19:05:53 <davidlenwell> #startmeeting refstack
19:05:54 <openstack> Meeting started Mon Mar  2 19:05:53 2015 UTC and is due to finish in 60 minutes.  The chair is davidlenwell. Information about MeetBot at http://wiki.debian.org/MeetBot.
19:05:55 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
19:05:58 <openstack> The meeting name has been set to 'refstack'
19:06:09 <davidlenwell> I was catching up on code reviews and lost track of time
19:07:13 <davidlenwell> I am a little dissorginized because I was camping all weekend for my birthday.. and I am still recovering
19:08:49 <Rockyg> o/
19:09:06 <davidlenwell> I'd like to start out by talking about this review.. https://review.openstack.org/#/c/156569/
19:10:01 <slypushenko__> We disscussed it with Vlad
19:10:38 <davidlenwell> it has the plus twos it needs to merge .. I'm of the opinion that we should be doing all tests and produciton design for mysql
19:11:38 <slypushenko__> We decided leave db_setup_wrapper as a single script
19:11:43 <davidlenwell> okay
19:11:55 <catherineD> davidlenwell: I gave a +2 because I think it is good for Local Test.  But I think we should not have that as "gate"
19:12:21 <davidlenwell> I can get on board with that
19:12:31 <davidlenwell> this is why I didn't just merge .. I wanted to talk it though
19:12:37 <slypushenko__> catherineD Why?
19:12:40 <catherineD> I think MySQL is one implementation .. there may be other backend db implemtation for future
19:12:54 <davidlenwell> iike what catherineD?
19:13:25 <catherineD> DB2?
19:13:26 <davidlenwell> ibm db2 ? ;)
19:13:26 <Rockyg> postgres
19:13:35 <catherineD> Rockyg: +1
19:13:36 <Rockyg> ;-)
19:13:47 <davidlenwell> see I don't think we need to come out of the gate supporting all those databases
19:13:54 <catherineD> All I am saying is keep the option open ..
19:14:02 <vladiskuz_> Sorry, but this patch have a bug with migration. I find it just now
19:14:39 <slypushenko__> We should be sure, that upload works fine.
19:15:26 <davidlenwell> So we need to manually test..
19:15:35 <davidlenwell> vladiskuz_:  go ahead and fix what you just found
19:15:35 <slypushenko__> It is a main idea to have functional test in gate
19:15:52 <zehicle> o/  sorry I was late
19:15:53 <slypushenko__> manual test is not an option
19:15:55 <vladiskuz_> we can have different gates to different database in the future
19:16:07 <davidlenwell> vladiskuz_:  ++
19:16:23 <davidlenwell> I do think we should gate on mysql .. ,because thats what we use in production
19:16:30 <davidlenwell> we need to know if something breaks
19:16:44 <slypushenko__> +1
19:17:41 <davidlenwell> okay .. lets move on .. I think thats enough time on this topic .. vlad go ahead and fix the bug you saw .. I want this in the gate
19:17:53 <davidlenwell> lets get an update on uuid progress..
19:17:58 <davidlenwell> #topic uuid
19:18:53 <davidlenwell> hogepodge: can you give us an update please
19:19:16 <hogepodge> The initial work and tagging has been merged.
19:19:33 <hogepodge> Clean up tasks to be done involve documentation
19:19:40 <catherineD> I look at the new test list with uuid ... and seems like the dynamically created test cases are not covered ..
19:20:01 <hogepodge> catherineD: yes, that's also work that needs to be done
19:20:44 <davidlenwell> so what are the work items that need to be completed before refstack can move from the release its using into trunk tempest?
19:24:03 <hogepodge> catherineD: do dynamically created tests fall under api?
19:24:26 <catherineD> no
19:24:44 <hogepodge> davidlenwell: I don't think there are any more uuid work items then
19:24:45 <catherineD> but they are mostly negative test ...
19:24:54 <mtreinish> catherineD: they're only for 2 classes of tests, negative api tests and 1 scenario tests
19:25:11 <davidlenwell> none of those tests are listed in the defcore capabilites list are they ?
