16:01:33 #startmeeting Solum Team Meeting 16:01:34 Meeting started Tue Jun 17 16:01:33 2014 UTC and is due to finish in 60 minutes. The chair is adrian_otto. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:01:35 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:01:38 The meeting name has been set to 'solum_team_meeting' 16:01:46 #link https://wiki.openstack.org/wiki/Meetings/Solum#Agenda_for_2014-06-17_1600_UTC Our Agenda 16:01:52 #topic Roll Call 16:01:54 Paul Montgomery 16:01:55 Adrian Otto 16:01:59 murali allada 16:02:01 Pierre Padrixe 16:02:01 Noorul Islam 16:02:03 Ed "Ed Cranford" Cranford 16:02:16 o/ 16:02:20 Arati Mahimane 16:03:01 Ravi 16:03:17 Chris Alfonso 16:03:32 Devdatta Kulkarni 16:04:38 how is everyone this fine morning/day? 16:04:48 Anish Karmarkar 16:04:49 doing great 16:05:09 hi anishk.. long time 16:05:23 #topic Announcements 16:05:47 We have new contributors joining the project. Anish and Gil form Oracle will be working with us going forward. 16:05:52 * anishk waves to Devdatta 16:05:59 please extend your welcomes! 16:06:08 cool.. welcome Anish, Gil 16:06:10 neat 16:06:10 welcome! 16:06:11 welcome Anish and Gil 16:06:11 welcome oracles! 16:06:14 Welcome 16:06:15 cool 16:06:15 Welcome! 16:06:17 * anishk is excited to be part of Solum 16:06:21 welcome ! 16:06:23 thanks! 16:06:46 anishk, do you want to take a moment to say a few words to the team while you have their attention? 16:07:04 sure 16:07:39 Both Gil and I joined OpenStack recently and are thrilled to be part of OpenStack and thrilled to be able to contribute to Solum in any which way we can 16:08:02 I know that Ed and I have extended some help to get Gil 16:08:08 We are very much interested in OpenStack and PaaS and have interest in making things interoperaable and portable 16:08:09 up and running with a dev environment 16:08:20 thanks for that 16:08:36 i like maintaining my dev env as a clean room 16:08:44 so if there is anything we can do to help you take flight, please let us know. This is a very helpful group. 16:08:51 i get ruffled if it doesn't "just work" 16:08:54 much appreciated 16:09:04 ok… 16:09:11 #topic Review Action Items 16:09:11 * ravips says hi to Anish and Gil 16:09:12 I should also mention that both Gil and I also participate in the CAMP TC 16:09:42 we made an action item a couple of weeks back that I'd like to check on 16:09:44 devkulkarni to work with adrian_otto to document the LP devleopment approach in BP+task+wiki to clearly outline our approach, and how it solves each use case. 16:10:06 adrian_otto: I will work with you on this in this week 16:10:07 we seem to be stalled on this one? IS that right devkulkarni1? 16:10:23 ok, we don't ahve anything blocking us now, do we? 16:10:25 yes. I have it on my list for this week 16:10:32 ok, thanks, I will carry that one forward. 16:10:33 not directly. 16:10:40 #action devkulkarni to work with adrian_otto to document the LP devleopment approach in BP+task+wiki to clearly outline our approach, and how it solves each use case. 16:10:54 (keeping the same typo, just for continuity) ;-) 16:11:12 #topic Review Tasks 16:11:36 I have not officially announced that we tagged the Juno-1 release 16:11:42 that came out last week 16:11:55 yeah, saw that message on the irc 16:12:01 that was worth having in the announcements section 16:12:05 thanks adrian_otto for that 16:12:17 we did not have a bunch of huge features in it, so I did not make a big fanfare on that one 16:12:35 btw, 16:12:40 the more important release was the one before that which rolled in a whole bunch of bugfix and paid down a lot of tech debt 16:12:52 the juno-1 plan that roshanagr had put on the wiki page had lot more things 16:13:10 devkulkarni1: yes, indeed 16:13:18 I don't think we have finished all those items 16:13:30 I rolled forward all blueprints and tasks, and open bugs into J2: https://launchpad.net/solum/+milestone/juno-2 16:13:56 we have a whole bunch of reviews that are blocked on Mistral 16:14:20 the short story here is that we have showstopper issues with Mistral that have not merged yet 16:14:29 We should bring in someone from Mistral to review them 16:14:33 question. do we know if Mistral supports multi-tenancy?? gokrokve — ?? 