22:01:39 <rocky2> #startmeeting refstack
22:01:40 <openstack> Meeting started Thu Feb  6 22:01:39 2014 UTC and is due to finish in 60 minutes.  The chair is rocky2. Information about MeetBot at http://wiki.debian.org/MeetBot.
22:01:42 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
22:01:44 <openstack> The meeting name has been set to 'refstack'
22:01:56 <davidlenwell> Hello everyone!
22:02:01 <zehicle_at_dell> o/
22:02:11 <rocky2> Hey!  Don't want to be the chair, but....
22:02:15 <zehicle_at_dell> can someone share the link to the page
22:02:25 <davidlenwell> https://etherpad.openstack.org/p/DefCoreRefStackUseCases
22:02:27 <zehicle_at_dell> rocky2, I can chair
22:02:38 <zehicle_at_dell> #link https://etherpad.openstack.org/p/DefCoreRefStackUseCases
22:02:54 <davidlenwell> I need to learn all those bot commands
22:03:35 <rocky2> I'm swtiching back and forth from #meeting to get them;-)
22:03:43 <davidlenwell> zehicle_at_dell: how do you want to start.. my meeting style isn't very formal as you may have seen
22:04:16 <zehicle_at_dell> can we start w/ a status update
22:04:18 <rocky2> Is Catherine here?  I'd love to get more detail on the t-cup effort
22:04:20 <zehicle_at_dell> then move into blocking issues
22:04:22 <catherine|3> Yes
22:04:31 <zehicle_at_dell> and plans for the rest of the week w/ help requests
22:04:39 <davidlenwell> I can go first
22:04:39 <rocky2> Yes
22:04:45 <davidlenwell> w/status
22:05:14 <rocky2> #topic status
22:06:04 <davidlenwell> I've been side lined all week per order of of ceo.. I've managed to free up today to only work on refstack.. I've started my data base designs and so far vomited out about 1.5 paragraphs of text for a higher level doc
22:06:48 <davidlenwell> I usually work really late .. so I imagine come morning there will be an email from me to everyone with further updates and links to documents
22:07:16 <rocky2> cool
22:08:23 <zehicle_at_dell> so, looking at the page
22:08:23 <rocky2> more info or should we move to next person?
22:08:42 <davidlenwell> I think those were the only items from me this week
22:09:17 <rocky2> Yup.  Catherine?
22:09:57 <catherine|3> David connected me to Alex and get a copy of the dockerfile
22:10:00 <davidlenwell> I introduced Alex_H and catherine|3 last week.. and sent over Alex_H's docker file.
22:10:35 <zehicle_at_dell> Alex_H, said that they'd made progress
22:10:38 <catherine|3> I made some modificatio nto the file and able to build a container and run  the tempest test
22:11:07 <rocky2> How long did the test take and what projects were included?
22:11:27 <rocky2> (if you don't have answers, don't worry)
22:12:13 <catherine|3> I am running the test and container from a virutal box from my laptop so it may be slow.  But it ttook about 4 hour to run 1100 test cases
22:12:30 <zehicle_at_dell> catherine|3, that aligns w/ our experience
22:12:44 <catherine|3> how long does it take?
22:12:45 <zehicle_at_dell> what are you using as the target?
22:12:57 <zehicle_at_dell> the team was saying 3 hours
22:13:08 <catherine|3> I build a all in one Hanvana OpenStack KVM cloud in my lab
22:13:08 <davidlenwell> Are you testing against a real cluster or a devstack install?
22:13:14 <davidlenwell> ahh
22:13:16 <zehicle_at_dell> real cluster
22:13:18 <catherine|3> not devstack
22:13:21 <catherine|3> a reak one
22:13:34 <catherine|3> but all in one (controller , compute ...)
22:13:42 <catherine|3> a real one
22:13:46 <davidlenwell> package install?
