16:06:32 #startmeeting Solum Team Meeting 16:06:32 Meeting started Tue Aug 12 16:06:32 2014 UTC and is due to finish in 60 minutes. The chair is tomblank. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:06:33 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:06:35 The meeting name has been set to 'solum_team_meeting' 16:07:10 Ah, adrian_otto just joined... 16:07:22 sorry for the delay 16:07:33 adrian_otto: Error: Can't start another meeting, one is in progress. Use #endmeeting first. 16:08:04 adrian_otto: i went ahead and started the meeting but can end and then you can restart as chair if you want 16:08:26 let's proceed, if you are willing to do the topic switching 16:08:38 sure... 16:08:43 #topic Roll Call 16:08:49 o/ 16:08:49 Adrian Otto 16:08:52 Roshan Agrawal 16:08:54 tom blankenship 16:08:56 Ravi Sankar Penta 16:08:57 Devdatta 16:08:58 Noorul Islam 16:08:58 Arati Mahimane 16:09:00 Ed Cranford 16:09:00 james li 16:09:17 Julien Vey 16:09:53 #topic Announcements 16:10:06 any announcements from the team? 16:10:57 ok, let's advance 16:11:48 #topic Action Items 16:14:59 #link http://eavesdrop.openstack.org/meetings/solum_team_meeting/2014/solum_team_meeting.2014-08-05-22.00.html last week's minutes 16:15:34 looks like ravips did already submit a ticket for keeping our wiki docs in rst 16:15:52 yes, opened bug https://bugs.launchpad.net/solum/+bug/1355523 16:15:54 Launchpad bug 1355523 in solum "Convert solum wiki pages to RST format" [Undecided,New] 16:15:56 thanks! 16:16:01 ravips: thanks! 16:16:18 #topic Blueprint/Task Review 16:16:39 #link https://launchpad.net/solum/+milestone/juno-2 Juno Development Tasks 16:16:52 I am planning on curring the .2 release today 16:16:57 *cutting 16:17:14 is there anything we should arrange to merge before then? 16:17:30 #link https://wiki.openstack.org/wiki/Meetings/Solum agenda — if anyone is looking for it 16:17:42 adrian_otto: what are the main features being released? 16:18:08 pipelines, private repos 16:18:20 I know pipelines all merged 16:18:25 did private repos get in yet? 16:18:30 no, not yet 16:18:30 not yet 16:18:31 well i'd like the logging changes for worker to get there so we can actually chase down what happens in those containers 16:18:44 +1 datsun180b 16:18:46 private repos is aaaaalmost there thanks to ravi 16:18:50 I made some more changes as per feedback from murali and Ed 16:19:07 we've been testing your changes intensely ravips 16:19:09 datsun180b: could you please share the patch link for logging? 16:19:18 any chance of merging that today? 16:19:27 we couldn't make it run on F20, james opened a bug against barbican, submitted a patch https://review.openstack.org/#/c/113404/, waiting for approval from barbiacan team 16:19:32 even if it requires putting in tech debt bugs to follow up afterward to address? 16:19:36 #link https://review.openstack.org/#/c/107464/ 16:19:50 adrian_otto: we might be rushing it.. 16:19:57 it should rebase cleanly against whatever else got in ahead of it but if not i can take care of it easily enough 16:19:58 thanks datsun180b 16:20:27 I can hold off on the release for a day or two if that would allow us to get it in 16:20:31 sorry i'm late 16:20:35 hi Gil 16:20:53 adrian_otto: that would help. I suggest we do that.. 16:20:59 so I can cut as many releases as we want 16:21:10 +1 16:21:11 oh okay. 16:21:18 so why don't I just cut one today with the current code, and then make a revision when we have the private repos 16:21:19 in that case you can cut one today without private repo 16:21:42 sgtm 16:21:47 the next one can have private repos and Ed's logging stuff 16:21:53 right 16:22:01 adrian_otto: +1 for cutting another with logging and private repos 16:22:06 cool 16:22:30 #agreed Cut a release today, and a subsequent release with Logging and Private Repos added 16:22:43 tomblank: ^^ need to repeat that for the bot to notice 16:23:03 #agreed Cut a release today and a subsequent release with Logging and Private Repos added 16:23:32 ok, so other status about ongoing work? 16:23:50 I'm having some troubles with marconi in devstack 16:23:58 opened a bug in marconi 16:24:26 julienvey: we're working on that. The issue doesn't seem to be related to marconi but the openstack-common client 16:24:46 flaper87: oh nice, thank you :) 16:24:54 julienvey: mind joining #openstack-zaqar ? We'll keep you updated 16:25:06 flaper87: I'm on it 16:25:12 awesome 16:25:55 any other status on ongoing work? if not, we'll move on to the next topic. 