19:00:11 #startmeeting Ansible Community Meeting 19:00:11 Meeting started Wed Feb 15 19:00:11 2023 UTC. 19:00:11 This meeting is logged and archived in a public location. 19:00:11 The chair is felixfontein. Information about MeetBot at https://fedoraproject.org/wiki/Zodbot#Meeting_Functions. 19:00:11 Useful Commands: #action #agreed #halp #info #idea #link #topic. 19:00:11 The meeting name has been set to 'ansible_community_meeting' 19:00:11 #topic Agenda https://github.com/ansible/community/issues/679 19:00:14 acozine, andersson007_, anwesha, baptistemm, bcoca, briantist, cidrblock, cyberpear, cybette, dericcrago, dmsimard, felixfontein, geerlingguy, gotmax, gundalow, gwmngilfen, ikhan_, jillr, jtanner, lmodemal, mariolenz[m], markuman, maxamillion, misc, nitzmahone, oranod, resmo, russoz, samccann, thaumos, zbr: The Ansible community meeting is starting now! 19:00:19 The ping list is stored at https://kutt.it/meeting-people. Feel free to add or remove yourself. 19:00:22 #info Agenda: https://github.com/ansible/community/issues/679 / Topics: https://github.com/ansible-community/community-topics 19:00:24 o/ 19:00:25 #topic Updates 19:00:29 #chair acozine 19:00:29 Current chairs: acozine felixfontein 19:00:38 hm, is chatbot AFK? 19:00:43 s/chatbot/zodbot 19:00:47 o/ 19:00:57 .hello2 19:00:58 maxamillion: Something blew up, please try again 19:01:01 maxamillion: An error has occurred and has been logged. Please contact this bot's administrator for more information. 19:01:01 o/ 19:01:07 .hello2 19:01:08 maxamillion: Something blew up, please try again 19:01:11 maxamillion: An error has occurred and has been logged. Please contact this bot's administrator for more information. 19:01:16 Hello hello 19:01:17 huh 19:01:22 .hello maxamillion 19:01:23 maxamillion: Something blew up, please try again 19:01:26 maxamillion: An error has occurred and has been logged. Please contact this bot's administrator for more information. 19:01:28 womp womp 19:01:36 #chair samccann maxamillion cybette_ anwesha[m] 19:01:36 Current chairs: acozine anwesha[m] cybette_ felixfontein maxamillion samccann 19:01:40 maxamillion: zotbot doesn't like you today :P 19:02:23 indeed 19:02:25 oops I misspelled zodbot 19:02:34 hello all 19:02:36 o/ 19:02:47 * kristianheljas is listening along 19:03:00 #chair oranod kristianheljas 19:03:00 Current chairs: acozine anwesha[m] cybette_ felixfontein kristianheljas maxamillion oranod samccann 19:03:09 furniture for everyone! \o/ 19:03:14 heh 19:03:28 my internet is being really slow/weird today, sorry if I fade in and out 19:04:43 #info Please note that there are over twenty PRs in community.general to add basic attributes to all modules. If you have time to review some of them, please take a look at this list: https://github.com/ansible-collections/community.general/pulls?q=is%3Apr+is%3Aopen+label%3Adocs_only 19:06:26 is there anything else someone wants to announce? 19:06:31 and also, what do you folks want to talk about today? 19:08:02 anyone have testing feedback on beta galaxy? 19:08:43 sorry, I didn't got around to test it... but then I don't really have roles anymore (I only have one I should deprecate, but didn't got around to implement that yet) 19:08:57 it does collections too =) 19:09:25 I have to try with the ones in my namespace; I probably won't manage to upload anything in the community namespace :) 19:09:45 is role/playbook focused collection inclusion in the agenda today? 19:10:33 we could talk about roles in (collections included in) Ansible (https://github.com/ansible-community/community-topics/issues/197), about moving things out of community.general/.network (https://github.com/ansible-community/community-topics/issues/167, or more specialized: terraform - https://github.com/ansible-community/community-topics/issues/192), or about marking plugins in collections 19:10:39 as private (https://github.com/ansible-community/community-topics/issues/154) 19:11:01 kristianheljas: it depends on how many want to discuss it (or if there's something else folks would like to discuss instead) 19:11:34 felixfontein: for the "add basic attributes" PRs, should reviewers test/confirm that the modules have the functionality listed? 