15:00:26 #startmeeting cloudkitty 15:00:27 Meeting started Fri Nov 9 15:00:26 2018 UTC and is due to finish in 60 minutes. The chair is peschk_l. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:28 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:30 The meeting name has been set to 'cloudkitty' 15:01:11 Welcome to his meeting. Today's agenda can be found at https://etherpad.openstack.org/p/cloudkitty-meeting-topics. If there is any topic you're intersted in, fell free to add it to the list 15:02:09 For the record, today will mostly be about the community. We'll be talking about easing contributions, documentation and community goals 15:02:55 We also need to discuss the meeting schedule, as it may need to be adapted depending on the timezone of the different contributors 15:03:25 First topic will be easing contributions 15:03:33 #topic easing contributions 15:04:57 As you probably have realized, it may be complicated for new contributors to make their first contribution 15:05:29 We are missing guidelines, and the IRC channel is not very active 15:06:34 A developer documentation would also be a great help 15:07:04 Our roadmap should also be easy to find 15:07:51 And I would like to make the usage of the spec repo systematic: It's a good way to discuss new features and to give the community an overview of the direction the project is taking 15:08:47 a worklist has been created in order to tackle this issue and to keep track of the work that is done on these matters: https://storyboard.openstack.org/#!/worklist/497 15:09:34 yes, I think it's a good thing to update the spec repo 15:09:56 The first one is about the documentation only. I proposed a spec for a complete refactoring: https://review.openstack.org/613550 15:10:00 Hi Linkid :) 15:10:09 hi :) 15:10:47 It would be great to have reviews/suggestions on this matter 15:11:34 I already had a bit of feedback, but mostly from colleagues. We'd like to have the opinion from people outside of Objectif Libre 15:11:59 Linkid: if you could give us your opinion on this spec it would be great 15:12:46 I'll read it as soon as possible. But I can't do it this week-end 15:12:48 the second story of the previously mentionned worklist is more about how interactions in the community could be eased in general: https://storyboard.openstack.org/#!/story/2004180 15:13:07 Linkid: no pressure! 15:14:17 This second story is by far not as specific as the first one, so any suggestion/ additional task / storyboard comment is welcome 15:15:52 We'd really like to gain some contributors. This means we need some visibility and to make clear that new contributors are welcome 15:16:00 yes, it is the same problem in many OS projects… maybe we can see what other OpenStack ones 15:17:03 the integration with some tools like puppet / ansible is a good one (it is already done, though) 15:17:59 I agree. but Iunfortunately I'm not sure that the integration is flawless 15:18:11 peschk_l: I saw that you also started a charm repo. Maybe we could use it 15:18:47 a lot of people ask us questions about kolla-ansible, it seems to be the prefered way of deploying it for POCs 15:19:07 (and ask for help to canonical people if necessary :)) 15:19:14 Linkid: yes a while ago, but there were some CI issues at that point... It would need to be refreshed 15:19:37 James Page was a big help back when I was hacking on the charm 15:19:59 But I don't think that I'll have time for the charm soon :/ 15:20:27 jferrieu has joined the cloudkitty development team a few weeks ago. He's looking at kolla-ansible 15:21:06 It would be agood thing to maintain a list of methods to try out cloudkitty easily 15:21:47 and back to 2014, I remember huats (IRC) made a video to show how to use CloudKitty. I think it's was good idea. Maybe we could do it with our current release 15:21:56 about the roadmap: any opinion on where we should maintain it ? 15:22:25 Linkid: I believe there are some on Objectif Libre's youtube channel 15:22:43 ah, great :) 15:23:09 https://www.youtube.com/watch?v=BLEcfhUwahM 15:23:34 Linkid: do you have an opinion about where the roadmap should live ? wiki/doc.. ? 15:24:06 I believe the doc would be somehow more "official" and easier to find, but it may be a bit more tedious to maintain 15:24:42 hmm… I don't know. Using the storyboard by tagging stories with "feature" is an option 15:25:36 IMO Storyboard should definitely be a complement, but not the only way to have a roadmap 15:26:56 We need an official list of planned features, a date (near future/distant future..), and a list of feature that won;t be implemented because they aren;t in cloudkitty's scope 15:27:59 (an exmaple of such a feature would be to limit ressources once a certain level/cost has been reached. This isn't really part of the rating scope) 15:28:10 I found those two links : https://www.openstack.org/software/roadmap/ & https://ask.openstack.org/en/question/2699/where-can-i-find-the-openstack-roadmap/ 15:28:49 (reading it) 15:29:14 OK, will have a look at it later 15:29:17 #link https://www.openstack.org/software/roadmap/ 15:29:29 #link https://ask.openstack.org/en/question/2699/where-can-i-find-the-openstack-roadmap/ 15:29:32 so that's a topic we'll have to discuss later 15:30:01 yep 15:31:04 are there any other questions on this specific topic or should we move on to the next one ? It would be about the upcoming features 15:31:56 it's good for me 15:32:25 Linkid, something on your side ? 