| *** sshnaidm|off is now known as sshnaidm | 08:11 | |
| *** gtema has joined #openstack-ansible-sig | 08:20 | |
| *** dtantsur|afk is now known as dtantsur | 10:02 | |
| *** gtema has quit IRC | 10:03 | |
| *** gtema has joined #openstack-ansible-sig | 10:35 | |
| *** gtema has quit IRC | 11:53 | |
| *** gtema has joined #openstack-ansible-sig | 12:30 | |
| *** gtema has quit IRC | 12:35 | |
| *** gtema has joined #openstack-ansible-sig | 13:06 | |
| *** gtema has quit IRC | 13:11 | |
| *** zbr has quit IRC | 13:27 | |
| *** zbr has joined #openstack-ansible-sig | 13:32 | |
| *** gtema has joined #openstack-ansible-sig | 13:54 | |
| *** johanssone has quit IRC | 15:48 | |
| *** johanssone has joined #openstack-ansible-sig | 15:49 | |
| *** priteau has joined #openstack-ansible-sig | 17:05 | |
| *** gtema has quit IRC | 17:14 | |
| mordred | sshnaidm, dtantsur, odyssey4me: WIP Make an OpenStackModule base class https://review.opendev.org/698044 | 17:22 |
|---|---|---|
| dtantsur | \o/ | 17:22 |
| mordred | I also put in a patch to announce patches in this channel | 17:22 |
| mordred | that's an initial stab at an idea - put it up for thoughts from folks | 17:22 |
| dtantsur | mordred: do you plan on using the same base for newer modules? because I suspect the auth code there needs rework | 17:24 |
| dtantsur | (at least it failed miserably for me the last time I tried using anything with ironic) | 17:24 |
| mordred | dtantsur: yeah - except I think we should fix auth :) | 17:24 |
| dtantsur | cool! I need to retry my testing with latest ansible. | 17:25 |
| mordred | dtantsur: or - maybe we can even further refactor so that some modules can override an auth method for when that makes sense or something | 17:25 |
| mordred | before we get too far down the road with that refactor we need to port in test jobs so we can see what we're breaking :) | 17:26 |
| sshnaidm | mordred, thanks | 17:26 |
| dtantsur | yep | 17:26 |
| sshnaidm | mordred, I think we have somewhere this job running.. will check | 17:26 |
| mordred | sshnaidm: yah - we've got a job in sdk that does some basic ansible module tests | 17:29 |
| mordred | I think that can likely be copy-pastated | 17:29 |
| mordred | and thne improved, because it's super simple | 17:29 |
| mordred | but is likely a good starting point? | 17:29 |
| sshnaidm | mordred, yeah, at least something | 17:30 |
| sshnaidm | mordred, will look to enable linters as well | 17:30 |
| mordred | (once we get further, we shoud obviously add bifrost and openstack-ansible and tripleo jobs - but let's wait until we've renamed) | 17:30 |
| mordred | sshnaidm: ++ | 17:30 |
| mordred | sshnaidm: also - we can actually hacking/pep8 these now :) | 17:30 |
| sshnaidm | mordred, yeah.. maybe worth to add a whole structure for collections with .keep files? | 17:31 |
| sshnaidm | so that people know where to submit files | 17:31 |
| *** dtantsur is now known as dtantsur|afk | 17:33 | |
| mordred | yah. and docs. and release notes. :) | 17:34 |
| sshnaidm | omg | 17:34 |
| sshnaidm | and swags | 17:36 |
| sshnaidm | I think choosing a mascot should be the first priority | 17:37 |
| *** gtema has joined #openstack-ansible-sig | 17:49 | |
| *** gtema has quit IRC | 17:54 | |
| *** priteau has quit IRC | 17:54 | |
| *** priteau has joined #openstack-ansible-sig | 17:55 | |
| mordred | sshnaidm: I'm making a patch to at least run pep8 on things so that we can have a gating job | 17:55 |
| mordred | whie we work on the other stuff | 17:56 |
| sshnaidm | mordred, cool, thanks | 17:56 |
| *** priteau has quit IRC | 17:58 | |
| sshnaidm | I think we need to expand it a little, at least to people signed in in etherpad: https://review.opendev.org/#/admin/groups/2063,members | 17:59 |
| sshnaidm | mnaser, can we do it? ^ | 17:59 |
| mnaser | sshnaidm: yeah i'm for that. i added you and ill let you use your discretion :) | 18:01 |
