19:01:05 <davidlenwell> #startmeeting refstack
19:01:06 <openstack> Meeting started Mon Oct 13 19:01:05 2014 UTC and is due to finish in 60 minutes.  The chair is davidlenwell. Information about MeetBot at http://wiki.debian.org/MeetBot.
19:01:07 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
19:01:09 <openstack> The meeting name has been set to 'refstack'
19:01:40 <davidlenwell> roll?
19:01:41 <catherine_d> hello davidlenwell
19:01:52 <davidlenwell> hello catherine_d
19:02:24 <davidlenwell> You will notice things are merging
19:02:29 <pvaneck> o/
19:02:50 <catherine_d> yes davidlenwell: Thanks a lot!!!!
19:04:34 <catherine_d> hope Rob will join today we need discussion on test list to get juliashapovalov1: going ...
19:05:07 <davidlenwell> we have pleanty to discuss without that
19:05:33 <catherine_d> agree..
19:05:52 <juliashapovalov1> hi all
19:06:00 <davidlenwell> we have nitty gritty things to get into .. but we also need to start the process deciding how we want to display our data
19:06:23 <zehicle> o/
19:06:25 <davidlenwell> juliashapovalov1: before we get to deep into things .. how much longer is on your contract to work on refstack?
19:06:31 <davidlenwell> zehicle: ^
19:07:05 <zehicle> they've got until the end of Nov.  Everyone is welcome to encourage Dell to continue past that
19:07:17 <davidlenwell> I'm sure we will ..
19:07:21 <davidlenwell> but we'll not count on it
19:07:26 <catherine_d> zehicle: that is great
19:07:32 <davidlenwell> how much of your time do we get zehicle?
19:07:34 * zehicle officially thanks Dell for sponsoring work on Refstack
19:07:57 <davidlenwell> +1
19:08:07 <zehicle> I'm about 20% on DefCore/Refstack.  but that's mainly on DefCore
19:08:17 <davidlenwell> okay
19:08:25 <davidlenwell> So we'll lead with that topic
19:08:39 <davidlenwell> #topic defcore update
19:08:59 <davidlenwell> zehicle: whats going on over there?
19:09:09 <zehicle> good news on the DefCore front, we've got a proposal that should work
19:09:28 <zehicle> basically, the Foundation has propose that we add a layer into the DefCore work called "programs"
19:09:38 <zehicle> we keep tests -> capabilities
19:09:51 <zehicle> we add tests -> capabilities -> programs -> platform
19:10:18 <zehicle> basically, it's going to be OK to have a subset of the core capabilities, in that case it would be a "program" trademark
19:10:33 <zehicle> like object (swift) or compute (nova/glance/cinder)
19:10:36 <davidlenwell> interesting
19:10:56 <zehicle> the platform mark would include a superset and have a less restricted trademark
19:10:58 <davidlenwell> okay... So from our perspective that doesn't change much .. it will be up to you how to structure those json files
19:11:13 <davidlenwell> That maps said things
19:11:17 <zehicle> It will likely not change them at all
19:11:33 <zehicle> we're have to add another hash set for programs
19:11:48 <zehicle> but the "core=true" is still valid for the platform mark
19:12:02 <zehicle> and we believe that's the key for interop
19:12:16 <davidlenwell> we == defcore commitie?
19:12:31 <zehicle> we = defcore + foundation
19:12:43 <davidlenwell> So its been voted on and stuff?
19:12:49 <zehicle> needs addittional socialization, but that was what I got so far
19:13:01 <zehicle> davidlenwell, not yet
19:13:07 <zehicle> board meeting is on the 20th
19:13:24 <zehicle> so I've got to get this setup for that.  Foundation is carrying the ball on descriptions of the plan
19:13:46 <zehicle> but I still have to craft up the proposal for board votes b/c the Foundation decription is too complex for a motion
19:13:58 <zehicle> either way, tests  + capabilities are still the core
19:14:01 <davidlenwell> okay .. thanks
19:14:21 <davidlenwell> #topic paris
19:14:24 <zehicle> it just provides an escape hatch for vendors who don't want compute or swift to still use the marks
19:14:38 <davidlenwell> who will be at the summit ?