19:12:09 Some of those modules are ones I've never used, so I don't have first-hand knowledge of whether or not they support check mode, etc. 19:12:40 acozine: basically yes... if you can confirm the values for some modules that would already be great (please mention which ones you know about) 19:13:18 I'll probably merge the PRs in a bit more than a week (for the next c.g release) if nobody objects (and will until then update all information that is reported as wrong) 19:14:31 kristianheljas: do you want to talk about playbooks/roles in collections in Ansible? since nobody else had any suggestion, we could do that ;) 19:15:13 I have some thought about it 19:15:13 @cybette:ansible.im cyb-clock chimes every 15 minutes during the community meeting 19:15:22 so i'll fo 19:15:25 go* 19:15:51 * maxamillion reads links 19:15:59 #topic Role or playbook-focused collections: include in the Ansible or not? 19:16:03 #link https://github.com/ansible-community/community-topics/issues/197 19:16:52 so basically the question is whether we want to include collections that consist mainly of roles and/or playbooks in Ansible 19:16:58 1) it think this is a good step towards standradizing roles and possibly even ending the "standalone vs legacy" discussion 19:17:12 another question thrown up by mariolenz[m] in that issue is whether we actually want to allow *any* roles in Ansible at all 19:17:27 it would be good indication if users would like to migrate over 19:18:16 kristianheljas: could you describe the "standalone vs. legacy" discussion? I'm not sure what you mean. 19:18:56 in case this is unclear to anyone: this is NOT about including standalone/"legacy" roles in Ansible, but about roles in collections 19:19:51 acozine: there was a discussion at one meeting whether to call distributed roles standalone or legacy, during galaxy beta discussions 19:20:53 I'm not sure how that discussion is related to including collections with roles (or allowing roles in collections included in Ansible) 19:22:21 Oh, i misread the context.... 19:22:39 Sorry about that 19:23:33 no problem 19:24:14 anyway, about this discussion, I'm for allowing collections mainly consisting out of roles and playbooks, but I think / agree we need more strict criteria/requirements for them 19:24:16 * kristianheljas thinks he's on board now 19:24:36 I like calling them standalone, but I'm not super passionate about the topic and don't really mind what we call them .... I used to but then I realized I haven't used a standalone/legacy role in almost two years so ... meh 19:25:03 I also think that collections in ansible should be allowed to have roles and playbooks already now (and I'm a bit biased here, since I added two roles and one playbook to a collection in ansible myself :) ) 19:26:07 (the only standalone/"legacy" role I used a lot was one I created myself, and I've converted it to a collection some years ago and am happily using the collection ever since...) 19:26:32 but then there are parts of the community who use standalone/"legacy" roles a lot more 19:26:35 felixfontein: that inclusion topic is nuanced and I tend to agree that inclusion into Ansible should be module/plugin focused 19:26:47 also there's the problem that collections cannot depend on roles, and roles cannot depend on collections, so it's hard to mix the two 19:27:01 I definitely don't want to ban collections that contain roles and playbooks from the Ansible package. But I don't see an urgent reason to include/add new ones unless they expand what Ansible can do. 19:28:09 as an example, a collection that is mostly roles and playbooks that allows users to automate a new technology - that I would vote in favor of 19:28:25 Thats not a ban decision, but whether to allow a colleciton that only has roles/playbooks 19:28:54 but a collection that is mostly roles and playbooks that don't have a unifying theme or purpose? that I would vote against, I think 19:29:09 I find it hard to think about this without a concrete example, though 19:29:54 in general I'd lean toward a module/plugin focus 19:29:55 well, for one thing it encourages people to continue making hundreds of modules for every endpoint and method in their obscure apis ... instead of just writing roles 19:29:58 maxamillion: acozine: I think I agree very much with both of you... I think there are valid examples for collections with mostly roles/playbooks, but I can't really think of a good example right now 19:30:01 @cybette:ansible.im cyb-clock chimes every 15 minutes during the community meeting 19:30:22 The ones in question are the Infra Validated content Roles, infra.Controller_confirguration, a collection of roles built on awx.awx, and