17:02:21 <thinrichs> #startmeeting CongressTeamMeeting
17:02:22 <openstack> Meeting started Tue Feb  3 17:02:21 2015 UTC and is due to finish in 60 minutes.  The chair is thinrichs. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:02:23 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:02:26 <openstack> The meeting name has been set to 'congressteammeeting'
17:02:55 <thinrichs> Let's get started with status updates.
17:02:58 <thinrichs> #topic status
17:03:06 <thinrichs> People with high priority blueprints should go first.
17:03:13 <thinrichs> cloudtoa_: want to get us started?
17:03:45 <cloudtoa_> I am working on the control-bus now, should have something submitted in the next couple of days (code).
17:04:25 <thinrichs> Nice.
17:04:29 <thinrichs> Anything you need from us?
17:04:45 <cloudtoa_> Not at this time.
17:05:25 <arosen1> cloudtoa_:  sounds good. Looking forward to seeing it on gerrit :)
17:05:33 <thinrichs> Okay.  So that will probably get in close to kilo2 (which is sometime this week).
17:05:53 <thinrichs> zhipeng: since it's late for you, want to go next so you can take off?
17:07:42 <thinrichs> Maybe he left himself logged in.  We'll come back.
17:07:59 <thinrichs> madhumohan: how are the modals coming?
17:09:13 <madhumohan> rebase is giving me a common error for all tests in test_congress.py: Parse Failure...
17:09:44 <arosen1> madhumohan:  is the patch posted on gerrit?
17:10:07 <madhumohan> I just sent a snippet of the error log.... I guess, there should be a small change i have missed in the test setUp() that is causing this error.
17:10:30 <madhumohan> arosen1:  yes... https://review.openstack.org/#/c/144922/
17:12:32 <arosen1> madhumohan:  where'd you send the snipit of the error log? I'm sure we can help you in #congress after the meeting with debugging it.
17:12:49 <thinrichs> Looks like a merge problem.
17:13:13 <thinrichs> I'd fix the pep8 errors (where we're creating a uuid).
17:13:47 <thinrichs> A failure there will cause nearly every test to fail.
17:14:10 <thinrichs> There has been a lot of churn in the policy engine recently.
17:14:35 <thinrichs> I'm happy to help if you like—just let me know.
17:16:39 <thinrichs> madhumohan: anything else?
17:18:17 <thinrichs> zhipeng_: here for Congress?
17:18:39 <zhipeng_> hey
17:18:45 <JimJunXu> Hi
17:18:50 <zhipeng_> yes just finished another meeting
17:18:57 <thinrichs> JimJunXu: hi!
17:19:06 <thinrichs> zhipeng_: how's the policy trigger work going?
17:19:32 <zhipeng_> I'm still working on it, and have to deal with some company upload issue
17:19:46 <zhipeng_> but I will try to upload the first version before the deadline
17:19:52 <zhipeng_> if I could sort that all out
17:20:21 <arosen1> zhipeng upload issues as in there is a firewall in the way or getting permission to contribute?
17:20:27 <thinrichs> zhipeng_: sounds good.  Don't hesitate to ask for help.
17:20:43 <zhipeng_> yes permission issue, mostly procedure stuff I need to follow
17:20:51 <zhipeng_> yes you guys are very nice
17:21:19 <zhipeng_> :)
17:22:23 <thinrichs> madhumohan: do you know how srangare is doing on the aggregates?
17:24:37 <thinrichs> Perhaps he stepped away.
17:24:43 <thinrichs> jwy: how's your progress going?
17:24:57 <thinrichs> On the horizon UI for policy?
17:25:28 <jwy> met with JimJunXu yesterday to discuss having a graphical representation of policies and converting those to datalog
17:26:23 <jwy> there were some unresolved issues still, as with the form builder, such as how to represent joins
17:27:05 <jwy> i think they are still interested in exploring this UI more, though, as a way to abstract policies for the user more than the form builder would
17:27:12 <jwy> JimJunXu: do you want to comment?
17:27:22 <JimJunXu> This is related to the current BP - 'Policy creation in Horizon'.
17:28:07 <JimJunXu> Need a solid congress example and to see what can be done by following the current approach.
17:28:50 <JimJunXu> the major issue in the UI can be representing the 'join' and 'not allow/error' directives
17:29:39 <JimJunXu> we started to look into it.
