16:00:12 <adrian_otto> #startmeeting Solum Team Meeting 16:00:15 <openstack> Meeting started Tue Sep 9 16:00:12 2014 UTC and is due to finish in 60 minutes. The chair is adrian_otto. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:16 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:00:18 <openstack> The meeting name has been set to 'solum_team_meeting' 16:00:22 <adrian_otto> #link https://wiki.openstack.org/wiki/Meetings/Solum#Agenda_for_2014-09-09_1600_UTC Our Agenda 16:00:28 <adrian_otto> #topic Roll Call 16:00:34 <adrian_otto> Adrian Otto 16:00:38 <gpilz> Gilbert Pilz 16:00:40 <muralia> murali allada 16:00:42 <james_li> james li 16:00:44 <datsun180b> ed cranford 16:00:53 <devkulkarni1> Devdatta Kulkarni 16:00:53 <mkam> Melissa Kam 16:00:57 <noorul> Noorul Islam 16:01:12 <roshanagr> Roshan Agrawal 16:01:25 <julienvey> Julien Vey 16:01:57 <adrian_otto> hello everyone 16:01:59 <stannie> Pierre Padrixe 16:02:29 <AnishKarmarkar> Anish Karmarkar 16:02:32 <adrian_otto> so first some small talk… Pierre and Julien... 16:02:45 <adrian_otto> in France, are telemarketers allowed to call your cell phones? 16:03:03 <adrian_otto> and if you do get a call, is there anything you can do? Any recourse? 16:03:18 <julienvey> yes, they can 16:03:41 <julienvey> and for phone call I don't know if we can do anything 16:03:59 <stannie> I really don't know 16:04:02 <adrian_otto> Americans get irritated when they get unsolicited calls on their mobile phones 16:04:18 <noorul> so does everyone :) 16:04:19 <datsun180b> That's probably in part due to the fact that we're charged for incoming calls 16:04:20 <roshanagr> It is no different in India 16:04:22 <adrian_otto> to such an extent that we have laws against it 16:04:27 <stannie> we usally don't do anything :( 16:04:34 <dimtruck> Dimitry Ushakov 16:04:39 <adrian_otto> interesting, ok. 16:04:58 <adrian_otto> Anyone else who would like to be recorded in attendance, please chime in at any time 16:05:08 <adrian_otto> #topic Announcements 16:05:09 <gpilz> as a public service, you should keep them on the phone as long as possible - then don't buy anything 16:05:16 <adrian_otto> any announcements from teammembers? 16:05:26 <adrian_otto> gpilz: :-) 16:06:12 <adrian_otto> #topic Review Action Items 16:06:15 <adrian_otto> (none) 16:06:26 <AnishKarmarkar> not for cell phones but if you use voip, forward them to Lenny (http://www.itslenny.com/) 16:06:32 <adrian_otto> #topic Blueprint/Task Review 16:06:44 <adrian_otto> Anish, that's brilliant! 16:07:17 <noorul> where is the bot which shortens the link? 16:07:39 <adrian_otto> #link http://www.itslenny.com/ 16:07:54 <adrian_otto> hummm, maybe that is not turned on in this channel 16:08:24 <adrian_otto> ok, any feature development work, tasks, or bugs that anyone would like to raise for group discussion? 16:08:53 <gpilz> I really need to get this reviewed: 16:08:55 <gpilz> https://review.openstack.org/#/c/117056/ 16:09:03 <noorul> I have a question, but I think I should leave it for Open Discussion 16:09:05 <gpilz> I've got other work in the pipeline that depends on it 16:09:17 <gpilz> a schedule, and a boss breathing down my neck 16:09:23 <noorul> 1818 lines 16:09:28 <adrian_otto> Gil, thanks for raising that 16:09:49 <adrian_otto> yes, it is a big patch which makes it difficult to review quickly 16:09:55 <julienvey> gpilz: will take a look, but you can still create other patches on top of that one 16:10:02 <gpilz> noorul - it looks like a lot, but there isn't a lot of complexity there 16:10:04 <adrian_otto> I have sat down to look at it, and got distracted with other tasks before finishing more than once 16:10:08 <datsun180b> it's mainly structure 16:10:08 <PaulCzar> I would like to see this reviewed and merged