22:13:55 <catherine|3> yes all go well
22:14:09 <catherine|3> Alex did a good job in resovling the dependency
22:14:16 <Alex_H> thanks
22:14:23 <catherine|3> Hi Alex
22:14:30 <Alex_H> Hye
22:14:43 <zehicle_at_dell> I was thinking to have Alex_H duplicate the tcup results against Rackspace
22:15:06 <catherine|3> all tempest requirement and test_requirement installed with no prolbem.  Test run but slow
22:15:17 <catherine|3> what is t-cup?
22:15:34 <zehicle_at_dell> tempest in a container
22:15:37 <davidlenwell> I'm sure thats because it was virtualized twice.. once to virtual box then in a docker container
22:15:39 <catherine|3> ic
22:15:58 <davidlenwell> if we could get an openstack cluster with docker computer nodes I imagine it would run a lot faster
22:16:03 <zehicle_at_dell> #info t-cup: tempest in a container, upload from probe
22:16:10 <rocky2> Tempest in a teacup is a reference to alice in Wonderland book, also.
22:16:17 <catherine|3> so far i tested with a static tempest.conf file.  Will test with David's generator next
22:16:31 <zehicle_at_dell> davidlenwell, my team is working towards that but it's not going to be available in time for us
22:16:48 <zehicle_at_dell> rocky2, :)
22:16:56 <davidlenwell> I might set up a node in my house to do that for my own tests
22:17:33 <catherine|3> will we generate the dcoker image every time from the dockerfile or will we save an image with tempest pre-isntall?
22:17:33 <davidlenwell> catherine|3: I will need to change that code to not depend on the database.. but to be fed creditials while its being generated ..
22:17:42 <zehicle_at_dell> how far off is being able to expand the test audience for tcup?  I'd like to have some people give it a try
22:17:53 <davidlenwell> we decided a while ago it was best to leave creditials out of the database
22:18:04 <zehicle_at_dell> +1
22:18:11 <rocky2> +1
22:18:53 <davidlenwell> thats maybe an hours work for me to make it work that way.
22:19:37 <rocky2> might be able to try it here once you think it will likely run
22:19:44 <zehicle_at_dell> would it simplify the tcup work to reduce the # of tests for now?
22:19:53 <davidlenwell> I think with the config generator its an easy thing
22:19:56 <rocky2> great idea!
22:20:00 <zehicle_at_dell> basically, just do a quick pass
22:20:08 <davidlenwell> zehicle_at_dell: I've been running just the api tests
22:20:11 <zehicle_at_dell> then we can get people to try it and also work on upload
22:20:28 <zehicle_at_dell> davidlenwell, makes sense.  that let's people use their cloud accounts
22:20:28 <davidlenwell> but that is because the early model of refstack wasn't designed to run behind enemy lines
22:20:34 <catherine|3> I can try again with just the API test
22:20:39 <catherine|3> and see how long
22:20:44 <davidlenwell> the api tests are super super fast
22:20:50 <davidlenwell> mine usually run in seconds
22:20:53 * zehicle_at_dell does openstack espionage
22:21:05 <davidlenwell> against a devstack install
22:21:18 <davidlenwell> zehicle_at_dell: +1
22:21:23 <catherine|3> There are abot 400+ API test
22:21:25 <zehicle_at_dell> cool, I think we have a natural scope limit that can help
22:21:30 <zehicle_at_dell> can we drop that to 250
22:21:54 <catherine|3> 400+ for JSOB and 400+ for XML
22:21:57 <davidlenwell> its not very time consuming
22:21:57 <zehicle_at_dell> just as a temporary measure
22:22:05 <rocky2> You could leave out the negative tests that have been separated out
22:22:23 <davidlenwell> also this config maker is smart
22:22:25 <catherine|3> no right now I test by path
22:22:31 <zehicle_at_dell> I don't want to keep it small, just to help people who are testing this get fast results
22:22:34 <davidlenwell> it logs into keystone and figures out which services are there
22:22:41 <catherine|3> say if the argument pass in is templest
22:22:42 <zehicle_at_dell> so we can get to uploading
22:22:43 <rocky2> Though, currently most of the api tests are not parametric, so they should be pretty fast.