15:33:02 nope 15:33:07 ok, moving on then 15:33:21 #topic upcoming features 15:34:15 I believe that the biggest feature of this release is the v2 storage 15:34:31 It has been implemented for two reasons 15:35:00 1 - CloudKitty no longer has a ressource/openstack service approach, but is rather metric-oriented 15:35:35 2 - We are slowly running into the same issues as the telemetry project had at some point: we handle a big amount of data 15:36:32 a developer documentation about the v2 storage interface can be found here https://docs.openstack.org/cloudkitty/latest/developer/storage.html 15:36:49 It definitely needs to be completed 15:37:59 anyway the major improvements brought by the v2 storage interface are the following 15:38:05 * pagination 15:38:10 * grouping 15:38:14 * filtering 15:38:41 4 years ago, in the conception, we thought of panda to handle data 15:38:58 arg, storage, sorry 15:39:44 it allows to group data by user / domain / anything rather than just by tenant 15:40:26 (yes pandas would be a bug improvement, but that would be a topic for another meeting :) But feel free to write a spec about it!) 15:41:14 anyway this new way of exposing our internal data means that the api needs to be refreshed 15:41:52 We've been wanting to move from pecan/WSME to something else for a long time 15:42:27 current favourite is flask/flask-restful, but this also needs to be discussed 15:43:04 I had a chat with the API work group and produced the followign spec afterwards: https://review.openstack.org/#/c/614275/ 15:43:47 bascially the idea would be to make the v2 api a new WSGI app and to dispatch request between these two apps 15:44:51 moving the v1 API to another framework would be a HUGE amount of work 15:45:17 With the two app approach, we can simply create a new v2 endpoint for each new feature 15:45:38 and slowly implement each endpoint of v1 in v2 also 15:46:03 once the v2 is considered complete, we would mark it as current 15:46:53 v1 api is completely compatible with the v1 and v2 storage interface. However, v2 API won't be retrocompatible with the v1 storage interface 15:47:22 more details in the spec. Here again, reviews would be a great help 15:47:35 Any opinions about this Linkid ? 15:50:28 We're slowly running out of time, let's move on to the next topic 15:50:39 #topic community goals 15:50:55 I clearly agree with your approach peschk_l 15:51:35 there are two community goals for this release: python3 first and upgrade checks 15:52:03 I believe that the only thing we are missing for python3-first iare functional tests running under python3 15:52:11 (thanks huats) 15:52:25 the thing is: we don't have functional tests, except from tempest 15:52:52 But tempest depends on devstack, and I'm not sure that devstack can be deployed on python3 right now 15:53:05 peschk_l: yep, I agree weth you about adding the v2 endpoint progressively :) 15:53:39 arg 15:53:52 Concerning upgrade checks, I submitted the following patch: https://review.openstack.org/#/c/613076/ 15:54:16 (thx Linkid, any strong opinion about the framework ?) 15:54:59 I thought about adding tests in the dashboard plugin too. I looked at it when I started removing non-free blobs 15:55:03 For now the only check is that v2 storage is used, If not, a warning is raised. Here again, reviews/suggestions are welcome 15:55:41 Linkid: yes, the dashboard Is also something I would like to talk about. It will be a topic for the next meeting/ to be discussed at the summit :) 15:56:02 argh, only 4 minutes left 15:56:09 no opinion about the framework. I already take a look at it in other OpenStack projects, and I think flask was suggested or used soewhere 15:56:16 *somewhere 15:56:59 I like flask because a lot of people are using it, which means easier contributions for new contributors 15:57:10 yep 15:57:43 I'm sorry, we need to move on to the next topic.. 15:57:53 #topic cloudkittyio 15:58:23 Again, this is about gaining contributors/visibility 15:59:56 The idea is to create a cloudkitty.io website using github. The content would be managed through github pull requests. There would be links to useful resources (how to try it out, contribution guide...) and a blog, where people could post cloudkitty-related articles 16:00:17 (their experience about deployments, custom integrations, tips and tricks etc...) 16:00:17 a page like this one would be great : https://www.objectif-libre.com/fr/services/cloudkitty/ (in english and with more pointers) 16:00:23 Great idea 16:00:32 starlingX use a page like that 16:01:16 the idea would be to post this kind of content https://www.objectif-libre.com/en/blog/2018/03/14/integration-monasca-et-cloudkitty/ and https://www.objectif-libre.com/en/blog/2018/05/23/how-cloudkitty-will-allow-you-to-rate-your-prometheus-metrics-a-step-by-step-tutorial-with-traefik-as-a-use-case/ 16:01:33 peschk_l: could we use the same git workflow as CloudKitty ? 16:01:36 it would be better on cloudkitty.io rather than a company's blog 16:02:16 Linkid: I was rather thinking about github pull requests. A lot of openstack users aren't familiar with the openstack/gerrit weorkflow 16:02:20 *workflow 16:02:21 yep, I told about the content ;) 16:03:26 ah, ok 16:03:27 good 16:03:46 it's important to have feedbacks and use cases 16:03:51 glad to see you also like the idea :) 16:04:16 arg, we need to close the meeting. We'll continue this discussion in a more unformal way 16:04:40 huats, Linkid, jferrieu any last words you'd like to add before we end this meeting ? 16:04:41 a small pelican website is easy to use 16:06:14 see you in Berlin :) 16:06:37 it's good for me, and to have a website separated from the corporate blog is a indeed a good idea to develop the autonomous identity of the project and its visibility 16:06:54 imho 16:07:09 I agree jferrieu :) 16:07:14 yep 16:07:25 well, this ends this meeting, next one will be held on december 7th 16:07:35 #endmeeting