16:14:36 and I'm not comfortable taking a dependency on a third party for of Mistral 16:14:47 s/for/fork/ 16:14:48 devkulkarni1: It does 16:15:04 gokrokve: so my understanding is that all desired code is in review 16:15:12 gokrokve: Can you point me to that, the v0.1 documentation doesn't show tenant awareness that I saw 16:15:26 is there any reason to expect that to be delayed for merge, or is this just going to be a short wait? 16:16:32 adrian_otto: you mean the patches that are up on Mistral repo? 16:16:39 can we get pointers to them.. 16:16:41 https://github.com/stackforge/mistral/blob/master/mistral/db/sqlalchemy/models.py#L79 16:17:13 Workbook is tagged by project_id, so users from one project do not see workbooks from another project 16:17:22 gokrokve: thanks. paulmo 16:17:48 yes, so these seem like relatively trivial patches. 16:17:56 gokrokve: But notice that the name field is the primary key in that workbook schema. So we'll have to do something like hash a tenant and append it to the name right? 16:18:07 seems like project_id is there.. but don't know whether the workbook names are unique across tenants or can same name be used for different tenants 16:19:38 sounds like a composite primary key is in order 16:20:30 +1 adrian_otto, or something similar 16:20:32 so let's park this one and come back to it in open discussion 16:20:49 Agree. I dont see that DB queries use project_id as a filter 16:21:13 It looks simple to fix, though 16:21:20 once we get those changes into Mistral, and arrange for a release to be tagged, we can pick op that dependency through pypi on an N.N.N version value 16:21:38 does anyone have any objections to this approach? 16:22:00 should we come up with workarounds that are intended to make forward progress without any waiting for this? 16:22:20 adrian_otto: sounds fine.. we could, will be great if we have asalkeld in that meeting 16:22:23 I'm documenting workarounds current 16:22:24 ly 16:22:48 paulmo: thanks. I love it when you exhibit initiative like that! 16:23:18 (probably just a SHA256 hash of the tenant and append that to the Mistral workbook name btw is my current thought) 16:23:20 devkulkarni1: let's plan to regroup with asalkeld later today, if we are expecting him, and just touch base on this one. 16:23:36 adrian_otto: okay. sounds good. 16:23:57 #link https://launchpad.net/solum/+milestone/juno-2 juno-2 milestone tasks 16:24:14 does anyone want to raise a specific task for discussion today? 16:24:34 or report noteworthy progress against any 16:24:55 sure 16:25:21 so i've been having trouble with the client, and fragments of conversation were something to do with yaml support in solum 16:25:32 i guess everyone in that conversation is asleep, aren't they 16:25:51 yes, unfortunately. 16:25:56 datsun180b: stannie postedpatch 16:26:07 oh did he? i'll take a look 16:26:09 https://review.openstack.org/#/c/100605/1 https://review.openstack.org/#/c/100606/ 16:26:18 thank you for linking those 16:26:35 waiting for jenkins to give green signal 16:26:37 perfect 16:26:38 and there is the one for "list" command that will come soon 16:26:39 #link https://review.openstack.org/100605 16:26:44 #link https://review.openstack.org/100606 16:27:11 ok, and that raises a related topic 16:27:11 well those should resolve my problems manually testing the unittest changes 16:27:21 about functional tests for the CLI 16:27:44 it's an expected behavior, we've to update the CLI with the YAML change in the API 16:27:45 #link https://bugs.launchpad.net/python-solumclient/+bug/1309493 16:27:46 Launchpad bug 1309493 in python-solumclient "Feature: Client changes for Plan YAML support and Plan versionning" [Wishlist,Incomplete] 16:27:52 yes 16:28:02 How would you feel about adding a func test to both solum and python-solumclient that exercise the CLI? 16:28:29 do you mean a func test in python-solumclient ? 16:28:37 stannie, in both 16:28:50 because in this case we change solum, and it regressed python-solumclient 16:29:09 but because we don't test python-solumclient in gate, that defect got in 16:29:33 yes that could really help 16:29:37 Need to take a look at how tempest is doing i 16:29:52 ok, tomblank are you willing to open a task for this one? 16:30:01 I wonder if the Solum CLI should be lower priority given that the Horizon UI (or Mistral CLI) will be the only way shortly to execute something in Solum, correct? 16:30:02 any volunteers for coding this up? 16:30:02 adrian_otto: yes, will do 16:30:03 categorically i don't know that solum ought to bear solumclient tests, but it makes sense that the client should assume a solum endpoint 16:30:13 tomblank: thanks 16:31:15 datsun180b: so what is your recommendation? 