16:26:02 ok, we can switch topics to what we would all like to see included in Juno-3 16:26:10 #topic Feature Planning, Juno-3 (For Paris) 16:26:22 this would be anything you'd like us to be announcing in Paris. 16:26:55 i bet we could swing functional tests 16:27:00 paris will be almost our 1 year project birthday 16:27:10 adrian_otto: we'd like to have the build farm ready 16:27:17 there was a feature roadmap that roshanagr had. are we not just following that? 16:27:34 unit -> build -> functional -> deploy 16:27:40 devkulkarni: do you have a link to that handy? 16:27:45 I'm checking on it 16:27:50 Did we finish everything that is currently bucketed under Jun-2 release 16:27:58 https://launchpad.net/solum/+milestone/juno-2 16:28:33 roshanagr: you had a wiki as well.. do you have that link handy? 16:28:56 git, Zuul, Environmments and CI/CD events 16:29:09 those are the ones we have earmarked for Juno 16:30:02 except for git we are not there with others 16:30:30 We can do custom language pack support 16:30:33 git push was implemented? 16:30:35 some additional things have happened instead (custom lang packs, pipelines, logging) 16:30:51 not in the sense of Solum hosting git repositories 16:31:07 Roadmap wiki link: https://wiki.openstack.org/wiki/Solum/HighLevelRoadmap 16:31:14 but more in the sense of 'users can push to their git repos' and Solum can take it from there 16:31:28 is Zuul integration still desired for Juno? 16:32:03 roshanagr ^^ 16:32:15 my thought is, we are using Mistral 16:32:18 for workflow 16:32:40 so for now we can punt on not using Zuul for that capability 16:32:53 devkulkarni: that is my thought as well but want to get other folks opinions. 16:33:26 Should we run through the items parked under Juno-2 and see where we are on each 16:33:27 the reason Mistral support was added was to allow customization of the workflow 16:33:57 we would rather not duplicate functionality already present in other ecosystem projects 16:34:13 so leveraging the Mistral work to integrate Zuul would be a logical next step 16:34:54 another possibility is to show an example integration of using Jenkins for CI 16:34:58 and integrate that 16:35:46 to demonstrate that for shops that have an investment in Jenkins for CI already can leverage that, and use Solum for STI capability, and continuous deployment. 16:35:47 sure. 16:36:15 STI = ? 16:36:21 source to image 16:36:22 adrian_otto: on integrating jenkins. my 2 cents is that this would be higher priority than Zuul integration. 16:36:26 ok :-) 16:36:57 tomblank: +1 16:37:37 + 1. Jenkins is standard in CI 16:37:43 At the time we st our sights on Zuul, we did not have something like the CI pipeline code contemplated, so we were hoping that Zuul might be an answer for that 16:37:58 I think it makes sense to swap in Jenkins for Zuul in the planning 16:38:05 +1 16:38:11 and put Zuul back on the wishlist for now 16:38:14 +1 16:38:24 lets get an agreed on that :) 16:38:31 if someone feels strongly, and wants to bring developers to work on that, we can see about advancing the priority again 16:38:48 ok, alternate points of view before considering an #agreed 16:38:50 ? 16:39:14 are there other new features that anyone considers higher priority than jenkins? 16:39:33 integration with trove, lb 16:39:41 were those part of juno-2 16:39:42 we should be clear about what we mean by "integrating jenkins" 16:39:43 or later? 16:39:46 are we talking that we would install and run jenkins ? or allow for an org to use their existing jenkins ? 