19:25:28 <catherineD> mtreinish: yes
19:25:29 <mtreinish> the uuid decorator is not used on the negative auto generated tests, this was a known quantity going into that work
19:25:41 <Rockyg> not for havana, but negative api tests should be in the future
19:26:25 <davidlenwell> we'll need to hash this out and figure some way of identifying them for the future .. for now lets discuss a change in refstack client to use the uuid's
19:28:10 <slypushenko__> It can be done easily
19:28:27 <catherineD> davidlenwell: I will take a look and provide recommendation
19:28:29 <davidlenwell> yes .. I know its an easy change
19:28:49 <davidlenwell> catherineD:  that is what I am looking for .. please make a new story in story board for the change and track it
19:29:08 <slypushenko__> I can do it tomorrow.
19:29:13 <catherineD> For me more important is how the tests section of the core tests should be defnine ... Today it is tests:[test1, test2 ...]
19:30:14 <catherineD> defined
19:30:46 <davidlenwell> catherine .. go ahead and put your thoughts into a story in storyboard and assign the work to slypushenko__
19:30:55 <catherineD> Yes I will ...
19:30:58 <hogepodge> catherineD: feels like parse and generation tools is what's needed. Similar to whats happening with capabilities files.
19:30:58 <davidlenwell> thank you
19:32:21 <catherineD> since we have mtreinish: here ... should we discuss about the issues refstack faces on testing ... https://docs.google.com/document/d/12GUhUphBSzPuQqp7WreZ-DXaINnUWSrYAYb4CyPmpAw/edit#
19:32:39 <catherineD> do we need to open bug in Tempest ?
19:33:47 <mtreinish> catherineD: for the unicode one that's definitely a bug
19:33:59 <mtreinish> there are open bps that will address some of the others as part of them
19:34:04 <mtreinish> like test-accounts part 2
19:34:16 <mtreinish> and improved ssh validation
19:34:49 <mtreinish> and the last one seems like a nova bug
19:34:51 <catherineD> mtreinish: I looked at the ssh validation .. mostly they are fro scenario tests ...
19:35:20 <catherineD> maybe it will apply to API tests ... I will take a closer look ...
19:35:28 <Rockyg> gotta run. will catch up on my return
19:35:42 <mtreinish> catherineD: http://specs.openstack.org/openstack/qa-specs/specs/ssh-auth-strategy.html
19:37:00 <catherineD> mtreinish: thx..
19:37:18 <davidlenwell> okay .. lets move on ..
19:37:41 <davidlenwell> catherineD has started workign on creating wireframes for some of the upcomming ui work..
19:37:53 <davidlenwell> catherineD:  I will review those today
19:38:44 <slypushenko__> Can it be share with team?
19:39:02 <davidlenwell> sure
19:39:16 <davidlenwell> its still early .. but I see no reason that it can't be done in the open
19:39:25 <davidlenwell> catherineD: would you share the link you sent me before?
19:40:03 <catherineD> First pass mockup https://drive.google.com/file/d/0BxvL0emVRKoUVW13TW5BWXg0MXM/view?usp=sharing
19:40:50 <catherineD> On this first pass, I concentrate more on the type of data that we need to fetch (for API definition) not  look and feel
19:40:51 <davidlenwell> remember these are a first pass
19:41:10 <davidlenwell> look and feel comes later .. lets keep it to pure functionality in the early mockups
19:41:12 <slypushenko__> ok, thx
19:42:42 <davidlenwell> okay .. lets move onto open discussion
19:42:47 <davidlenwell> #topic open discussion
19:43:21 <zehicle> we're going to start moving some of the defcore items out of the refstack repos
19:43:38 <zehicle> just a heads up, no action required here
19:44:09 <zehicle> at some point, I'd like to talk about how the UUIDs will change the JSON files
19:44:18 <davidlenwell> great
19:44:39 <davidlenwell> im sure the moves will have coorasponding commits deleting the files from refstacks repo?
19:45:23 <catherineD> zehicle: how UUID changes the JSON files is what refstack-client needs for update .... have we decided on how?
19:45:26 <slypushenko__> zehicle I think we just should replace test names with uuids
19:46:01 <catherineD> slypushenko__: I think the long test id should still be there for description purposes ..