infra.ah_configuration Roles+modules that do hub config with modules and roles 19:30:22 Is the context, Agree on standalone collections 19:30:43 I've used a role for nginx load-balancers that I don't think has any modules but is very useful 19:30:59 on standalone multiple purpose without a unifying purpose would not meet criteria 19:31:46 jtanner: yeah, writing modules "because that's what we need to do to get our stuff into Ansible" is a bad pattern 19:31:54 when I start thinking about this topic I'm quickly getting to the point where I'm wondering whether we should stop adding almost any content to Ansible, and instead throw things out to keep it more slim... I'm really torn on this 19:32:05 and that bad pattern has been around for a long time 19:32:22 yeah, my first response was "isn't the package big enough already?" 19:32:23 I would be fine on that point to felixfontein I am more interested in Getting into Ansible DOCS, for discovery 19:33:07 felixfontein: I feel this, at some point the anisble package needs some sort of split when it becomes "too large" - whatever that amount is 19:33:08 Right now if we started sliming ansible itself, I'd say the ansible docs still should accept new collections to be searchable and readable for use 19:33:20 did anyone ever define big enough? i always had the sense that you all would keep adding to it until you needed a new internet provider to download the package in a single day 19:34:06 the bigger the package, the bigger the bloat around the stuff you need 19:34:12 SeanSullivan[m]: hmm, we should never include collections only because they want to end up in the Ansible docs... we should only include them if we think they are useful for ansible (community package!) users 19:34:40 right now I have some doubts why a collection for installing AH makes any sense part of the ansible community package 19:34:46 felixfontein: It is useful for the community, and seeing larger amounts of customers using it 19:35:26 but if users start to use ansible package less and more galaxy, that might prove useful 😛 19:35:34 less maintanance, etc.. 19:35:38 I just meant, With collections, Hub/Galaxy, Execution environments, I see more people moving away from The ANsible Ansible, and to core + collections 19:36:00 SeanSullivan[m]: but these folks won't use the Ansible community package to set everything up, will they? 19:36:12 ACD as it was initially called, was supposed to be a temporary stop gap ... 19:36:28 jtanner: how 'discoverable' are collections that are only in galaxy (NG)? As in if I google for something that is only in Galaxy NG would it show up? 19:36:44 jtanner: there are unfortunately too many things Ansible/ACD should have been :) 19:36:59 samccann: you'd have to piece that together yourself by scraping and diffing 19:37:01 cuz I feel like part of Sean Sullivan point is that content only in galaxy can't be found by google (tho maybe I'm putting words in their mouth on this one) 19:37:47 and pypi resolves that somehow? 19:37:47 IT CAN be found, it if you aren't searching for something specific, how likely ARE you going to find it 19:38:29 Sean Sullivan: sadly I dunno what community Wisdom might do in this regard, but seems like it might help? 19:39:09 I think "allowing documenation of collections not included in Ansible to be shown somewhere on docs.ansible.com" is a valid thing, but should not be the main reason for including something in the Ansible community package 19:39:20 It might, And what I said before, Ansible inclusion the docs were not the ONLY reason for us proposing, but It was one of those advantages that weighed in favor of doing so 19:40:24 felixfontein: what would you describe the reasons to be included 19:40:34 i personally think role-only or role-heavy colletions should be considered as 1st tier content for "ansible" just as much as any module-heavy collection. i use geerlingguy's roles far more than i do any module in the ansible package 19:41:08 so the meta discussion is - what is the future of Ansible community package? Do we let it keep growing and growing? And do we expand it out to collections of roles/playbooks? 