16:39:57 What "feature" would we get from Jenkins immediately? 16:39:57 you do get jenkins via zuul if you want to kill two birds with one stone 16:40:08 is it "enable already running jenkins to send notifications to solum" 16:40:17 PaulCzar: allow usage of existing Jenkins to be consumed within Solum's customizable CI 16:40:28 or "enable CI via Jenkins in Solum workflow" 16:40:30 roshanagr: good question. We would get the ability to add your own existing CI solution to solum 16:40:46 the feature is actually there, but this would be a way to demonstrate it 16:40:58 adrian_otto: thanks ! 16:41:12 clarkb: good point. imo, starting with jenkins as a first step would be easier than 'zuul + jenkins' 16:41:25 to tell the story of two CI camps: 1) I am a CI master already, let me hook up, and 2) I don't do CI yet, and I want to start. 16:41:43 most software dev shops fall into those categories today 16:42:00 julienvey: i think it is "enable CI via Jenkins in Solum workflow". 16:42:03 so, we would have an example mistral workflow that would call out to an external jenkins for the CI portion of the workflow for the approriate task ? 16:42:04 adrian_otto: on 1) - why would an existing Jenkins CI shop want to hook up to Solum CI 16:42:16 PaulCzar: yes 16:42:34 at least thats my view 16:42:43 roshanagr: they might already be managing user/integration testing with jenkins, and want Solum to do the source->image and deploy 16:42:48 roshanagr: for continuous deployment, and creation of docker images from a hosted service 16:43:00 s/user/unit/ 16:43:06 to allow portability of the output to numerous deployment targets (OpenStack, other clouds, etc.) 16:43:36 adrian_otto: paulCz: thanks, get it 16:43:36 PaulCzar: exactly. 16:44:23 what about integration of APM? 16:44:48 any additional questions on this and are we agreed to put Zuul integration on our backlog, and replace it with a simple Jenkins integration task that shows how to connect an existing Jenkins system to Solum? 16:44:48 APM = Application Performance Monitoring. The prevailing leader here today is NewRelic 16:45:36 adrian_otto: what about it? are you asking should we pull it for Juno? 16:45:37 Is APM within the scope of STI? 16:45:56 #agreed Put Zuul integration on our backlog, and replace it with a simple Jenkins integration task that shows how to connect an existing Jenkins system to Solum. 16:46:05 roshanagr: those two are logically two separate things, no? 16:46:22 roshanagr: no it is beyond the scope. I wonder if we might scope that out for Juno, and focus on other features. 16:46:26 devkulkarni: yes, that is what I thought as well 16:46:39 adrian_otto: yes 16:47:23 yes. did we have trove and load balance integration in Juno? If not, we should pull that in 16:47:34 I think that will be useful 16:47:44 +1 devkulkarni 16:47:56 Trove integration will be great. 16:48:00 devkulkarni: yes, that is listed in Milestone 2014.2.3: Juno-3 16:48:16 ok 16:48:31 with build farm, jenkins integration, trove, LB, CI/CD events... that's a lot to complete prior to Juno... 16:48:53 APM: my bad, the roadmap has parked it under Juno 3. I take an action item to update the roadmap wiki 16:48:55 I suggest we make a prioritized backlog of the features 16:49:00 but maybe i'm just being pessimestic :) 16:49:10 there’s also an interesting new project from google called cadvisor that we might want to look at integrating - https://github.com/google/cadvisor 16:49:17 and we will advance through them according to those priorities 16:49:29 tomblank: CI/CD events is a big topic. we can split it into multiple pieces and deliver them piece by piece 16:49:34 s/pessimestic/ Pessimistic 16:49:39 PaulCzar: let's add that to the agenda to discuss next week 16:49:57 devkulkarni: agreed... 16:50:07 have to go :( bye 16:50:11 adrian_otto: +1 16:50:14 ok, anyone super interested in a Juno feature that we have not yet identified? 16:50:16 PaulCzar: please give a short intro to it. 16:50:23 what is it? why we should look at it? 16:50:24 etc. 16:50:34 julienvey: thanks... 