19:46:11 <slypushenko__> and update capabilities.html page
19:46:53 <slypushenko__> catherineD No. we can easily resolve uuid to test name using github repo
19:47:12 <hogepodge> catherineD: +1 on the long name
19:47:23 <zehicle> catherineD, I was assuming we'd just change them
19:47:27 <zehicle> we'll need a xreference too
19:47:32 <hogepodge> catherineD: slypushenko__: uuid should be only source of truth.
19:47:42 <zehicle> I was thinking the xref file would have extra info
19:47:47 <zehicle> like links to the tests
19:47:50 <catherineD> I can see some thing like uuid: [testid1, testid2 ..]
19:48:04 <slypushenko__> If we keep long names we should care about its consistency
19:48:24 <hogepodge> slypushenko__: if we tag tempest hash at top of file that becomes an anchor
19:48:43 <catherineD> test1, test2 is accomodated the path relocation ...
19:48:53 <hogepodge> the assumption is that the only source of truth in test identification should be uuid, but we need a human readable component.
19:49:17 <zehicle> remember, Tempest may not be the only test source
19:49:27 <zehicle> esp when tests move out of Tempest
19:49:36 <slypushenko__> if we have tempest tag and uuid we can resovle long names automaticaly... So we don't need to keep it
19:50:36 <catherineD> slypushenko__: that is good for display not for subunit results ...
19:51:16 <catherineD> subunit results are still parsing the test_id and then UUID ...
19:51:18 <davidlenwell> zehicle: if tests move out of tempest they will still have uuid
19:51:30 <zehicle> yes.  that's what I expect
19:51:40 <zehicle> we're "coding" that into the process
19:51:43 <davidlenwell> at least thats how we wrote it in the spec
19:51:57 <zehicle> tests must have UUID & be part of OpenStack community process
19:52:19 <hogepodge> zehicle: as a defcore action item for Wednesday we should define a v2 schema for capabilities.
19:52:35 <catherineD> hogepodge: +1
19:52:41 <hogepodge> (heck, maybe every defcore release will require a schema revisit)
19:52:54 <davidlenwell> true sotry
19:52:55 <catherineD> once that is defined it would be easier for refstack-client to make the change
19:53:15 <zehicle> hogepodge, Egle and I were not planning to meet on Wed
19:53:24 <zehicle> you can keep the meeting if you'd like and work on that
19:53:36 <hogepodge> zehicle: ok, good to know.
19:53:48 <hogepodge> zehicle: I keep signing myself up for writing bluebrints. :-P
19:54:41 <catherineD> hogepodge: zehicle: before we run out of time ... I would like to make a comment about someone's suggestion on docker image for tests ...
19:54:43 <zehicle> if you think there's enough people, go ahead and hold the meeting.  there's plenty to do
19:54:49 <zehicle> ok
19:55:53 <catherineD> IMOP, refstack-client is pretty easy to install so having a docker image does not help much ... the hard part is the tempest.conf which docker image won't be able to solve
19:56:34 <zehicle> catherineD, yes.  that's been our experience
19:57:46 <hogepodge> I'm open to new attempts. Maybe they know something we don't.
19:58:04 <hogepodge> But yes, it's a problem. (hence the [interop] tag idea I've been bouncing around)
19:58:15 <catherineD> hogepodge: agree just want to point out the painpoint
19:58:17 <hogepodge> (which would aim to solve the tempest.conf problem)
19:58:36 <zehicle> that's where I'd be interested in seeing if they help.
19:58:47 <hogepodge> catherineD: thanks. total pain point (see the backlog from #openstack-qa over the last hour)
19:58:49 <zehicle> having a container can reduce troubleshooting
19:59:10 <zehicle> I think it would be really really good to have a known working config
19:59:14 <hogepodge> zehicle: a web service would too. :-D
19:59:16 <zehicle> even if most of the tests are not passing
19:59:23 <zehicle> or disabled
19:59:30 <zehicle> so that people can check their install
19:59:30 <hogepodge> interopallthethings.org
19:59:34 <davidlenwell> okay .. we're out of time again .. but I like this disucssion .. lets move it to #refstack
20:00:12 <zehicle> :)
20:01:15 <davidlenwell> #endmeeting