22:22:44 <davidlenwell> when it builds the config .. it leaves out stuff it knows the cloud doesn't have
22:22:51 <catherine|3> then it would be about 1200+ test cases
22:23:01 <zehicle_at_dell> how about this, let's try to get the run down to under 30 seconds for now
22:23:04 <davidlenwell> does that make sense ?
22:23:13 <zehicle_at_dell> and then we can expand it once we have the infrastructure working
22:23:32 <catherine|3> If I just run tempest.api.images
22:23:54 <catherine|3> that would be about a few mins
22:23:56 <rocky2> zehicle_at_dell: +1
22:24:16 <catherine|3> I can drop dowm more by just test a handfule of test case with
22:24:27 <davidlenwell> the extent of my testing with what refstack can do now has been a vmware node on my laptop testing another vmware node running devstack.. so results may very
22:24:44 <zehicle_at_dell> yy
22:25:15 <zehicle_at_dell> once we get the workflow between tcup and refstack working, then we can crank it back up'
22:25:19 <davidlenwell> I think for getting the software working .. limiting the tests is fine
22:25:24 <catherine|3> Zehicle_at_dell What is the status of mapping testcase to capability?
22:25:26 <davidlenwell> yeah .. that
22:25:39 <catherine|3> yes we can limit the test
22:25:46 <catherine|3> for now
22:25:46 <zehicle_at_dell> #agree limit # of tests while we get integration between t-cup and refstack going
22:25:50 <davidlenwell> once the workflow is there I don't care how long tests take
22:25:54 <davidlenwell> I want all the tests
22:26:17 <zehicle_at_dell> once we have the workflow working for our team, I'm planning to start expanding the number of test people
22:26:18 <rocky2> I have an old Dell Studio XPS with 6g that will be my "lab" and "outside" machine.  I plan on running at least some devstack based refstack tests on it.
22:26:24 <catherine|3> so we test all test and filter the result per DefCore definition of capability?
22:26:29 <zehicle_at_dell> I'd like to get about 10 people in the initial pass
22:26:37 <zehicle_at_dell> then open it up and see if we can get 100+ scans
22:26:54 <davidlenwell> I agree with that
22:26:56 <zehicle_at_dell> but we have to be confident that it's working - dont wnat to have to staff a support desk
22:27:05 <davidlenwell> I think we have that now between you guys and piston
22:27:15 <rocky2> When we get close, I think I can get a nodepool setup for testing.  We can decide a couple of configurations and either crowbar or compass a couple of configs that we can use for testing.
22:27:18 <davidlenwell> we have running clusters we can test againts at anytime
22:27:35 <zehicle_at_dell> I'd like to poke the TC and Board to give it a try
22:27:44 <zehicle_at_dell> between them, we should find 10-20 takers
22:27:51 <zehicle_at_dell> and a friendly audience
22:28:00 <rocky2> very true
22:28:32 <zehicle_at_dell> do we need to coordinate more on activity for tcup?  it seems like the right people are talking
22:28:40 <catherine|3> So far we can only run test.  Not dealing with the results yet
22:28:45 <davidlenwell> I agree..
22:28:56 <davidlenwell> that is the right path catherine ..
22:29:04 <rocky2> yes
22:29:04 <davidlenwell> dealing witht he results is also a solved problem
22:29:14 <catherine|3> so the results are still i nthe container.  We need to move it out some wjhere
22:29:17 <davidlenwell> I've written code than can extract the results from rest repository
22:29:34 <davidlenwell> I just need to add the part that sends it home to refstack
22:29:39 <davidlenwell> which is easy once the api is finished
22:29:53 <catherine|3> yes but test repository is now in the contaimer.. e need to resovle container id , ip ....