in, fixes problems with 14.04 ( the gate, etc ) - https://review.openstack.org/#/c/119391/ 16:10:32 <adrian_otto> so in the future, consider submitting groups of patches that are smaller 16:11:32 <adrian_otto> ok, let's take a moment to review each of those now 16:11:40 <noorul> ok, its approved 16:12:21 <adrian_otto> ok, so 119391 has been processed 16:12:37 <datsun180b> nice 16:12:41 <adrian_otto> so now we have 117056 16:12:45 <datsun180b> that means we get one more green gate 16:12:49 <gpilz> thanks 16:13:20 <adrian_otto> Gil, I have some comments on your patch 16:13:30 <gpilz> okay 16:13:31 <adrian_otto> so I'll make those and follow up with you after we adjourn 16:13:32 <devkulkarni1> gpliz: by moving assemblies to camp/v1_1 16:13:37 <devkulkarni1> the client will break 16:13:52 <gpilz> I haven't moved v1/assemblies 16:13:52 <james_li> datsun180b: after a while we can consider make trusty ‘vote' 16:14:04 <devkulkarni1> btw, I will also review the patch in detail later 16:14:13 <devkulkarni1> gpliz: oh okay 16:14:14 <noorul> Why are we not improving v1 ? 16:14:16 <datsun180b> i hope "a while" is "about two hours" 16:14:21 <noorul> Did we release v1 yet? 16:14:28 <gpilz> if you read the spec for Solum CAMP, you will see that the intention is to no affect Solum clients in any way 16:14:45 <devkulkarni1> +1 16:14:51 <adrian_otto> noorul: do you mean v1 of Solum? 16:15:14 <datsun180b> it's just an orthogonal way to present and manipulate the extant resources from what i understand 16:15:27 <noorul> Yes, 16:15:33 <noorul> is v1_1 for CAMP 16:15:38 <noorul> then where is solum version? 16:15:53 <julienvey> i think v1 is the api version, not solum 16:15:53 <gpilz> Solum is …/v1/foo 16:16:02 <gpilz> CAMP is camp/v1_1/foo 16:16:03 <adrian_otto> noorul: good question. "CAMP 1.1" is the OASIS spec version currently in 2nd public review 16:16:07 <noorul> Yeah, I mean solum api version 16:16:10 <adrian_otto> it is not pinned to Solum's versioning 16:16:23 <datsun180b> i think v1_1 tracks only camp and has no deference to solum's own versions 16:16:28 <adrian_otto> v1 was submitted to OASIS like 2 years ago 16:16:31 <gpilz> and you can turn CAMP off 16:16:37 <adrian_otto> but was not proposed as a standard 16:16:48 <adrian_otto> 1.1 will be the first public standard version 16:17:31 <adrian_otto> noorul: does all that make sense? 16:17:37 <noorul> to which version of solum api does this tied to? 16:18:04 <datsun180b> none of them 16:18:04 <adrian_otto> v1 16:18:12 <adrian_otto> but it's another API 16:18:18 <adrian_otto> not just a fascade 16:18:26 <datsun180b> camp's versions will increment whenever oasis wants it to 16:18:36 <datsun180b> regardless of solum's version march 16:18:44 <adrian_otto> correct, datsun180b 16:19:37 <adrian_otto> noorul: clear now? 16:19:48 <noorul> The structure confuses me 16:19:54 <noorul> should that be in solum/camp ? 16:20:03 <noorul> rather than in solum/api/controller/.. 16:20:20 <adrian_otto> (that was one of the things I flagged for comment) 16:20:35 <gpilz> noorul - I need to leverage the existing Pecan controller tree 16:21:02 <gpilz> so I start at the root of Solum - the '/' in '/v1' 16:21:16 <gpilz> that's the perfect place to enable/disable CAMP support 16:21:42 <gpilz> if you disable camp, the root controller just never sees the 'camp' controller - and you get a 404 16:23:42 <devkulkarni1> and python-solumclient does not need to change because it is not supposed to support the camp endpoints 16:23:55 <gpilz> right 16:24:40 <gpilz> also - i need to leverage the unauthenticated access support in solum/api/auth.py 16:25:05 <devkulkarni1> for which part? 16:25:09 <gpilz> because the resources that describe which CAMP versions are supported should be available to unauthenticated clients 16:25:15 <adrian_otto> platform endpoints 16:25:18 <noorul> why do you think you cannot import from solum.api in camp.api ? 