17:30:33 <thinrichs> I've said this before, but it's worth repeating....
17:31:02 <thinrichs> The UI is super important here.  People are always skeptical that operators/devs/etc. will be able to write policy.
17:31:09 <thinrichs> So we need to make that as easy as possible.
17:31:38 <thinrichs> Let us know if there's anything we can do to help.
17:31:49 <jwy> thinrichs (or anyone else, really): could you give several example policies that cover various data sources to see how that would translate to a graphical rep?
17:32:13 <jwy> and various combinations of joins and "not"s
17:33:26 <thinrichs> I was looking to find the link to our use case document.
17:33:28 <thinrichs> #link https://docs.google.com/document/d/1ExDmT06vDZjzOPePYBqojMRfXodvsk0R8nRkX-zrkSw/edit#heading=h.fbota2qx0jad
17:33:59 <thinrichs> Perhaps it is time to update that so people have a handful of worked out policy examples.
17:34:08 <cloudtoa_> So some of our customers want to use Congress just as a simple "state-checker."  For instance, if something is configured in "x" (like a VLAN on a port) then that VLAN should exist on that port.  So a data source driver examines config, and another examines corresponding state.  Like a verifier that alerts you when something doesn't work.
17:34:39 <thinrichs> Makes sense to me.
17:34:41 <cloudtoa_> These should be relatively simple policies to write and the value and purpose should be clear to an operator.
17:34:54 <cloudtoa_> It might be a good starting point in the UI...
17:35:16 <cloudtoa_> If there is something shaky in OpenStack now that is tedious to check, that could be a good nexus for us.
17:35:37 <JimJunXu> yes, that is one good case.
17:35:51 <jwy> good info, thanks
17:36:14 <thinrichs> Let's try to add that to the use case doc.
17:37:00 <thinrichs> If we used the vxlan example, we're checking neutron for the config, right?
17:37:04 <cloudtoa_> Roger.. will add after meeting.
17:37:06 <thinrichs> What are we checking for the actual state.
17:37:08 <thinrichs> ?
17:37:40 <arosen1> yup sounds like it - the vlan value.
17:37:55 <arosen1> on the switch vs expected from the cloud operator.
17:38:13 <thinrichs> Anyway—if you're going to add it to the doc we'll see what example you come up with, cloudtoa_.
17:38:18 <alexsyip> I think the question is: which part of the system knows if a vlan exists on a particular port.
17:38:57 <arosen1> I'd say probably the datasource driver that talks to what ever knows about vlans.
17:39:09 <cloudtoa_> Well, the VLAN thing is just an example.  Surely there are other things as well.  A VM is deployed but something goes wonky with some other component... it's a complex dist system, there has to be numerous examples...
17:39:24 <alexsyip> So which part of the system knows about vlans ?
17:40:17 <cloudtoa_> Nova can tell you the segment ID associated with a virtual network... that would be source 1.
17:40:43 <cloudtoa_> Then another source would tell you where hosts are connected in the network (via LLDP harvesting, I imagine).
17:41:02 <cloudtoa_> Then another source could tell you what is configured on the switch access-port associated with that host.
17:42:10 <thinrichs> Sounds good.  I'm looking forward to reading more in the doc.
17:42:20 <thinrichs> Let's move on and finish up status updates.
17:42:29 <thinrichs> stevenld: do you have anything to report?
17:42:55 <stevenldt> yes. I'd like to update on murano-driver
17:44:01 <stevenldt> I submitted the code for review for the driver, however I'm being blocked by the issue where Congress is in the tracked projects and python-muranoclient is not in global_requirements
17:44:35 <thinrichs> stevenldt: go ahead and add python-muranoclient to the global requirements.
17:44:48 <thinrichs> It's a .txt file, and you're free to edit it as you see fit.
17:45:22 <thinrichs> You're talking about requirements.txt, right?
17:45:46 <thinrichs> That's where we typically add the python clients we are using.
17:45:46 <stevenldt> would adding python-muranoclient to global requirements require approval from Murano team?
17:46:28 <stevenldt> as well add the python clients to global requirements as we need?
17:46:30 <thinrichs> So you're saying the OpenStack blessed list of possible requirements?
17:46:36 <arosen1> whoops sorry i got disconnected. I'm not sure how much of my stuff posted.