22:29:59 <davidlenwell> so you are right to concentrate on the testing .. get that working .. I'll focus on the api .. get that working and we can meet in the middle
22:30:05 <zehicle_at_dell> let's see if we can get the tcup documented so other people can try it
22:30:13 <zehicle_at_dell> and make upload from tcup a goal for next sprint
22:30:16 <davidlenwell> so I'm thinking some pyton code will get triggered from the container when its finished
22:30:19 <catherine|3> yes my next focus is fetching data out of container
22:30:36 <davidlenwell> that code will call home
22:30:55 <zehicle_at_dell> catherine|3, do we have enough API for you to start that?
22:31:14 <davidlenwell> https://github.com/dlenwell/refstack/blob/master/refstack/common/tester.py  shows how to get data from testr after the test is run
22:31:24 <zehicle_at_dell> :)
22:31:40 <davidlenwell> but we'll need to finish the api methods for recieving data on the refstack side
22:31:42 <zehicle_at_dell> #topic API for uploading
22:32:17 <davidlenwell> So that is the restful api that I am supposed to of design last week.. I am mid-design on it right now today and will have a draft at the end of my day.
22:32:46 <davidlenwell> it will include methods for recieving results from tcup
22:33:01 <rocky2> Do we have an idaddress to call "home"?  I'd like to look at result files, processed and not when we get to that point.
22:33:02 <davidlenwell> the results them selves probably have enough info to assosiate them the right way
22:33:03 <catherine|3> rigjt now I am testing stand alone not touching API yet
22:33:10 <davidlenwell> I think it will be api.refstack.org
22:33:22 <rocky2> great!
22:33:23 <zehicle_at_dell> +1
22:33:29 <davidlenwell> catherine|3: thats good because the api doesn't exist fully
22:34:12 <catherine|3> Perhaps we should have a technical meeting sometime to dive deep some nore
22:34:13 <zehicle_at_dell> davidlenwell, I'm assuming that the POST can stand alone?
22:34:17 <davidlenwell> so one task is to create a "called when tests are finished" script..  python would be idea since I've already written the parts we need
22:34:36 <davidlenwell> zehicle_at_dell: That is a safe assumption
22:34:37 <rocky2> +1
22:34:57 <zehicle_at_dell> catherine|3, I can work to arrange.  Topics you'd like to cover?
22:35:19 <davidlenwell> I thought this was a technical meeting ?
22:35:47 <zehicle_at_dell> it is - I think catherine|3 is thinking something face-to-face or voice w/ a white board
22:35:53 <davidlenwell> oh
22:35:58 <davidlenwell> I'd love to work on that
22:36:06 <catherine|3> Topic is the flow of integration ... Wjere is the host that hosted the container.. where the test data destination ...
22:36:09 <davidlenwell> piston would love to host it also if sf is easy travel
22:36:17 <rocky2> I'd like to have a session on how we specify/subset tests from tempest
22:36:29 <davidlenwell> catherine|3: so that part of refstack is very flexible
22:36:44 <rocky2> SF works for me
22:36:54 <zehicle_at_dell> We're having a board meeting on 3/4
22:37:00 <zehicle_at_dell> we could do it on 3/3 or 3/5
22:37:05 <catherine|3> Where are you located at David?
22:37:13 <zehicle_at_dell> was planning to have a DefCore meeting anyway in SFO on 3/3
22:37:17 <davidlenwell> I live north of LA but work for piston in SFO
22:37:19 <zehicle_at_dell> so this would fit very nicely
22:37:37 <davidlenwell> I have to be in sf for an ironic sprint 3/5 and 3/6 anyways .. so that works for me too
22:37:43 <catherine|3> I leave one hour away from SF
22:37:51 <catherine|3> I live
22:38:28 <davidlenwell> awesome .. so I am happy to have and host refstack tech sessions anytime
22:38:30 <catherine|3> so I thnk that works but that woild be a month from now
22:38:51 <zehicle_at_dell> Let's plan for 3/3  - I may be pulled into other DefCore work at times
22:38:54 <catherine|3> meanwhile we IRC and email?