16:31:31 We can have the tests in solum client repo 16:31:38 I'd argue that python-solumclient is needed in order for most users to interact with Solum 16:31:44 that the client bear its own functional tests, and require that to run them you have a solum endpoint specified 16:31:45 and execute them for both solum and client gate 16:32:02 noorul: yes, that's what I would be happy with 16:32:18 that's all the client is basically, an http client that understands the context of solum and can parse responses and shape requests to communicate with it 16:32:51 do we not have an API func test in Solum? Is that the problem? 16:33:02 like paulmo mentioned, other uis can supplant it 16:33:39 datsun180b: please help me understand. I'm not following. 16:33:55 python-solumclient isn't (or shouldn't) be required to interact with solum 16:34:12 it's not. 16:34:36 right, so keeping functional tests for it in solum's codebase is what i take issue with 16:34:40 datsun180b: I think though that we can use solumcli as the reference implementation of a solum client, therefore it makes sense to ensure it always works 16:34:56 datsun180b: see noorul's suggestion 16:34:57 s/use/think of/ 16:35:39 keeping the client's functional tests in the client's repo is all i'd suggest. i guess that's violent agreement with noorul 16:35:45 I think it's fine to kick off tests that live in the python-solumclient repo, so that we don't get out of sync when we adjust the API 16:36:04 ok, good, so we are aligned on this 16:36:13 thanks for explaining your concerns datsun180b 16:36:32 any other task updates to touch on? 16:36:37 I am having trouble with the devstack gate. Functional tests are failing for my patch - https://review.openstack.org/#/c/96928/. Functional tests are running fine on mine and Ed's local devstack env. Any pointers would be of great help ! 16:37:14 * noorul is looking 16:37:37 aratim: did you see this: http://logs.openstack.org/28/96928/8/check/gate-solum-devstack-dsvm/f9b9711/console.html.gz#_2014-06-13_16_47_36_421 16:38:09 foreign key constraint violation 16:38:17 yes adrian_otto: I have seen that 16:38:27 But I think the real problem arises here - http://logs.openstack.org/28/96928/8/check/gate-solum-devstack-dsvm/f9b9711/console.html.gz#_2014-06-13_16_47_36_820 16:38:34 what do you make of "Cannot delete or update a parent row: a foreign key constraint fails" 16:38:51 We might need some cascading deletes sprinkled in the schema 16:39:28 paulmo: I have not see that in other openstack projects 16:39:33 I might be wrong 16:39:51 oh these are actually DELETE FROMs and not just UPDATE WHEREs 16:40:08 We had this issue for component 16:40:10 * paulmo nods. I'm not sure if there are unwritten rules around the usage but it is handy if you carefully place them where it makes sense. 16:40:30 * noorul agrees 16:40:33 Otherwise code will have to manually walk the chain and delete dependencies. 16:41:26 paulmo: the functional tests are doing exactly that 16:41:37 They'd have to. :) 16:41:46 +1 aratim 16:42:04 adrian_otto: we can discuss this later on #solum, dont want to take up all the time 16:42:29 aratim: do you want me to record any action items on this? 16:42:46 sure 16:42:55 that would be great! 16:42:57 maybe noorul can offer some advice,and help get you unstuck since we did see this before 16:43:22 noorul: is it okay if I give aratim an action item to follow up with you on this? 16:43:39 The first failure is for component_put and I think that is not related to foreign key 16:43:57 adrian_otto: I am okay 16:44:25 yes noorul, the problem is here - http://logs.openstack.org/28/96928/8/check/gate-solum-devstack-dsvm/f9b9711/console.html.gz#_2014-06-13_16_47_36_820 16:44:37 Assembly is not getting deleted 16:44:38 #action aratim to follow up with noorul to plan a solution to https://review.openstack.org/#/c/96928/ 16:44:53 ok, thanks for that. 16:45:12 any other task updates? We have just one more item on the agenda before open discussion. 