23:01:18 #startmeeting product-team 23:01:18 Meeting started Wed Mar 18 23:01:18 2015 UTC and is due to finish in 60 minutes. The chair is sarob. Information about MeetBot at http://wiki.debian.org/MeetBot. 23:01:19 Anyone here for the Product Work Group meeting? 23:01:19 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 23:01:22 The meeting name has been set to 'product_team' 23:01:29 roll call 23:01:37 o/ 23:01:40 Carol Barrett 23:01:48 Roland Chan 23:01:48 * sgordon here 23:02:08 Hi 23:02:17 * sarob \o/ 23:02:28 o/ 23:02:44 excellent 23:02:50 agenda 23:02:53 #link https://wiki.openstack.org/wiki/Meetings/product-team 23:03:27 barrett1: want to kick off what happenned at the ops midcycle? 23:04:02 go for it! 23:04:19 or anyone that was there... 23:04:22 Sure - I sent several emails to the group with info along the way 23:04:31 #topic ops midcycle 23:04:32 I also posted info in our etherpad 23:05:02 We were able to talk with Operators, PTLs/Developers and Foundation folks 23:05:02 which one? 23:05:08 etherpad 23:05:23 Checking 23:06:13 https://etherpad.openstack.org/p/kilo-product-management-socialization 23:06:49 line 639 onwards 23:06:50 There was a placeholder at the end for feedback from Ops Meetup so I added the email info there 23:07:23 im with you 23:07:42 I also edited the wiki page to add a link to the roadmap that includes the PTL interview info and PTL comments about what the PWG could do for them 23:08:17 We still are missing info on some key projects on intentions for K, L and M 23:08:19 better color, thx 23:09:09 i like the ttx feedback 23:10:03 im confused to the not meeting with TC part 23:10:21 since the TC is a cross section of the PTLs mostly 23:10:22 yes, I think their buy in as a group is needed. 23:10:27 mostly 23:10:37 but i think what he is getting at is that is more coincidence than anything 23:10:37 Thierry didn't see a reason why we would need to meet with the TC, they don't approve this 23:10:44 as the PTL = TC link is long gone 23:10:59 the xcpoject meeting, right after the TC meeting is where all the PTLs are supposed to gather and work together 23:11:06 so there are PTLs, of projects which are likely important to us, who arent on the TC 23:11:14 (plenty of them in fact) 23:11:15 right 23:11:16 Thierry suggested the xproject team because all PTLs are there and it's their meeting. 23:11:20 Rockyg +1 23:11:26 rockyg: good point 23:11:39 support != approval 23:11:42 im fine starting with the xproject team 23:11:51 the same peoples 23:11:54 similar 23:12:17 rolandchan: meaning? 23:13:05 meaning there is value in having the TC say they support something even if they aren't in an approval chain for that thing. 23:13:33 got it 23:13:49 like being responsible rather than accountable 23:14:13 maybe consulted or informed is better example 23:14:25 What we could take into the xproject meeting is goals, process, sample roadmap (from interviews), sample use case (which would be based upon Operators feedback and be the basis for populating the roadmap). 23:14:26 Wouldn't the TC want to be involved when a long-term roadmap is being proposed (based on PTL feedback) or since the roadmap is built using PTL feedback it doesn't require approval? 23:14:32 yep. I think "consulted" is perfect. 23:14:55 I believe that Thierry thinks if the PTLs all agree, he can feed the info back to th TC without issue (optimistically) 23:15:02 Does the TC "approve" the goals for each release for each project? 23:15:13 barrett1: nope 23:15:19 PTLs job 23:15:25 PTLs rule 23:15:29 dude 23:15:31 sarob ++ 23:15:37 I don't understand why the TC would be involved with the Roadmap either 23:15:42 Makes sense if we don't encounter disagreements. I guess the bigger question is do we think it will be instances where this agreement may arise? 23:16:03 blah auto correct 23:16:12 i wanted to socialize with the TC what we are doing 23:16:17 rather than get approval 23:16:19 This agreement = disagreements 23:16:22 There will be priority issues on getting mulitple PTLs on the same page 23:16:34 Rocky ++ 23:17:03 why not socialize on the openstack-dev ML? 23:17:07 The other thing the PTLs talked about was the need to balance new features with bug fixing - they struggle with this 23:17:11 there 2 23:17:15 sarob: I agree, we don't have to include them in the actual process but socialization would be great 23:18:33 okay, lets follow ttx advice and start talking to the xproject group as we get the rough roadmap, use cases, workflow ready 23:18:57 while also publishing to openstack-dev ML 23:19:07 sarob ++ 23:19:08 ++ 23:19:20 wfm 23:19:23 \o/ 23:20:23 sarob: +1 23:20:45 #action team: once rough roadmap, use cases, workflow ready, we join the xproject meeting and publish URL to openstack-dev ML 23:21:22 what deadline 23:21:29 might want to introduce the product team to the PTLs before that. In case some are up on our group 23:21:31 do we want to give ourselves? 