19:14:42 <zehicle> o/
19:14:51 <catherine_d> o/
19:15:02 <davidlenwell> oh awesome1
19:15:04 <davidlenwell> !
19:15:12 <fcarpenter> I will be in Paris
19:15:17 <davidlenwell> catherine_d: I was worried you wouldn't be able to go because that talk didn't get picked up
19:15:53 <davidlenwell> So qa is going to give us some time in their track to discuss uuid's on tests
19:15:55 <catherine_d> yes they count me in this time ... :-)
19:16:50 <catherine_d> yes ... for this meeting or schedule a separated to go thru the test -> capability mapping and the reason why we need UUID
19:16:52 <davidlenwell> I'll coordinate with them closer to the summit .. lets make sure we can all attend
19:17:12 <zehicle> FWIW, I have an extra pass
19:17:28 <davidlenwell> I want open discussion in a design session about that issue
19:17:56 <zehicle> I know that rockyg will be there too\
19:17:59 <davidlenwell> we can corner folks outside in the hallways or breakout areas to finalize things
19:18:26 <catherine_d> davidlenwell: meanwhile I gave -1 to many of juliashapovalov1: 's reviews we need to discuss to get her going ...
19:18:46 <davidlenwell> wow that is really good news .. I was worried refstack wouldn't have propper representation internationally
19:19:06 <davidlenwell> catherine_d: let me take a look at it
19:19:17 <davidlenwell> after the meeting
19:19:32 <catherine_d> Let me give a quick summary ...
19:19:52 <davidlenwell> #topic open reviews
19:20:16 <juliashapovalov1> catherine_d: I've addressed your comments and sent an e-mail with description also
19:21:27 <catherine_d> I believe for Havana the tests had alreayd been selected by DefCoare in the spreadsheet https://docs.google.com/spreadsheet/ccc?key=0Av62KoL8f9kAdEJTWnFWejJySnFXZmxWdnowTDhUSVE&usp=drive_web#gid=2
19:22:16 <zehicle> and I think juliashapovalov1 has done pulls that match that against the json file
19:22:22 <catherine_d> personally, I do not agree with  juliashapovalov1:'s response  .. as such, we should discuss
19:22:25 <juliashapovalov1> there is no more havana/stable here: https://github.com/openstack/tempest
19:22:47 <juliashapovalov1> so the kist was adjusted to master
19:22:50 <juliashapovalov1> that's all
19:23:05 <juliashapovalov1> only names update
19:23:18 <catherine_d> I did a quick comapre of Havana stable tempest and today's Here is what I see ....
19:24:45 <catherine_d> in Havana there were 1444 tests (including JSON and XML) .... when comparing to today's master tempest ... 772 test match (FQN names) 144 tested had been deleted ....
19:24:59 <catherine_d> 528 tests had been either rename or deleted ...
19:25:57 <catherine_d> the difficult part is the 528 tests ... How do we knwo which one for SURE is just the same test but being rename ... and which one are new tests ..
19:26:33 <catherine_d> we do not have the resource to read the tempest code for the 528 tests ...
19:26:56 <catherine_d> to categorize them into the correct group ...
19:27:44 <juliashapovalov1> i spend a lot of time reviewing them manually
19:28:04 <catherine_d> I know you did ....
19:28:11 <rockyg> o/
19:28:32 <juliashapovalov1> and updated list contains much less then 528 files renamed
19:28:40 <catherine_d> and that is very difficult work ...
19:28:45 <juliashapovalov1> please check the comments
19:29:03 <davidlenwell> can we take this discussion offline in channel after the meeting?
19:29:21 <catherine_d> davidlenwell: I think so ... we need separate meeting ...
19:29:31 <davidlenwell> agreed
19:29:51 <davidlenwell> or a mailing list discussion might suffice
19:30:02 <rockyg> fits?
19:30:06 <davidlenwell> yep
19:30:10 <catherine_d> let's start with that ..
19:30:35 <davidlenwell> catherine can you eamil the list clearly explaining your point of view and reference the review.
