15:01:14 #startmeeting ansible-sig 15:01:15 Meeting started Tue Feb 18 15:01:14 2020 UTC and is due to finish in 60 minutes. The chair is sshnaidm. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:01:16 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:01:18 The meeting name has been set to 'ansible_sig' 15:01:31 who is available today? 15:01:34 o/ 15:02:08 o/ 15:02:38 dtantsur, odyssey4me jrosser mnaser cloudnull just fyi, openstack ansible modules meeting started 15:02:42 o/ 15:03:32 agenda is: 15:03:35 #link https://etherpad.openstack.org/p/openstack-ansible-modules 15:03:49 so from action items: 15:04:18 we'll deal with pushing collection and ptg stuff later this week, as mordred pointed above 15:04:21 o/ 15:05:03 job for build&push collections is in WIP https://review.opendev.org/#/c/708372/ 15:05:08 yah. stupid virtualenv 15:05:10 o/ 15:05:48 re: virtualenv - https://github.com/ansible/ansible/pull/67407 may help make it easier to do better things 15:06:24 thanks odyssey4me 15:06:44 we have also a few reviews pending, please get some in your time 15:06:47 odyssey4me: yah- in fairness to the virtualenv devs, they heard and responded pretty quickly 15:07:04 sshnaidm: ++ reviews 15:07:38 yeah, I think we can proceed as a start with mordred patch, at least it has +2 from me and I saw gtema worked on it 15:08:04 for using a class for openstack module 15:08:24 sshnaidm: cool - I thnik that'll open the door for us to clean and simplify in a bunch of places 15:08:33 #link https://review.opendev.org/#/c/698044/ 15:08:39 mordred, totally 15:09:09 another problem that we hit a few times and should be handled there is matching various options and openstacksdk versions 15:09:13 also - those checks were complex enough that last time I tried to required_if them it didn't work -but maybe required_if is better todauy - or maybe you're smarter :) 15:09:20 yah 15:10:04 mordred, I think it worked for me, maybe it's ok now 15:10:42 sweet 15:10:46 for being more compatible with various openstcksdk versions, I'd propose a job matrix to test code 15:10:55 I totally agree - doing as much with those declarations is nicer 15:10:59 ++ 15:11:10 openstacksdk versions to test: 15:11:10 master 15:11:10 0.36.1 (stable/train) 15:11:10 0.27.0 (stable/stein) 15:11:10 0.17.3 (stable/rocky) 15:11:10 Ansible: 15:11:12 devel 15:11:14 2.9 15:11:16 2.8 15:11:31 as I noticed the minimum required version is 0.12 and it's rocky 15:12:24 yeah - I mean - one definitely needs at least whatever version we did the shade merge from - I think that's 0.12 15:12:28 https://github.com/openstack/ansible-collections-openstack/blob/master/plugins/module_utils/openstack.py#L112 15:12:46 mordred, so it shouldn't work for queens at all? 15:13:00 for queens you'd need to use shade 15:13:02 BUT 15:13:30 it's important to point out that it has never been these modules intentions to release match with openstack releases- master openstacksdk should totally work with queens - or even diablo 15:13:46 so while I do think we shoudl support some older sdks within reason 15:14:05 it's also important that we point out to people in docs that they do _not_ need an older version of something to work with an older cloud 15:14:49 mordred, so you mean we don't want "max_version" 15:14:53 mordred: yeah, the issue has actually been people adding arguments to modules which don't work with the sdk at older versions - for example https://review.opendev.org/708119 is upping the minimum if certain parameters are used 15:14:55 only min_version 15:14:57 (I know we have vendors shipping stuff side-by-side, so running modules with older versions of libraries may be unavoidable in some cases - but it would be really nice if we could get those vendors to update sdk :) ) 15:15:08 sshnaidm: ++ 15:15:25 odyssey4me: ++ 15:15:42 great 15:16:01 #action sshnaidm to point about min and max version in docs 15:16:31 wrt docs 15:16:47 I just copypasted this from ansible: https://review.opendev.org/#/c/704558/ 15:17:09 so if somebody wants to update this doc, please feel free to hijack this patch :) 15:17:20 consider it as a wiki :) 15:17:41 #link https://review.opendev.org/#/c/704558/ 15:17:57 haha 15:18:22 heh, *hint* at ekultails and mordred there :p 15:19:32 well, we need something to show to new contributors, after they are shocked after github->gerrti switch 15:20:25 maybe I'll add there a "transition" section with links how to create gerrit acc, etc 15:20:37 ok, moving on 15:20:40 so we agree on proposed jobs matrix, right? 15:21:18 yes. although we might want to start with ansible-devel being non-voiting - just because sometimes the devel branch can itself be flaky 15:21:48 mordred, yeah, good point, maybe same with master openstacksdk? 15:21:59 or no, it's actually under our control 15:22:12 I think we can run these jobs later on openstacksdk branches and master, to prevent breakages for ansible module actually 15:22:29 or at least fast fixing them 15:22:42 dtantsur, no objections, right? ^ 15:23:20 yeah - master openstacksdk should be fine - it's gated :) 15:23:28 great 15:23:36 ++ 15:23:45 so, this is it for agenda today, please feel free to raise topics for open discussions 15:23:45 I just replied to that docs patch above (and I'll take a stab at a new patch update) - but we should totally do co-gating with openstacksdk 15:24:12 mordred, cool 15:25:35 gtema, thanks for working on tags stuff https://review.opendev.org/#/c/705343/ 15:25:49 I'm quite lost in this tags thing.. 15:26:11 wlcm, still thinking for the better way and would like to ask perhaps not to merge anything tag related 15:26:27 sshnaidm mordred wrt ansible-devel - I suspect that once the collections split out happens, core will be more stable... that said, it might be better for us to do a periodic job for devel rather than gate on it 15:27:35 odyssey4me, from my experience periodic jobs are less important for people 15:27:53 odyssey4me, and it's visible only if someone is looking for it in particular 15:28:03 odyssey4me: yeah - I'd hope it'll be more stable then - and also the things in it that impact *US* should be fairly low 15:28:05 odyssey4me, I'd compromise on non-voting job 15:28:25 gtema, you can always -2 these patches :) 15:28:34 I know 15:29:05 gtema, I'll mention this in ML after a mtg, but you know... 15:29:08 sshnaidm: fair enough - although non-voting jobs tend to be ignored... but at least they're more visible 15:29:38 odyssey4me, yeah, and you can get feedback for your change immediately 15:29:45 mordred: any chance to refer in the zuul job definition project on the different connection? github repo with job definition does require-project in opendev 15:30:13 sorry for missusing channel 15:30:52 gtema: yes - cross-source stuff works perfectly fine 15:31:22 do I need required-projects: -opendev/opendev/refstack-client? 15:31:42 opendev/openstack/refstack-client 15:31:44 ok, thks 15:31:59 opendev.org/openstack/.. I think 15:32:12 yeah - opendev.org/openstack/refstack-client ... you'll need to have those in your zuul - is this in opendev's zuul? 15:32:22 nope, in mine 15:32:38 enabling refstack tests of my cloud 15:32:41 kk. yeah - just make sure you have both repos in the main zuul config 15:32:49 thks 15:33:00 all right 15:33:01 you don[t ACTUALLY need the opendev.org prefix if it's unique 15:33:15 mordred, or, is it just a default? 15:34:02 ok, I think I can finish the meeting and feel free to continue 15:34:05 zuul will figure it out 15:34:14 #endmeeting