23:22:14 so i think we are making a lot of progress on rough roadmap 23:22:22 Just a hi, we're here, this is what we're working on. the next time you see us here on the agenda will be with the drafts... 23:22:24 not as clear on where we are at on use cases / workflow 23:22:31 rockyg: we should at least have the rough workflow ready for review 23:22:34 for deadline how about intersecting the xprojec team meeting the week of 4/13 or 4/20? 23:22:54 rough roadmap being = documenting existing priorities 23:23:02 def before liberty summit 23:23:03 (for now) 23:23:06 Yep 23:23:18 sgordon: yes 23:23:44 Can we take work flow and rough roadmap (per Steve) to xproject meeting the week of 4/13? 23:23:47 Agreed on both topics (sgordon and barrett1) 23:23:57 4/20 might be hard, depending on how the RCs are going. 4/13 or about 3 weeks before summit would be good 23:24:03 we dont start thinking about new priorities and goal until the PTLs feel we are a responsible group 23:24:21 sarob: +1 23:24:30 4/13 is a month from now 23:24:37 4/13 sounds good 23:24:41 do we need that much time? 23:24:42 It seems like we could help the PTLs by bringing them Operators feedback in a consistent form and create Use Cases that utilize that feedback 23:25:07 i was thinking like next week or the one after 23:25:28 Sooner the better works for me 23:25:45 The challenge is also how much progress we want to make on the coordination piece by liberty. I'd like xproj feedback for sure on that portion of the workflow. 23:25:58 we can loop back to xproject with summit ready stuff for 13apr maybe 23:26:20 shamail: +1 23:26:23 +1 23:26:37 +1 23:27:12 Should we get a repository setup and post our docs there ahead of that meeting? 23:27:19 The one after is good... It gives people time to leverage one full week to try and fill out as much of the rough roadmap as possible but still leaves time for work before summit 23:27:41 barrett1: +1, happy to draft some use-cases as well 23:27:56 barrett1: we are planning on using the xproject repo right? 23:28:15 I'm hoping to have some we can use from WTE too 23:28:41 Does anyone have a format for Use Cases that they have used and like? 23:28:58 well now... 23:28:59 We cod do a directory called product or product-wg under the xproject repo 23:29:08 http://git.openstack.org/cgit/stackforge/telcowg-usecases/tree/template.rst 23:29:09 sarob: OK, will we create a sub dir there? 23:29:11 go, sgordon! 23:29:21 rocky ++ 23:29:38 the main thing i had an AI to add to it (from a telco wg perspective) was personas/user story section 23:29:48 barrett1: yup 23:29:56 sgordon: +1 on the format 23:30:23 Would we have core reviewers in xproj? 23:30:34 steve ++ think the narative is helpful 23:30:44 shamail: ugh, good point 23:30:58 shamail: yeah, we would need to 23:31:04 workflow details 23:31:13 #link https://docs.google.com/document/d/13JPDDiBGGXf5dtP0u8C-1So2Mjb3yEmGhv_ijVqyEf0/edit# 23:32:28 soo, 23:32:30 If we need core in xproj then we probably can't commit before reviewing with them. 23:32:42 shamail: yup 23:33:25 so lets get the rough draft of what the workflow looks like ready for xproject review 23:33:26 On the process, what is the "community voting" in 1.ij? 23:33:31 within one to two weeks 23:33:42 sarob ++ 23:33:50 every pass 23:33:58 the technical community reviews 23:34:03 OK 23:34:09 why the DefCore check? 23:34:20 random thought 23:34:43 that defcore state should be a quick double check 23:34:46 That was teased out through doc comments... Probably no real relation to DefCore arm. 23:34:48 Atm* 23:35:11 Given that DefCore applies to "adopted" capabilities, it's not always likely that these new features will be already included 23:35:12 defcore is going to spit out money 23:35:13 nice 23:35:37 Sarob: 😄 23:36:12 I suggest we take that box out 23:36:14 barrett1: true... at best we might want to scan failed tests from DefCore to maybe find new items 23:36:16 barrett1: defcore and product work are on the opposite ends of the innovation cycle 23:36:34 sarob: Agree 23:36:54 i was keeping defcore in our thoughts as we work through the rough draft 23:36:55 The other check would be to see if defcore thinks it wants to deprecate something we have plans for in the future roadmap 23:36:57 was all 23:37:18 i dont think we should ask defcore for a review 23:37:32 just that the product team check to see the state of defcore 23:37:44 I think we'd need to create a matrix for each use case showing which projects are likely to be impacted 23:37:57 barrett1: +1 23:38:13 barrett1: interesting, id like to see how that would look 23:38:24 We could include it at the end of each use-case or commit message? 