19:31:57 <catherine_d> yes I will ... zehicle: I just wnat to confirm ... Havana tests should start from the defcore test list https://docs.google.com/spreadsheet/ccc?key=0Av62KoL8f9kAdEJTWnFWejJySnFXZmxWdnowTDhUSVE&usp=drive_web#gid=2
19:32:09 <catherine_d> is this correct?
19:32:51 <davidlenwell> also remember that we won't merge any file into the defcore files without approval from zehicle or another defcore member
19:33:09 <zehicle> catherine_d, it's not required
19:33:22 <zehicle> the capabilities groups are the key item for this pass
19:33:22 <davidlenwell> zehicle:  who is taking joshua's place on defcore?
19:33:31 <zehicle> davidlenwell, don't know yet
19:33:40 <zehicle> technically, I can run it solo
19:33:46 <davidlenwell> So for now its you
19:33:52 <zehicle> however, since I'm elected, it's good to have a second
19:33:58 <davidlenwell> yeah
19:34:20 <davidlenwell> I'll defer to you on that
19:34:27 <catherine_d> ok then we have a principal differences ...
19:34:41 <zehicle> I don't know who on the board has the time or the interest - it's political risk to take it on
19:35:17 <davidlenwell> okay so this brings me to another topic that I want to be open about
19:35:32 <davidlenwell> our life after defcore
19:37:09 <rockyg> dum da da dum
19:37:11 <davidlenwell> Do our sponsering parents care enough about interop without defcore to continue to allow us to work on it?
19:37:54 <davidlenwell> I can get away with working on it forever.. but I know not everyone shares that level of autonomy.
19:38:15 <rockyg> Good question.  Randy Bias had proposed and interop standard for OpenStack AWS.  I think we could bring those tests in as a separate "interop" thing.
19:38:16 <zehicle> ?? Defcore's an on going process.  I was not expecting it to go away
19:38:46 <zehicle> I do hope that we stream line it.  but RefStack has a long list of use-cases that we are not even touching yet
19:39:03 <rockyg> Yah.  And I think with big tent, interop is going to be much more important
19:39:12 <davidlenwell> thats really my point is that our primary focus should be interop and not defcore
19:39:56 <zehicle> rockyg, w/ the new programs addition into DefCore - we're likely to see more interest on project interop too
19:40:03 <zehicle> which aligns w/ big tent
19:40:20 <davidlenwell> excelent
19:40:21 <rockyg> Interop references.  Sets of tests that define minimum interoperability standards for different kinds of cloud architectures.
19:40:54 <davidlenwell> I'm not sure I'd use the word minimum here
19:41:05 <davidlenwell> or define
19:41:14 <rockyg> :-)
19:41:17 <davidlenwell> test and compare .. all clouds are reletive
19:41:18 <catherine_d> rockyg: who define the set of tests for interops?  could it be DefCore?
19:41:22 <davidlenwell> none is the true standard
19:41:40 <davidlenwell> our job is to show how they are differenent not to tell users which one is the right way
19:42:38 <rockyg> I suspect that defcore with TC and board approval will start the whole thing with the marks, but there will likely end up a committee that sets up "references"
19:42:44 <zehicle> davidlenwell, +1  I think that's the right vision for RefStack
19:42:47 <davidlenwell> If im a company like netflix and I know that my hybrid cloud solution relies on this list of capabilities I need to be able to search and see what my options are
19:42:58 <rockyg> +1
19:43:00 <zehicle> Defcore had to provide a filter, but that's a different layer
19:43:40 <zehicle> it's only relevant to the extent that the tests are a reasonable proxy for interop
19:44:01 <davidlenwell> I'm fine with defcore using our tools to enforce whatever it wants .. but refstack/interop's main goal is to give application developers and operators who want hybrid cloud solutions a map
19:44:07 <zehicle> which is still to be determined IMHO.  I'm hopeful, but I think that we need the community to decide that it's important to invest in tests that do that
19:44:53 <zehicle> davidlenwell, +1 about getting data
19:45:17 * zehicle thinks that we need to make sure the test teams are on board since it's a different use case for their tests
19:45:35 <rockyg> davidlenwell: which is also why refstack just happens to provide aggregate data from the tests; to keep the board satisfied as the interop stuff takes on a life of its own
19:45:40 <davidlenwell> I have a feeling that we'll be spending a good part of the summt discussing this
19:46:16 <rockyg> Maybe we should reach out on either the fits or defcore list to get some agenda started?