| sshnaidm | mnaser, thanks! | 18:01 |
| *** gtema has joined #openstack-ansible-sig | 18:28 | |
| *** gtema has quit IRC | 18:33 | |
| *** gtema has joined #openstack-ansible-sig | 19:06 | |
| *** gtema has quit IRC | 19:11 | |
| *** gtema has joined #openstack-ansible-sig | 19:17 | |
| sshnaidm | mordred, I think it's worth to hold on moving modules now, let's wait until namespace and publishing is ready, so we can delete them from ansible immediately. Otherwise we can step into sync hell, to sync patches from github to repo | 19:18 |
| mordred | sshnaidm: that will never be a possibility | 19:19 |
| mordred | sshnaidm: we can *freeze* them upstream once namespace and publishing is ready | 19:19 |
| mordred | but I don't think we can just delete them can we? | 19:20 |
| sshnaidm | mordred, we delete them and symlink to namespace in galaxy, what will stay there is only 2.9 version | 19:20 |
| sshnaidm | mordred, so people still use it w/o knowing it's deleted, as I understood from gundalow | 19:21 |
| mordred | they'd need to install the collection from galaxy no? | 19:21 |
| mordred | but - gundalow is smarter than I am | 19:21 |
| * gundalow waves | 19:21 | |
| gundalow | and reads | 19:21 |
| sshnaidm | mordred, yeah, he mentioned symlink and I thought they shouldn't | 19:21 |
| *** gtema has quit IRC | 19:22 | |
| gundalow | mordred: once the collection is up and running (ie installable via `ansible-galaxy collection install FOO.BAR`) you can raise a single PR for ansible/ansible which `git rm`'s all the files your moved to collection 2) adds `migrated_to: namespace.collection` in BOTMETA.yml | 19:23 |
| mordred | ah - gotit | 19:23 |
| sshnaidm | gundalow, how do I request a namespace btw? I hope "openstack" is available | 19:24 |
| gundalow | sshnaidm: https://github.com/ansible/galaxy/issues/new/choose | 19:25 |
| sshnaidm | gundalow, ack, thanks! | 19:25 |
| mordred | I've got openstack already | 19:26 |
| mordred | https://galaxy.ansible.com/openstack | 19:26 |
| mordred | so I thnk we should be ok on that front | 19:26 |
| mordred | gundalow: ansible-galaxy collection install openstack.cloud will result in modiule FOO in that collection being available as openstack.cloud.FOO now? or simply as FOO? (I think that was still up in the air a bit when we spoke in atlanta) | 19:27 |
| sshnaidm | mordred, oh, nice | 19:28 |
| sshnaidm | mordred, I think we talked about openstack.service_type in the mtg | 19:29 |
| mordred | sshnaidm: I don't think there is any value in having more than one collection | 19:29 |
| mordred | people want "the openstack modules" not "the openstack.compute" modules | 19:30 |
| mordred | and they all have the exact same depend - so there's not really much value in installing only some of them, no? | 19:30 |
| sshnaidm | mordred, well, as I see we have already a stuff in openstack namespace | 19:31 |
| sshnaidm | various things | 19:32 |
| sshnaidm | so maybe really worth to separate them | 19:32 |
| mordred | those are all roles | 19:32 |
| mordred | auto-imported from any repo named "ansible-role-$foo" on github - but almost certainly quite stale as that process has never worked well for us :) | 19:32 |
| mordred | (given that we don't really use github) | 19:32 |
| *** Shrews has joined #openstack-ansible-sig | 19:33 | |
| mordred | so what we should really do is delete what's there - it certainly shouldn't be depended on by anyone since the plumbing is not maintained (those exist there as a result of me doing investigations as to whether things worked) | 19:33 |
| sshnaidm | mordred, I see.. from user perspective I agree it's better to install one collection, although it may go with a lot of unnecessary requirements | 19:34 |
| mordred | then - making additional collections for roles and modules for different things like openstack-ansible or tripleo seem completely reasonable | 19:34 |