16:25:42 <devkulkarni1> gpilz, adrian_otto: ok make sense 16:26:16 <adrian_otto> devkulkarni1: the other reason is for automated version discovery 16:26:22 <gpilz> noorul - it's not a matter of importing - I want to share a single pecan instance 16:28:14 <kebray> gpilz are you asking for solum-camp API to share same single running instance of pecan as non-camp-solum api? 16:29:25 <gpilz> yes - if you read the Solum CAMP spec you will see that it requires cross-compatibility 16:29:57 <gpilz> if you create a plan using the CAMP API, that plan should be visible via the Solum API (authorization policy allowing) 16:29:59 <adrian_otto> kebray: gpilz is referring to the spec in the solum-specs repo for this feature 16:30:17 <gpilz> similarly, if you create a plan via the Solum API, it should be visible in the CAMP API 16:30:45 <datsun180b> #link https://review.openstack.org/#/c/112153/ 16:30:57 <gpilz> the easiest way to do this is share a single pecan instance 16:31:01 <devkulkarni1> thanks datsun180b 16:31:06 <roshanagr> is cross compatibility a requirement? 16:31:08 <adrian_otto> #link http://git.openstack.org/cgit/stackforge/solum-specs/tree/specs/juno/solum-camp-api.rst 16:31:20 <gpilz> roshangar - yes, we don't want to split our user base 16:31:23 <datsun180b> either way, i linked the review of the merged code 16:31:38 <gpilz> CAMP is just an alternate API to the same underlying services 16:31:55 <noorul> so the spec is already in 16:32:00 <datsun180b> right 16:32:01 <gpilz> yes 16:32:31 <roshanagr> gpilz: how does now having cross compatibility split the user base 16:32:49 <roshanagr> as long as a user picks one API and sticks with it, he should be good 16:32:49 <adrian_otto> the lack of compatibility would drive a split 16:32:59 <gpilz> you force users to pick which API they are going to use forever more 16:33:25 <gpilz> either that or force them to move there apps from the Solum flavor to the CAMP flavor or vice versa 16:33:40 <kebray> adrian_otto, then if that's the argument, why not argue for one and one API? Why have two with intention of no split? That implies they should be made the same. 16:33:44 <roshanagr> we give users a choice in the beginning, and once they have made the choice, they stick with that choice. 16:33:49 <gpilz> we want a "many tools / one system" model not a "many systems" model 16:34:04 <roshanagr> what is the overhead of cross compatibility? 16:34:20 <adrian_otto> kebray: solum will effectively be an extended version of CAMP 1.1, but will not be required to follow the spec if it does not align with the goals of the project 16:34:25 <gpilz> roshanagr - there isn't any that i have seen (so far) 16:34:42 <adrian_otto> our community is reluctant to track API's to specifications, so this is the compromise 16:35:20 <devkulkarni1> +1 adrian_otto .. also, if some other specs come along and want to provide a different 'view' of underlying resources I think it should be okay 16:35:26 <roshanagr> gpilz: good to know, thanks. 16:35:28 <kebray> +1 for not tracking the solum API to a spec. 16:35:47 <adrian_otto> with our current approach we don't limit Solum's freedom, and we open the possibility that CAMP compliant clients can use Solum 16:36:44 <kebray> by tying camp API to same instance of pecan, won't that limit Solum's freedom in terms of solum API evolution vs. camp standard (where we are acknowledging we don't want them to be tied). 