19:46:25 <zehicle> that would be great.  I'm more worried there won't be attention to this since there are other fires
19:46:29 <davidlenwell> okay .. So lets move on to open discussion.
19:46:36 <davidlenwell> #topic open discussion
19:46:38 <zehicle> fits list is basically dead
19:46:54 <davidlenwell> lets make a new list
19:46:56 <davidlenwell> who can do that?
19:47:03 <davidlenwell> interop should be a list
19:47:04 <catherine_d> API to upload data
19:47:07 <rockyg> that's a political thing.  It was too hot during the initial work to post there.
19:47:09 <zehicle> what do you want the list to do?
19:47:22 <davidlenwell> replace fits
19:47:31 <zehicle> what do you expect to be discussed there?
19:47:55 <rockyg> let's make an openstack-interop list
19:48:24 <davidlenwell> fits requires too much explination and is attached to a politically hot issue
19:48:39 <davidlenwell> nobody aposed an interop tester
19:49:12 <catherine_d> davidlenwell: they will when we define what is core for interop ...
19:49:19 <davidlenwell> opposes
19:49:36 <davidlenwell> catherine_d: no core
19:49:50 <rockyg> just test sets
19:49:54 <rockyg> and tools
19:50:01 <davidlenwell> compatability mapping has nothing to do with core
19:50:10 <davidlenwell> capability mapping I should say
19:50:43 <davidlenwell> knowing what public and private cloud solutions support what is something that takes a tone of research or operators to figure out
19:51:04 <catherine_d> Let's discuss work items for coming week ..
19:51:18 <rockyg> So, really, other than specific test sets, we would likely want to provide a list of capabilities that map to sets of test to run for those capabilities
19:51:34 <davidlenwell> rockyg: yes
19:52:16 <davidlenwell> ideally in the future those writing tests will some how add to the assosiation so that we are not playing guessing games with new tests
19:52:16 <rockyg> And, then next level up of clumping capabilities.  We have that, others run against and determine whether they are where they want to be.
19:52:54 <rockyg> davidlenwell: I think that's what all the moving and renaming of tests is about in QA/Tempest
19:53:09 <davidlenwell> rockyg: yes
19:53:34 <rockyg> When projects own functional tests, thenit will be even more clearly defined for capabilities
19:54:25 <zehicle> davidlenwell, +1000
19:54:44 <zehicle> I never wanted the board to be playing in the capabilities grouping business
19:55:16 <zehicle> but we had to do it because it was too hard to score otherwise
19:55:27 <davidlenwell> cool .. well we are about out of time and I know the next group starts on time
19:55:35 <davidlenwell> So does anyone have any final thoughts?
19:56:23 <catherine_d> Upload data API ...
19:56:27 <davidlenwell> yes ..
19:56:38 <davidlenwell> I've just fixed my review from last week with a working model
19:56:44 <davidlenwell> I'll post the functions later today
19:56:48 <catherine_d> and refstack.org site do we target them for Paris
19:57:13 <catherine_d> do that we can show our work (at least demo ..)
19:57:15 <davidlenwell> I have an hp cloud account we can host it on until we can migrate to infra
19:57:22 <rockyg> Do we have refstack.org, or are we gonna end up with interop.openstack/org
19:57:29 <davidlenwell> as soon as the api lands I will put it up
19:57:36 <rockyg> interop.openstack.org?
19:57:37 <davidlenwell> we still have refstack.org for now
19:57:46 <davidlenwell> we won't be changing domain names before paris
19:57:48 <rockyg> kewl
19:57:50 <catherine_d> great
19:58:03 <davidlenwell> so catherine_d . go review that
19:58:08 <davidlenwell> and watch for the next one
19:58:10 <catherine_d> I will ...
19:58:17 <davidlenwell> #endmeeting