15:00:11 #startmeeting ansible-sig 15:00:12 Meeting started Tue Mar 31 15:00:11 2020 UTC and is due to finish in 60 minutes. The chair is sshnaidm. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:13 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:15 The meeting name has been set to 'ansible_sig' 15:00:25 hi all, who is available today? 15:01:03 * cyberpear eavesdrops 15:01:10 * mordred waves 15:01:20 o/ 15:02:12 ok, let's move on 15:02:34 no specific agenda for today, I'd like to give some updates from Ansible 15:02:57 I've been on Sunday Ansible contributors virtual mtg, we talked about collections 15:03:33 so all community modules should be part of ACD (Ansible community distribution), which will be released much more frequently than half-year release of Ansible 15:03:50 how it will sync with all releases of various modules is not clear yet 15:04:06 an openstack-style coordinated release? 15:04:32 dtantsur: it seems the process is that they pull in the collections when they build ACD tarball 15:04:34 dtantsur, that's unclear currently what will be a cadence 15:04:46 yeah, kind of 15:05:02 I'll post lately some links with WIP ideas from Ansible folks 15:05:02 * mordred has a crazy thought related to releases for us to come back to in a bit 15:05:05 ++ 15:05:34 once the actual ACD build process is nailed down - I'd like to run at least some part of the ACD build in our jobs so we can test that we're still working with what they're expecting 15:06:23 mordred, yeah, it seems reasonable 15:06:41 I hope other modules won't break ours and this is independent system 15:07:19 but building ACD with our latest code seems a good test for me 15:07:50 just need to understand how things are taken into ACD 15:08:16 yup 15:08:26 which I think means it needs to be defined first - since that's still WIP for them ;) 15:08:28 and I must to admit openstack ansible modules are in good shape unlike many other modules.. 15:08:38 mordred, yep 15:09:05 at least from CI and tests perspective 15:09:22 which takes I to next point 15:09:36 I think we're all covered in all branches except of rocky 15:09:56 Rocky is EM, right? 15:10:08 EM..? 15:10:12 extended maint 15:10:14 extended maintenance 15:10:19 alive-but-not-really 15:10:27 and yeah - I don't really think people shold be doing much with rocky openstacksdk 15:10:39 (I don't really think they should be doing much with train openstacksdk - but I know I'm going to lose that battle) 15:10:49 :D 15:11:01 I'm using it because Queens is too old. 15:11:04 true vikings consume only master 15:11:28 well - just as a quick reminder for everyone who might not know - latest sdk should ALWAYS work with *all* older versions of openstack 15:11:34 and RH doesn't ship any newer SDK than Rocky for RHEL7 15:11:40 it is not necessary to install an sdk matching your version of cloud 15:11:51 cyberpear: yeah - that I can't help with - and is a fair reason to use it ;) 15:12:04 while rocky is blocked for incoming patches including breaking patches, I'm fine :) 15:12:14 but better to settle it 15:12:26 it's some python deps hell there 15:12:33 nothing serious 15:12:54 yeah- I am not opposed to fixing those 15:13:10 having CI jobs able to work is not a bad thing, even if it's in EM 15:14:06 "not a bad thing" :) 15:14:45 and for testing new features of SDK with current modules I pass openstacksdk version to test roles: https://review.opendev.org/#/c/715620/ 15:15:04 please use it for now, until we come with a better solution 15:15:07 ++ 15:15:47 mordred, can we have something pushed to galaxy already? like 0.x version 15:16:13 I'm starting to look at building/pushing jobs 15:16:39 so maybe we can set up releasing procedure in the near future 15:17:17 but I think it's fine to push it manually for now 15:17:33 we do 15:17:40 already there 15:17:46 oh, great 15:18:06 https://galaxy.ansible.com/openstack/cloud 15:18:38 which version is it? 15:18:43 it's been a bit shaky - there's been issues with the tarball built being inconsistent - so we should make sure we're producing consistent tarballs before we tag anything 1.0 15:18:49 0.0.1-dev83 15:19:56 ok, so we need to add "--pre" like: ansible-galaxy collection install openstack.cloud --pre 15:20:13 until it has release 1.* 15:21:17 so that's it for me for now 15:21:32 any topics to talk, folks? 15:22:20 * dtantsur cannot think of anything 15:22:49 * mordred had one - but just spoke with sivel and turns out it may not be a thing 15:23:02 mordred, what was it? 15:23:04 stop intriguing us :) 15:23:11 but the tl;dr is that it's possible to install an ansible_collection as part of the install of a python library 15:23:17 ansible scans sys.path 15:23:21 via pip install? 15:23:28 so I was pondering the idea of bundling the collection with sdk 15:23:30 BUT 15:23:43 it's a purposefully undocumented feature and is intended to only be there to support ACD 15:23:53 so I think it's best to stay away from that for now 15:24:05 good, undocumented features are never abused by curious people 15:24:17 * dtantsur may be overly sarcastic today 15:24:21 :) 15:24:26 oh yes, you are 15:24:57 nice! I was afraid for a second that I'm always overly sarcastic. 15:25:02 btw, if you want to install it from git, it's like: git clone https://github.com/namespace/collection ~/.ansible/collections/ansible_collections/namespace/collection 15:25:31 and it should work if you have plain structure in your repo 15:26:02 yeah 15:26:38 * dtantsur wonders why collections/ansible_collections, are there any other sort of collections in ansible? 15:27:02 nope, it's a known issue 15:27:10 * cyberpear also annoyed by it 15:27:26 seems more like philosophical question 15:28:07 ok, so if no more questions, I think we can end the meeting 15:28:15 woot 15:28:16 #endofmeeting 15:28:23 #endmeeting