16:45:46 ok, moving on 16:45:53 #topic Blueprint Discussion 16:46:19 before our mid-cycle meetup for Icehouse I was conducting a Bluepring review during our team meetings 16:46:32 this gave the team a chance to sync up on our important work items 16:47:01 aratim: http://logs.openstack.org/28/96928/8/check/gate-solum-devstack-dsvm/f9b9711/logs/screen-solum-api.txt.gz#_2014-06-13_16_47_22_808 16:47:06 since we switched to tasks using bug tickets, we have done less sync-up, and I wanted to get all your thoughts about if we should make a change 16:47:31 and put our key pursuits back into blueprints and ask them to be owned and reported on each week by its owner 16:47:49 that approach going give us a better sense of unity in understanding how we are progressing against those goals 16:47:53 what do you all think about this? 16:48:19 we would not use Blueprints as a task list, but as a status instrument 16:48:31 +1 16:48:39 I like this better 16:48:44 i mean the new way 16:49:07 devkulkarni1: ok, why do you feel that way about it? 16:49:14 the reason being, many times we did not have details in the blueprints and so it was not really possible to get the overall picture anyways 16:49:26 It seems like we could collapse some of the tools somehow… we'd have to update blueprints, rst specs and bugs, right? 16:49:26 now we are going towards spec repo 16:50:07 paulmo: we don't necessarily need to update the BPs 16:50:23 but I would use them to form the weekly agenda 16:50:28 the bug tasks help to have granularity, before we had a big bps without granular tasks and it was hard to dispatch the work. Can't we use both system ? Bug tickets for granular tasks and blueprint for highlevel task ? 16:50:36 to address adrian_otto your question about how to get overall picture and status update.. 16:50:38 the other thing I can do is just use the meeting agenda to cover what I think are important topics 16:50:47 adrian_otto: +1. i would like to get back to reviewing/giving updates on the blueprints/specs. we could also review the tasks associated with each BP/spec during that discussion. 16:51:27 I like keeping it informal and trusting adrian_otto (and the entire team) to bring important topics for the meeting discussion 16:51:31 stannie: yes, that's what I'm thinking 16:51:38 ok great 16:51:53 ok, so let's try a half-step at first 16:51:55 stannie: yes, agree... 16:52:13 I will call out some agenda points and just ask for updates on them 16:52:37 the benefit should be that new and less active contributors can stay in the loop with us, and be able to help out more 16:52:59 sounds good. 16:53:01 if we drift too far apart in terms of understanding what's happening, it's hard for those less involved contributors to jump in and help 16:53:22 adrian_otto: good point. 16:53:24 ok, so we will re-evaluate in a couple of weeks and see if we should move further in that direction 16:53:45 I don't want to force updates if they don't make sense either 16:54:21 so if there is something you will be expected to report on, you can update the agenda page, and just put (skip) at the end of one, and I won't raise it for discussion if there is nothing to say about it. 16:54:45 are we happy with this? I'm going to note this with a #agreed 16:54:58 agreed 16:55:02 agreed 16:55:03 +1 16:55:05 agreed 16:55:11 agreed 16:55:13 yeah 16:55:18 +1 16:55:34 #agreed adrian_otto will identify key pursuits for the project, and have them as standing agenda items in our meeting agendas, requesting updates for higher team visibility. 16:55:37 thanks 16:55:41 #topic Open Discussion 16:56:04 anything I could be doing better for you all? 16:56:18 any thoughts on mid-cycle meetup? do we have one? when? where? 16:56:32 i think they have been helpful to the project in the past... 16:56:39 after juno-2 probably.. 16:57:03 that sounds like a good time to meet 16:57:21 late July would make sense 16:57:35 or perhaps the first week of Aug 16:57:36 where? 16:57:42 Austin 16:58:12 It's Rackspace's turn to host, and we have the most attendees in Austin, so that would help us from a budgeting perspective 16:58:15 +1 on Austin. 16:58:19 +1 16:58:23 :) 16:58:28 naturally i have to agree 16:58:36 Bring sun tan lotion! 16:58:52 and plaid shirts 16:59:03 anish_karmarkar: you don't have to swim 16:59:08 * anish_karmarkar first week of aug works lot better for me than last week of july 16:59:17 but let us know if that's something that you think we could arrange 16:59:24 I will circulate a poll to find the best date 16:59:27 with this being holiday season, we should pick dates soon... 16:59:42 as long as there are no violent objections to using Austin as the meeting place this time 16:59:42 adrian_otto: +1 thanks... 16:59:58 BP for supporting private github repos: https://blueprints.launchpad.net/solum/+spec/support-private-github-repos , plz review and let me know if you see any issues 17:00:05 we should invite mistral folks 17:00:11 yes 17:00:13 ravips: awesome!! 17:00:14 thanks 17:00:19 ravips: sweet! 17:00:31 thanks everyoen for attending this week 17:00:38 keep an eye out for the ML 17:00:42 #endmeeting