16:37:06 <gpilz> kebray - the Solum API is free to do whatever it wants 16:37:10 <adrian_otto> that's a slippery slope 16:37:12 <gpilz> the onus is on me to keep up 16:37:27 <adrian_otto> gpilz: well put 16:37:56 <devkulkarni1> kebray: not really. it will be upto each 'view' implementor's responsibility to make sure that their stuff does not break Solum engine 16:38:01 <adrian_otto> it CAMP became an obstacle for Solum, and it's contributors for the CAMP related code are unavailable or unwilling to adapt it, then it could be deprecated. 16:38:11 <adrian_otto> s/it CAMP/if CAMP/ 16:38:32 <gpilz> adrian - that's one reason I allow CAMP support to be disabled with a single-line config change 16:38:33 <noorul> kebray: I have the same concern 16:39:17 <adrian_otto> my advice is not to worry too much about this until it materializes as a problem 16:39:21 <kebray> adrian_otto, seems deprication/maintenace would be easier if it's a separate API instance (tests aren't tied to same running instance,so separate concerns)... also better for the solum service operator too to understand and run. 16:39:25 <devkulkarni1> noorul, kebray: so how do you think this concern can be addressed? 16:39:34 <adrian_otto> worst case we update solum to remove problematic code. 16:39:54 <adrian_otto> best case, we evolve them in a coordinated way 16:40:11 <adrian_otto> I don't think we are going to have major API changes 16:40:16 <AnishKarmarkar> given the cross-polination between Solum and CAMP (not to mention common participants), what is the likelihood that they'll diverge enough to be incompatible? 16:40:33 <adrian_otto> most of the future features will manifest as additions to the DSL, not thew API 16:40:52 <AnishKarmarkar> s/polination/pollination/ 16:41:13 <adrian_otto> AnishKarmarkar: I doubt it will be a barrier 16:41:37 <kebray> devkulkarni1, have camp run as a separate API instance/endpoint, using under the hood public and/only public API calls to Solum where possible.. if something can't be done via public solum API call, then CAMP and Solum aren't align, and for each feature a discussion needs to be had. 16:41:39 <noorul> we started with camp 16:41:51 <noorul> then we had issues, explored workflow 16:42:02 <roshanagr> gpilz: will I be correct to think that the need to share same instance of pecan is driven by need for cross compatibility? 16:42:07 <adrian_otto> most concerns on this subject assume that standards organizations and software development efforts are not in any way coordinated. 16:42:12 <adrian_otto> in this case they actually can be. 16:42:44 <gpilz> roshanagr: yes 16:42:47 <adrian_otto> noorul: those issues were not with ALM 16:42:57 <adrian_otto> they were with workflow, which CAMP is not designed for. 16:43:34 <AnishKarmarkar> keep in mind that the exit criteria for CAMP (any version) is that there be multiple compatible implementation. So in the future, if CAMP does something that can't work with Solum, I doubt it will make it to the final spec 16:43:59 <roshanagr> gpilz: thanks. so cross compatibility does increase the coupling between Solum and CAMP API. Which is where I ask to evaluate the cost benefit of cross compatibility 16:44:37 <roshanagr> We can still have CAMP as an alternate API on Solum, without having tigher coupling or cross compatibility 16:44:45 <roshanagr> tighter 16:44:50 <gpilz> roshanag: keep in mind that it isn't necessarily a single user that is being affected 16:45:03 <noorul> i think 15 minutes more left 16:45:12 <noorul> I have question 16:45:30 <gpilz> if a group of people are working on an application together - you would like to allow each person to use the tools that suit them best 16:45:43 <noorul> adrian_otto: i still remember you saying that RAX is using solum internally 16:45:51 <noorul> adrian_otto: Is there a documentation on that? 16:46:21 <adrian_otto> noorul: only my statements in the announcement section of that meeting where it was mentioned 16:46:23 <devkulkarni1> kebray: sure that is a possibility. it will mean anyone who wants to provide alternative views would not be directly leverage Solum models. may be it might be okay to do that when the view is so different that it will need its own models anyways. 