| gundalow | mordred: If you can always use the FQCN (fully qualified collection name), though if you've set your `collection:` keyword to include `openstack,cloud`, then you can just use `module_name` in your Playbooks | 19:34 |
| mordred | there is one and only one requirement for any openstack ansible module in th emodule space - and that's openstacksdk | 19:34 |
| mordred | all of them depend on it | 19:34 |
| mordred | and none depend on anything else | 19:34 |
| mordred | gundalow: when you say "set your collection: keyword" - who is your there? is that a person installing from galaxy? or is that us in our galaxy.yaml file? | 19:35 |
| sshnaidm | mordred, it's now :) | 19:35 |
| sshnaidm | mordred, it's users I think | 19:35 |
| mordred | cool | 19:35 |
| sshnaidm | it's kinda "import" | 19:36 |
| gundalow | https://docs.ansible.com/ansible/latest/user_guide/collections_using.html#using-collections-in-a-playbook | 19:36 |
| gundalow | > To avoid a lot of typing, you can use the collections keyword added in Ansible 2.8: | 19:36 |
| gundalow | yup, `import` is a good analogy | 19:36 |
| gundalow | they = playbook writers | 19:36 |
| mordred | ah - perfect. that's exactly how I'd want that to work | 19:36 |
| sshnaidm | mordred, by deleting "all there" you meant in openstack namespace in galaxy? | 19:37 |
| mordred | sshnaidm: so - just to be clear from earlier - I think there should be one and only one collection that contains the equiv of the current os_ modules - and any new similar ones we grow - and all of those should be built on top of sdk and not any other libraries to keep everything streamlined. BUT - I think there could certainly be additional collections that contain other things that are more | 19:38 |
| mordred | specific (and probably more operator focused) | 19:38 |
| mordred | sshnaidm: yes - those roles in the openstack namespace in galaxy are only there accidentally and by happenstance, not by any actual supported process | 19:39 |
| mordred | we can just leave them and ignore them like we have been doing of course | 19:39 |
| mordred | but I do worry that someone is going to start using them and then get confused if they don't update or something similar | 19:39 |
| sshnaidm | mordred, well, it's up to you, maybe better to clean if they aren't used | 19:40 |
| mordred | official openstack releases of software depending on github is bad thing. but - that's where collections make this awesome - since we can publish directly from zuul - so ay roles that need to be published can go that route | 19:40 |
| mordred | sshnaidm: yah. that's sort of where I was leaning - but should probably make sure with odyssey4me and mnaser that nobody is using those anyway | 19:41 |
| mnaser | hmm | 19:42 |
| mnaser | mordred, sshnaidm cant we just add a warning? | 19:42 |
| mordred | for which thing? | 19:43 |
| mordred | the galaxy roles? | 19:43 |
| mordred | I just double checked - anything that's there got imported 2 years ago and never updated | 19:43 |
| sshnaidm | we can delete them and listen if there are yellings.. | 19:44 |
| sshnaidm | mordred, all that seems reasonable to me, though I'd like to raise up versioning and naming in Thu meeting again, to have more acknowledgment from various projects | 19:44 |
| mordred | ++ | 19:45 |
| sshnaidm | mordred, I think maybe it worth to move them keeping the history? | 19:45 |
| mordred | yeah - sorry I wasn't there this past thursday - too much plague | 19:45 |
| gundalow | where did "depend on GitHub" come from? | 19:45 |
| mordred | sshnaidm: I mean - all of those roles still exist in git repos - which is how people should be installing them | 19:45 |
| mordred | gundalow: the old role import mechanism is "import role from github" | 19:45 |
| gundalow | ah | 19:45 |
| mordred | the new collection upload mechansim solves this issue for us :) | 19:46 |