16:50:50 okay, we should advance to Open Discussion now 16:50:58 #topic Open Discussion 16:51:03 I have one topic 16:51:15 i have a topic 16:51:20 gpilz has been working on the CAMP API additions to Solum, which is great 16:51:38 but it requires abstractions like assembly 16:51:40 to be in place 16:51:46 I believe with pipelines 16:51:52 I'd like to get https://review.openstack.org/#/c/112153/ reviewed 16:51:54 we are getting rid of assembly notion 16:52:13 devkulkarni: they are mutually exclusive 16:52:16 I think gil's work won't be able to go anywhere if we get rid of assemblies 16:52:26 you don't need to use an assembly to do an STI process 16:52:39 it's for the post deployment part of the lifecycle 16:52:41 adrian_otto: yes — I would like them to be mutually exclusive 16:53:17 we should not eliminate Assemblies form the Solum API 16:53:24 but we don't need to use them for Pipelines 16:53:47 s/form/from/ 16:53:58 longer term, assemblies might wrap on top of pipelines, but I think short term we will probably need to maintain both 16:54:04 yes — as long as this is clear to everyone we should be good 16:54:33 devkulkarni: do you have suggestions on how to achieve that clarity? Are we there now, or is there more needed? What could help get there? 16:54:56 so I think we should surface this issue when asalkeld is part of the meeting 16:55:26 pipelines are a way of customizing how the application is built? 16:55:29 should we start a ML thread on this? 16:55:34 I might be wrong, but I think we all need to be on the same page with regards to pipelines and assemblies 16:55:37 #action adrian_otto to add discussion of Pipelines and Assembly to next meeting agenda 16:55:42 devkulkarni: I think the issue is that there are some people who have a vested interest in continuing to improve assembly, even though pipline is ‘the future’ 16:55:52 #action adrian_otto to add discussion of Pipelines and Assembly to next meeting agenda 16:55:55 PaulCzar: I disagree 16:56:01 so we need agreement from the community to continue to working on both 16:56:17 I think the notion of assembly is separate from notion of a pipeline 16:56:33 vested interest or not, they identify different levels in an application lifecyle 16:56:36 if we don't have assembly, we only have stack 16:56:46 so think on that, and we will revisit that topic next week 16:56:56 from a CAMP perspective an assembly represents a running application in its steady state 16:57:10 how it got to that steady state is not something that concerns assembly 16:57:12 and we've been calling that a DU informally in solum 16:57:35 at least, a deployed DU 16:57:44 personally I am more concerned with the churn 16:57:53 datsun180b: no, a du is the execution of a component of an app in the scope of a single host, not a group of hosts 16:57:54 that changes to the abstractions 16:57:56 datsun180b: we call it an assembly 16:58:05 An assembly can comprise of multiple DU's hooked together with other services 16:58:14 roshanagr: +1 16:58:16 changes to the abstractions cause in the minds of everyone 16:58:24 see this is why i'm quiet 16:58:33 hehe 16:58:47 ok, so we have two minutes remaining for open Disucssion 16:58:57 ha ha ha 16:59:03 https://review.openstack.org/#/c/112153/ ? 16:59:15 gpilz: I will review it today 16:59:19 From an end user point of view, 'solum assembly create' seems a better command than 'solum pipeline create' 16:59:21 do i need to get this approved before I can submit more changes? 16:59:28 +1 aratim 16:59:35 adrian: thanks 16:59:53 gpilz: if you want to avoid possible rework, yes 16:59:59 if you do not care about rework, no. 17:00:00 gpilz: code can be submitted before spec is merged (at least we have done that before, so should be okay) 17:00:12 submit it as a WIP 17:00:14 but leave your commits in WIP state until spec approval 17:00:16 i love re-work 17:00:21 thanks everyone, tom, let's end 17:00:23 okay 17:00:29 #endmeeting