidm> yeah
15:32:11 <sshnaidm> and it has pretty good coverage of ironic there
15:32:22 <dtantsur> and we'll add bifrost jobs
15:32:32 <dtantsur> that is, when somebody actually updates bifrost to use the collections
15:32:46 <dtantsur> i.e. when it's too late and we get broken
15:32:55 <sshnaidm> I'll need to look more about bifrost
15:33:09 <dtantsur> sshnaidm: that's essentially where tripleo is moving :)
15:33:12 <sshnaidm> does it use current os_ironic modules?
15:33:18 <dtantsur> ironic + ansible (optionally keystone)
15:33:20 <dtantsur> yes
15:33:25 <sshnaidm> dtantsur, oh, you mean novaless and metalsmith, etc?
15:33:33 <dtantsur> on tripleo side - yes
15:33:39 <dtantsur> bifrost has been like that since inception
15:34:12 <sshnaidm> ok, so I think Steve Baker is actually did it for tripleo
15:34:18 <dtantsur> yep
15:34:24 <sshnaidm> in Ussuri we have it by default iirc
15:34:28 <dtantsur> actually, I think the ironic modules have been developed for bifrost
15:34:31 <dtantsur> initially
15:34:34 <dtantsur> yeah
15:34:53 <sshnaidm> so you mean in tripleo we use actually os_ironic* modules?
15:34:58 <dtantsur> nope, in bifrost
15:35:05 <sshnaidm> ah, ok
15:35:29 <dtantsur> I guess we should move on from the ironic topics :)
15:35:45 <sshnaidm> dtantsur, yeah, so it seems like important to have these jobs on os_ironic..
15:35:49 <sshnaidm> dtantsur, ok
15:35:55 <sshnaidm> I think I finished with today agenda actually
15:36:05 <sshnaidm> so now free talk if anyone has
15:36:16 <dtantsur> I stayed till the end, do I receive a treat? :D
15:36:53 <sshnaidm> dtantsur, of course
15:36:57 <dtantsur> \o/
15:37:27 <sshnaidm> first ironic ansible module goes to dtantsur !
15:37:42 * dtantsur bites the first ironic ansible module
15:38:19 <sshnaidm> dtantsur, don't scare him, will run away..
15:38:24 <sshnaidm> ok, so let's continue talks here if there are any
15:38:30 <sshnaidm> #endmeeting