| gundalow | I think I mentioned this before PaulB has already got publish Collections via Zuul woring | 19:46 |
| mordred | yah | 19:46 |
| mordred | we'll definitely be swiping that from him | 19:46 |
| mordred | he's also getting speculative roles with galaxy going too - very excited about it | 19:46 |
| sshnaidm | mordred, just fyi what we talked about: http://lists.openstack.org/pipermail/openstack-discuss/2019-December/011435.html | 19:48 |
| mordred | sshnaidm: thanks! | 19:51 |
| mordred | sshnaidm: I think 3 isn't *quite* accurate - but I think we should talk about that on thursday so we can get to final clarity | 19:51 |
| mordred | but thanks for writing that up! | 19:52 |
| *** gtema has joined #openstack-ansible-sig | 19:53 | |
| *** gtema has quit IRC | 19:57 | |
| sshnaidm | mordred, why not? | 20:01 |
| *** gtema has joined #openstack-ansible-sig | 20:06 | |
| mordred | maybe I still missed something | 20:09 |
| mordred | gundalow: when you mentioned migrated_to: in BOTMETA above ... does that have some sort of symlink magic associate with it? | 20:10 |
| *** gtema has quit IRC | 20:10 | |
| gundalow | mordred: that's in ansible/ansible | 20:10 |
| mordred | my understanding from reading the various things is that on upgrade to 2.10 a user will need to update their playbooks to have a collection: entry in it and also will need to install the collection, right? | 20:11 |
| *** gtema has joined #openstack-ansible-sig | 20:12 | |
| mordred | (or they will need to update their playbooks to use fully-qualified names) | 20:12 |
| gundalow | unknown, we hope to have some "kitchen sink" installer and some magic so 2.9 playbooks will work out of the box with 2.10 (Ansible Community Distribution) | 20:12 |
| mordred | gotcha | 20:12 |
| mordred | so with the botmeta change in ansible/ansible it could allow people to do the kitchen sink installer and that installer would know what additional collections to install | 20:13 |
| mordred | yeah? | 20:13 |
| *** gtema has quit IRC | 20:17 | |
| *** gtema has joined #openstack-ansible-sig | 20:22 | |
| *** gtema has quit IRC | 20:26 | |
| *** gtema has joined #openstack-ansible-sig | 20:30 | |
| *** gtema has quit IRC | 20:34 | |
| *** gtema has joined #openstack-ansible-sig | 20:36 | |
| sshnaidm | gundalow, ^^ will it just install all collections in botmeta together with ansible? | 20:41 |
| sshnaidm | I think it's logical to do | 20:41 |
| *** gtema has quit IRC | 20:41 | |
| gundalow | sshnaidm: that's the hope :) | 20:55 |
| gundalow | mordred: yup | 20:55 |
| *** gtema has joined #openstack-ansible-sig | 21:08 | |
| *** gtema has quit IRC | 21:15 | |
| *** gtema has joined #openstack-ansible-sig | 21:24 | |
| *** gtema has quit IRC | 21:31 | |
| *** gtema has joined #openstack-ansible-sig | 21:39 | |
| *** gtema has quit IRC | 21:44 | |
| *** gtema has joined #openstack-ansible-sig | 21:45 | |
| *** gtema has quit IRC | 21:50 | |
| *** gtema has joined #openstack-ansible-sig | 21:56 | |
| *** gtema has quit IRC | 22:00 | |
| *** gtema has joined #openstack-ansible-sig | 22:02 | |
| *** gtema has quit IRC | 22:10 | |
| *** gtema has joined #openstack-ansible-sig | 22:13 | |
| *** gtema has quit IRC | 22:22 | |
| *** gtema has joined #openstack-ansible-sig | 22:29 | |
| *** gtema has quit IRC | 22:33 | |
| *** gtema has joined #openstack-ansible-sig | 22:40 | |
| *** gtema has quit IRC | 22:45 | |
| *** gtema has joined #openstack-ansible-sig | 22:47 | |
| *** gtema has quit IRC | 22:56 | |
| *** gtema has joined #openstack-ansible-sig | 22:57 | |
| *** gtema has quit IRC | 23:01 | |
| *** gtema has joined #openstack-ansible-sig | 23:06 | |
| *** gtema has quit IRC | 23:12 | |
| *** gtema has joined #openstack-ansible-sig | 23:13 | |
| *** gtema has quit IRC | 23:18 | |
| *** gtema has joined #openstack-ansible-sig | 23:28 | |
| *** gtema has quit IRC | 23:37 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!