17:46:52 <stevenldt> as well == or we
17:47:06 <arosen1> stevenldt:  i'm running into that same issue for the cloudfoundry client. Is it possible to have the driver and test work in the tree with just mock?
17:47:32 <arosen1> I think that should work. This way we don't have to have these clients in the requirements.txt file for now.
17:48:04 <arosen1> I don't think these stackforge clients are going to be allowed in global requirements anytime soon. At least from how that email on the list sounded.
17:49:11 <stevenldt> arosen1: so you are saying that I should replace the whole muranoclient with a mock?
17:49:24 <kitho> Are there any driver test that is using mock now ?
17:50:02 <arosen2> sorry about that got disconnected again :(
17:50:06 <arosen2> stevenldt:  i'm running into that same issue for the cloudfoundry client. Is it possible to have the driver and test work in the tree with just mock?
17:50:12 <arosen2> I think that should work. This way we don't have to have these clients in the requirements.txt file for now
17:50:17 <arosen2> I don't think these stackforge clients are going to be allowed in global requirements anytime soon. At least from how that email on the list sounded.
17:50:37 <alexsyip> What happens if we add the murano client to our requirements.txt ?
17:50:42 <alexsyip> Does something fail ?
17:51:10 <arosen2> yup
17:51:14 <stevenldt> alexsyip: yes, the tempest fails
17:51:15 <alexsyip> what fails ?
17:51:20 <alexsyip> Why does tempest fail ?
17:51:26 <arosen2> there is a test to ensure requirements.txt makes the reqirement repo
17:51:36 <arosen2> devstack will fail too.
17:52:00 <alexsyip> Why does openstack require that our requirements.txt is a subset of the global requirements.txt ?
17:52:26 <arosen2> stevenldt:  this way you can run openstack on the same box and not have to worry about using virtual env
17:52:50 <arosen2> it ensures all the libs use the same python versions
17:53:13 <alexsyip> we can’t be the first project to run into this.
17:53:43 <alexsyip> How do other projects deal with this issue?
17:54:09 <arosen2> I think there are 3 options
17:54:10 <thinrichs> Clearly we need to work out what to do when we want to integrate with a system not blessed by OS.
17:54:28 <arosen2> 1) remove congress from a being an openstack tracked project
17:54:52 <arosen2> right now jenkins posts patches for us keeping our requirements files in sync with global requirements
17:55:14 <thinrichs> (Running short on time.  Still need to talk about hackathon.)
17:55:15 <arosen2> 2) figure out a way to get the tests to work with out adding this to the requirements file.
17:55:30 <arosen2> k we can sort this out later. We'll need to figure something out for the cloudfoundry driver anyways.
17:55:37 <arosen2> I'll sync up with stevenldt later about this.
17:55:42 <alexsyip> ok
17:55:46 <alexsyip> sounds good
17:55:53 <stevenldt> sure. I'll ask more in #congress
17:56:00 <thinrichs> It's definitely worth discussing.  In the long term we need a solution.  Follow up in #congress.
17:56:12 <thinrichs> sarob: how's the hackathon going?
17:56:21 <thinrichs> hackathon *planning* going?
17:57:37 <thinrichs> sarob: you there?
17:58:00 <sarob> yup
17:58:16 <sarob> i have a room reserved
17:58:21 <sarob> for both days
17:58:32 <arosen2> what are the date we are targetting again?
17:58:37 <arosen2> dates*
17:58:44 <sarob> this thurs and friday
17:58:50 <arosen2> ah okay.
17:59:18 <thinrichs> Have we confirmed people working on high-priority blueprints can attend?
17:59:38 <sarob> no one has confirmed yet
17:59:42 <sarob> :(
17:59:44 <sarob> maybes
18:00:00 <sarob> im asking for 3-4 hour blocks
18:00:11 <sarob> then run over irc and hangout
18:00:30 <sarob> i will forward once i have any confirmations
18:00:46 <sarob> thats all
18:00:49 <thinrichs> Let's follow up on the ML.  We want to make sure we have a good number of people who can attend.
18:00:55 <sarob> other than clarizen fun :)
18:01:04 <sarob> ML, right on
18:01:14 <thinrichs> Anyone working on Congress.
18:01:25 <thinrichs> Out of time.  (Over-time, actually.)  Let's follow up on the ML and #congress.
18:01:27 <thinrichs> Thanks all!
18:01:31 <thinrichs> #endmeeting