22:38:54 <davidlenwell> I'll be in sf 2/16 - 2/20
22:39:16 <davidlenwell> if you can make it into the city we could have a live hacking session.. maybe 2/19
22:39:17 <zehicle_at_dell> I'm not required if you want to meet
22:39:25 <catherine|3> oh may be we will visit you in SF 2/16 - 2/20?
22:39:30 <zehicle_at_dell> but I may be able to drag Alex_H to the 3/3 meeting
22:39:42 <davidlenwell> zehicle_at_dell: I'm good with doing it both times
22:39:47 <rocky2> A mini-hackathon for Davidlenwell and Catherine|3 mid Feb?
22:39:47 <catherine|3> that is great ..
22:39:55 <zehicle_at_dell> let's lock in the 3/3 meeting so I can start working on logistics
22:39:57 <davidlenwell> rocky2: yes .. that sounds great
22:40:06 <catherine|3> +1
22:40:08 <zehicle_at_dell> if you can pull together one 2 weeks before, that would be great
22:40:18 <davidlenwell> yeah .. lets do both
22:40:36 <rocky2> +1 twice
22:40:47 <zehicle_at_dell> Alex_H, can you check your schedule?  we probably can have you tag along if we work it in advance
22:41:09 <Alex_H> ok
22:42:03 <zehicle_at_dell> I'm always behind on the #topics!  should have #topic f2f meeting!
22:42:44 <zehicle_at_dell> davidlenwell, when you have the upload API, can you also make it public for POST?
22:42:51 <davidlenwell> yes
22:42:58 <rocky2> OK.  Well, we seem to have that worked out, so how about #topic Wiki ?
22:43:04 <davidlenwell> I'll just set it up to accept any subunit in a post
22:43:08 <zehicle_at_dell> #topic wiki
22:43:27 <davidlenwell> I should be able to grab the info about who its from and when it was run and all that fun data from the rresults
22:43:42 <zehicle_at_dell> #agree API to POST tempest results will be setup as a public endpoint so we can test it
22:43:46 <davidlenwell> about running a wiki on refstack .org
22:43:50 <catherine|3> so the latest code is still at https://github.com/dlenwell/refstack ?
22:44:06 <rocky2> Great. Yup.
22:44:07 <davidlenwell> catherine|3: I belive I have the newest code yes
22:44:15 <zehicle_at_dell> why not use the OpenStack wiki?
22:44:17 <catherine|3> ok thx
22:44:19 <rocky2> Once there, I'll start populating like crazy.
22:44:20 <davidlenwell> So that brings me to another topic ..
22:44:39 <davidlenwell> yes .. use the openstack wiki .. thats a no brainer
22:45:02 <davidlenwell> but if we want to start running things that are services I'd rather the infra team runs them .. we can use sub domains
22:45:36 <rocky2> Just link wiki.refstacl.org to wiki.openstack.org/wiki/Refstack (close, but I don't think it's correct)
22:45:53 <davidlenwell> sure .. we can make a cname or something
22:46:04 <davidlenwell> right now joshua has control over the domain name
22:46:27 <zehicle_at_dell> davidlenwell, I'm already talking to the OpenStack foundation about that
22:46:43 <zehicle_at_dell> Jonathan Bryce & Mark Collier
22:47:13 <zehicle_at_dell> we were thinking that we could get it proved out here for this cycle and then start migrating it to $$$.openstack.org
22:47:14 <davidlenwell> its easy to get them to run things if you make a puppet config for it and ask monty very nicely
22:47:42 <zehicle_at_dell> so we'd get under regular governance after Atlanta
22:47:50 <davidlenwell> perfect
22:47:52 <rocky2> +1
22:48:14 <davidlenwell> this is me saying .. I don't want to admin services because I am not best suited as an admin..