23:38:39 use case heatmaps? 23:38:40 sarob - I am working on one for Rolling Upgrades with Zero downtime. I can share it 23:38:53 barrett1: okay 23:39:10 more data good 23:39:17 kewl. I wanna see that one, too 23:39:21 Once we have the Use Case and the Matrix with appropriate core-reviewer approval, would that be the 1st milestone? 23:39:22 * sarob say more data good 23:39:23 Ditto 23:40:07 barrett1: umm, let me noodle that 23:40:41 lets continue this on the ML 23:40:43 My thought is, with that in hand, we could start the process of evaluating existing BPs for changes and creating new ones needed to implement the use case 23:40:50 so we can finish the agenda 23:41:13 sounds good 23:41:49 #topic state of tags, our purpose/focus 23:42:10 #link http://lists.openstack.org/pipermail/product-wg/2015-March/000198.html 23:42:42 who wants to take this up for the group discussion? 23:43:14 well, tl;dr we have a big tent and people coming into it to see what's what want some guidance via tags 23:43:15 I can start but Rockyg and sgordon plz chime in 23:43:26 what tags, who defines them, etc. 23:43:31 is pretty much a blank slate 23:43:41 in particular there does not seem to be a rush of people to be the who 23:43:59 There was a discussion on a specific tag "production ready" that Rocky sent over... I think we all agreed on the cons associated with such a tag. 23:44:01 My thoughts are that we have a unique perspective in that we are some of the most experienced in delivering product as opposed to code 23:44:13 From the Ops Meetup - it seems like there is a desire from Operators for a small form of an integrated release. And tags that describe the interoperability of other projects with that core. 23:44:26 * apologizes for cell phone lag... Listening to others now. 23:44:41 I wonder if there needs to be 2 (or more) classes of tags - One for developers and One for Operators? 23:44:48 the other variation of this discussion which was raised on the -dev thread 23:45:08 was whether each of the "old" criteria the TC evaluated should be considered to see if it should be a possible tag 23:45:13 Shamail, and we also know the pitfalls of assigning the wrong words to a collection of code ;-) 23:45:29 :-) 23:45:43 barrett1, i see it as different groups defining tags rather than classes of tags 23:45:56 so the user group would be a good example of a body that likely should set some 23:45:58 sgordon: what's an example of a group? 23:46:06 similarly i think this body is well placed to provide a different perspective 23:46:37 If the Operators defined an OpenStack-Core tag, how would they cause it to be used by the developers? 23:46:49 sgordon: I think the TC have an idea of something like foundational, then other tags that can be applied to other projects that meet the old TC criteria but don't sound like check boxes 23:46:54 what does used mean here barrett1 23:47:11 i mean really, to me, "using" them means going to the list of projects 23:47:19 and see ooh this has tag-1, tag-2, and tag-3 23:47:30 Having the tag associated with portions of code 23:47:48 but we're not talking about tagging portions of code no? 23:47:53 we're talking about tagging projects? 23:48:16 entire projects? 23:48:17 if we have to have the tags set in code then i think we have a problem 23:48:18 Please let's not use "core". It's overloaded already. and if it were used, it would be nova, glance, keystone and either nova-network or neutron and cinder or swift 23:48:31 Current scope is project... :) 23:48:36 yeah 23:48:51 rockyg: Based upon the the Ops feedback Horizon would be in that set too 23:49:09 I think everyone has good points on tags and possible organization of tags... I would like to add another question: what's product WG's role (if any)? 23:49:37 well i was thinking about this earlier 23:49:55 Would our role be in using Tags to help Operators find implemented Use Cases? 23:49:55 primarily i think we need to determine what we care about 23:49:56 We can always participate as individuals or parts of other teams... But do we need a WG stance? 