19:41:42 SeanSullivan[m]: from my POV: something that is useful for a non-trivial fraction of the Ansible package's users. unfortunately this is very hard to convert into explicit rules 19:41:47 oof 19:41:50 that's ... a fun one 19:41:53 heh, the favorite topic ;) 19:42:09 samccann: I think so, As If we are allowing Quality Top tier roles included, it should just allow them, but if we want to rethink the thing, I think its worth discussion 19:42:12 my ears ring every time you all get close to this tangent 19:42:26 jtanner: <3 19:42:30 heh yeah we've had this talk before 19:43:05 https://github.com/ansible-community/community-topics/issues/82 19:43:38 but not to keep poking the stick, it feels like community Wisdom could be the game-changer here. If it is what I think it is - an AI that would help users find content 19:44:01 you are talking about the wisdom project? 19:44:06 heh, too bad such AIs don't exist 19:44:19 So if I could type 'somewhere' I need a role that would do bla' and community wisdom could find it... we aren't dependent on docs.ansible.com (and google search results) 19:44:25 LOL 19:44:41 * samccann frequently speaks of that which she doesn't not know 19:44:47 https://www.redhat.com/en/engage/project-wisdom ... "It will be accessed as a feature within the Ansible extension for Visual Studio Code, and will work with new and existing documents." 19:44:52 the most famous AIs nowadays are good at inventing their own facts :) 19:45:13 @cybette:ansible.im cyb-clock chimes every 15 minutes during the community meeting 19:45:33 i think wisdom will operate similar to github copilot moreso than chatgpt 19:45:39 jtanner: yeah I think documents in that sense means what's inside a collection, not docs.ansible.com 19:45:55 After seeing ChatGPT in BING, its clearly useful/doable as a Search Helper 19:47:02 Havent tried bing, but ChatGPT is a great google search result condenser. But as with google, youd need to verify any claims made 19:47:25 which makes it pretty worthless as a search result condenser IMO 19:47:50 but that's pretty off-topic I guess :) 19:48:03 it is 19:48:16 I will put a comment into the community-topics issue. For now I have to leave, to prep for my next meeting at the top of the hour. 19:48:23 Thanks everybody for a lively conversation. 19:48:26 thanks acozine! 19:48:44 * acozine waves 19:48:44 So if we take the #82 as closed, then goes back, what should be criteria for majority roles based collections 19:48:45 #unchair acozine 19:48:45 Current chairs: anwesha[m] cybette_ felixfontein kristianheljas maxamillion oranod samccann 19:49:09 to steer it back to original subject, and do we have an open issue to post to about it 19:49:31 none of these ai models are going to tell you which role or collection or module actually works and isn't broken unless there is some supporting data to provide that 19:49:39 Sean Sullivan: Yes, https://github.com/ansible-community/community-topics/issues/197 19:50:06 /facepalm, I'm #2 on that thread 19:50:21 :) 19:51:27 Me wonders, how much review load might these inclusions bring as they are easier to create IMO 19:51:43 jtanner: isn't that also a problem for roles/playbooks being added to the Ansible package tho? We don't have quality control on that afaik. 19:52:13 I think for roles they'd need to reach a certain acceptance level at some point though, and at what point, do we just include every single Geerling collection 19:52:19 :-) 19:52:22 what gaurantees do i have on the plugins in collections? 19:52:45 I thought there were basic CI tests as part of the inclusion criteria for plugins but I could be mistaken 19:53:13 And The roles we have in 3/4 collections are included in 4 CI tests, and I'd be fine with leaving aap_utilities based on AAP installation out of it 19:53:20 roles had CI capabilities long before collections did 19:53:34 molecule, lint, etc 19:54:46 roles existed long before collections did, so that's not surprising :) 19:57:40 If we keep allowing roles/playbooks in the existing collections it might also be a bit unfair to the focused ones 19:58:32 kristianheljas: you mean, if we do not want to include focused ones? 19:58:46 mainly based on the argument that they don't have plugins/modules 19:59:18 I'm not sure a majority wants to exclude roles/playbook focussed collection for *that* reason 19:59:27 yeah, that's correct 20:00:09 not that the reason "we don't have good standards for roles/playbooks in Ansible yet" is *that* much better with this regard, though 20:01:19 anyway, time's up! 20:01:22 thanks everyone for the discussion! 20:01:42 please don't forget to add your opinions to https://github.com/ansible-community/community-topics/issues/197 (if you haven't done that already)! 20:01:46 #endmeeting