22:48:36 * zehicle_at_dell setting up space on OpenStack wiki for RefStack info
22:48:53 <davidlenwell> I'd preffer a change in the services running required a commit to infra-config to change
22:49:02 <rocky2> I've got a colleague that can help getting the stackforge zuul/gerrit/jenkins stuff set up for our code base when we're ready to start firing off some CI tests
22:49:26 <davidlenwell> if we go into official openstack channels we won't need to do that
22:49:37 <rocky2> even better
22:49:38 <davidlenwell> stackforge it
22:49:44 <davidlenwell> essentially
22:50:09 <davidlenwell> if the goal is under governance after atlanta we have to think that way now
22:50:20 <davidlenwell> and not go off and set up our own stuff all over the place
22:50:22 <rocky2> So, would this go under a "program" like QA or Infra?
22:50:33 <davidlenwell> zehicle_at_dell: ???
22:50:51 <zehicle_at_dell> sorry
22:50:59 <rocky2> Doh.  Forget that last thing I typed.
22:52:08 <zehicle_at_dell> we'd need some tests ;)
22:52:23 <davidlenwell> I think by then we'll have lots of results to work with
22:52:42 <zehicle_at_dell> tests of the refstack code base
22:52:53 <davidlenwell> oh that .. yes
22:53:10 <davidlenwell> do we have someone who likes to write tests?
22:53:16 <rocky2> Again, I've got folks.  Just sayin'
22:53:16 <zehicle_at_dell> #link https://wiki.openstack.org/wiki/Governance/DefCoreCommittee/RefStack
22:53:51 <rocky2> Cool.
22:53:58 <zehicle_at_dell> 5 minute warning
22:54:00 <davidlenwell> rocky2: lets talk offline about getting that started
22:54:10 <rocky2> Will do.
22:54:20 <zehicle_at_dell> davidlenwell, +1 on getting started on governance
22:54:37 <davidlenwell> we didn't write tests because in the hacky stage of things before things are defined I feel like its a huge waste of time
22:54:49 <zehicle_at_dell> right now, we're talking about a very small surface w/ POST and basic compare
22:55:03 <davidlenwell> thats doable
22:55:22 <davidlenwell> I can write the test for that before integrating the api code .. which is the order I'd preffer to do them in
22:55:37 <davidlenwell> that way you always have tests for every api call
22:55:41 * zehicle_at_dell sends hug
22:56:10 <rocky2> Since you're writing a chunk of the api right now, just some simple unit tests that generate a happy path output from a fixed input would be a nice start.
22:56:14 <zehicle_at_dell> last comments/topics?
22:56:33 <davidlenwell> lets wrap up in #refstack
22:56:37 <rocky2> I love people who believe in TDD!
22:56:54 <rocky2> And especially those who practice it!
22:57:00 <davidlenwell> like I said it depends on if I am developing a product or if its r&d hacking
22:57:45 <zehicle_at_dell> great meeting everyone - I feel like we got some things locked down
22:57:53 <davidlenwell> agreed
22:57:54 <zehicle_at_dell> looking forward to f2f next month
22:57:58 <davidlenwell> same
22:58:04 <catherine|3> Will we continue meeting next week or on #refstack?
22:58:06 <rocky2> Understood.  That's why most GUI automated test creation is a waste.  It's always "R&D" so why waste the extra time?
22:58:16 <zehicle_at_dell> please keep #refstack open
22:58:30 <zehicle_at_dell> we should use this channel for official scheduled meetings
22:58:31 <rocky2> Yes.  Next week.  Thanks all!
22:58:38 <zehicle_at_dell> #endmeeting
22:59:13 <zehicle_at_dell> #endmeeting refstack
23:13:59 <clarkb> #endmeeting