16:46:25 <roshanagr> gpilz: we could have an organization/group discuss and arrive at a single preferred answer to which API they prefer 16:47:04 <gpilz> roshanagr: why make them do that? 16:47:50 <adrian_otto> noorul: I'm happy to follow up with you to give you additional perspective. 16:47:56 <noorul> adrian_otto: ok 16:48:09 <devkulkarni1> roshanagr: reaching such a consensus would be difficult in most cases (different orgs/groups have different needs) 16:48:20 <roshanagr> gpilz: it is a workable compromise between flexibility for the user and implementation separation of concerns 16:49:24 <roshanagr> devkulkarni: yes, however for a given group, it is fine to expect them to arrive at a concensus. Unless we feel the cost of cross compatibility is so low that this is a non issue 16:49:29 <adrian_otto> ok, so in a moment I will bring us to Open Dicussion 16:50:25 <adrian_otto> I think it's worth making an action item to address open concerns regarding the tactical addition of CAMP API support. We can do that as a follow-up IRC meeting. 16:50:31 <adrian_otto> does that sound fair? 16:50:46 <gpilz> sounds fair to me 16:50:46 <devkulkarni1> +1 16:50:49 <roshanagr> adrian_otto: yes, sounds good 16:51:00 <devkulkarni1> I would suggest postpone it to the meeting when adrian_otto you will be around 16:51:29 <kebray> +1 16:51:57 <adrian_otto> #action adrian_otto to schedule an IRC meeting for discussion of CAMP API support, and our tactical integration options. 16:52:20 <adrian_otto> look for an ML thread with a link to a poll for selecting the date/time 16:52:27 <adrian_otto> thanks 16:52:33 <adrian_otto> #topic Open Discussion 16:52:53 <roshanagr> got to go, thanks 16:53:02 <adrian_otto> roshanagr: cheers 16:53:13 <adrian_otto> any other topics for today? 16:54:00 <datsun180b> whoo, cz's patch merged 16:54:08 <adrian_otto> :-) 16:54:16 <datsun180b> can we make the trusty gate voting now? should we? 16:54:36 <adrian_otto> datsun180b: good question 16:55:03 <PaulCzar> we should let it bake for a few weeks 16:55:11 <adrian_otto> do we expect most of Solum's user base to be using trusty as their environment for running Solum? If so, then it makes sense to gate on it. 16:55:27 <datsun180b> well i know i've got problems with my 12.04 vm right about now 16:56:23 <adrian_otto> PaulCzar: I'm happy to leave it as non-voting until we feel confident. I worry about it becoming an impediment to getting code in. 16:57:09 <datsun180b> okay so it bakes 16:57:11 <noorul> adrian_otto: yes and then we have to wait for infra :) 16:57:14 <datsun180b> for how long? 16:57:27 <adrian_otto> noorul: yes, indeed. 16:57:40 <adrian_otto> datsun180b: we can revisit it next week? 16:57:52 <noorul> There is one advantage 16:58:03 <noorul> it will be -1 jenkins job 16:58:03 <datsun180b> sgtm 16:58:28 <noorul> may be the feedback from jenkins will be faster if we could remove fedora one 16:59:08 <adrian_otto> noorul: yes, that's possible, although I think they run in parallel. 16:59:22 <noorul> adrian_otto: what is the benefit? 16:59:29 <adrian_otto> in tha case it would only help if the fedora test were the longest. 16:59:56 <adrian_otto> from a performance perspective. 17:00:08 <noorul> adrian_otto: What is the benefit of having two jobs 17:00:15 <adrian_otto> I'm happy to drop it if we have at least one popular OS doing gate. 17:00:25 <noorul> +1 17:00:27 <adrian_otto> thanks everyone for attending today 17:00:32 <adrian_otto> #endmeeting