23:50:06 e.g. stable api (i can use n-1 client against this project) 23:50:15 just as an example 23:50:42 we also dont necessarily have to decide right now, we may come up with common things we care about exposing via tags later 23:51:04 my only concern (more generally) is again that in the absence of *anyone* stepping up and defining the tags it's kind of the wild west 23:51:12 sgordon: I like the idea of using tags to communicate what's important to Operators. 23:51:12 and that is what the operators seemed to be concerned about as well 23:51:27 +1 23:51:34 Devils Advocate: that is moving us from a group sharing PTLs thoughts as use-cases to a group actually suggesting roadmap items (and how to tag them) 23:51:41 sgordon: +1 23:52:03 Shamail, debatable - one of the tags from our pov might be "has roadmap" ;p 23:52:24 Nice :) 23:52:36 shamail: I'm fine with that in any case. 23:52:37 shamail: If we're not suggesting roadmap items then I don't think we're helping the PTLs and the OpenStack 23:52:55 there's a balance to be struck here but i see it actually as within what we discussed as our remit 23:53:13 barrett1: agreed in the long-run but crawl, walk, run 23:53:19 because it's ultimately about assisting consumers of openstack to interpret what's going on 23:53:30 the roadmap activity obviously strikes at that 23:53:34 i think tagging does too 23:53:44 but like i said that doesnt mean we need a definitive list of tags today 23:53:58 I thought tagging was more intended for people who want to build a version from the trunk...no? 23:54:01 i would imagine the list of tags, more generally, will grow over time anyway 23:54:21 And that they tags could somehow also help out with documentation... 23:54:27 i believe it to be much more general than that 23:54:36 sgordon: for examples of what make good tags, I think the former gradation list, http://governance.openstack.org/reference/incubation-integration-requirements.html, could be turned into tags 23:54:57 So if we decide we want to help with the tagging system... (E.g. Help tame or document the Wild West)... Any ideas on how to proceed? 23:54:57 i mean put it this way 23:55:07 was the integrated release only to help those building from trunk? 23:55:36 sgordon: don't think so... It was also supposed to help consumers know that this code runs in production somewhere 23:55:38 jogo, agreed - just as Rockyg said though that seems to be something actively being looked at in the -dev community already 23:55:45 And follows the OpenStack way 23:55:45 Shamail, right - if even that 23:56:13 depending on which version of the integration requirements was used at a point in time :) 23:56:19 so can i bring it back 23:56:27 to what we are doing 23:56:49 I also think that we may be able to help provide a clearer path between the conversion to big tent and how the tags are used to define and the tent 23:57:01 i dont think tagging has much to do with our group right now 23:57:02 sgordon: yup, but tags along those lines could be useful. I as a dev am interested in the answer to: As an operator I will only look at projects that have the following tags 23:57:28 we are creating a multirelease roadmap 23:57:54 I guess the question is how do we create a link from the roadmap to the projects? 23:58:06 maybe theme based tags could be one of the outputs 23:58:16 but other than that 23:58:21 so someone could actually get the roadmapped capabilities in their deployment 23:58:24 tagging is out of scope for us i believe 23:58:30 The devs know that the one big integration approach is starting to collapse under its own weight and they think tags will help when the tests are more directed to limited collections of projects. They just don't haven't been able to connect the tent with the tags tightly yet. 23:58:41 sarob: that's a good starting point at an intersection 23:59:02 we are in the last minutes 23:59:08 sarob: what's an example of a theme? 23:59:24 telco 23:59:26 CBP 23:59:33 storage cloud 23:59:35 sarob: CBP? 23:59:47 barrett1: the "ilities"... Availability, Scalability, etc. 23:59:50 private cloud 23:59:50 oops, i mean thats a vertical using our 00:00:01 shamail: +1 00:00:17 OOoh, the ilities are good tags to consider 00:00:20 theme is a feature that many verticals have in common 00:00:23 I'll take a pass at editing the work flow per the discussion here and sending out the project matrix 00:00:37 cool 00:00:43 good place to stop 00:00:53 be active on the ML my friends 00:00:56 I have updated a slideware version of the roadmap too - could use comments/edits on that 00:01:04 ++ 00:01:15 #endmeeting