00:00:57 <ekcs> #startmeeting congressteammeeting
00:00:58 <openstack> Meeting started Thu Apr  6 00:00:57 2017 UTC and is due to finish in 60 minutes.  The chair is ekcs. Information about MeetBot at http://wiki.debian.org/MeetBot.
00:00:59 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
00:01:01 <openstack> The meeting name has been set to 'congressteammeeting'
00:01:19 <ekcs> Hi all!
00:01:24 <aimeeu> Hi ekcs!
00:01:30 <masahito> hi
00:02:03 <ramineni_> hi
00:02:20 <thinrichs> Hi all
00:02:24 <ekcs> good to have everyone back =)
00:02:34 <thinrichs> (Sorry I've missed the last couple.  Been a busy few weeks.)
00:02:53 <ekcs> thinrichs: =)
00:03:09 <ekcs> Not much from me in terms of special topics today. #link https://etherpad.openstack.org/p/congress-meeting-topics
00:03:24 <ekcs> please feel free to add your topics. I’ll wait a couple minutes.
00:06:27 <ekcs> ok let’s get started then.
00:06:33 <ekcs> #topic pike-1 milestone
00:07:30 <ekcs> Just a reminder that target week is next week. So let’s submit and review patches intended for pike 1.
00:07:36 <ekcs> anything to discuss there?
00:09:07 <aimeeu> I don't think I'll have the OPNFV Copper bash tests ready by then. I'm running into issues testing them against DevStack master.
00:09:35 <aimeeu> The tests were written for Newton so I have to update many of the calls.
00:10:13 <ekcs> got it. thanks for the update aimeeu !
00:10:38 <ekcs> a little surprised to hear that many things needed change between newton and now. do you have an example handy?
00:11:27 <aimeeu> deprecated API calls in neutron
00:11:48 <aimeeu> and also some of the images used by glance have changed to newer versions
00:13:04 <ekcs> oh I see. So not in the interactions with congress then.
00:13:37 <ekcs> but still part of the test case.
00:14:16 <aimeeu> yes - set up
00:15:26 <ramineni_> ekcs: that reminds recently someone reported bug , congress desn start if glance service is not up
00:15:49 <ekcs> aimeeu: got it. well thanks! so your current step is update the copper bash tests to work in master. then after that we’ll look at how to do it in tempest?
00:16:16 <ramineni_> ekcs: anyone noticed this type of error, bit strange
00:16:34 <aimeeu> ekcs: yes. I plan to upstream the tests to the examples folder and then figure out how to write as Tempest tests
00:17:31 <thinrichs> ramineni_: that sounds familiar to me, though it's been too long.  Maybe I'm thinking of congress failing to start if you include a driver in the config that doesn't actually exist.
00:17:41 <ramineni_> https://bugs.launchpad.net/congress/+bug/1671344
00:17:41 <openstack> Launchpad bug 1671344 in congress "congress service fails to start if glance is not running" [High,Confirmed] - Assigned to Anusha (anusha-iiitm)
00:18:40 <ekcs> aimeeu: sounds good!
00:19:11 <ramineni_> thinrichs: the driver exists and service is not started for the particular service, is it expected to fail?
00:19:36 <ekcs> ramineni_: yes I saw the bug. thanks for looking into it. it says you confirmed it, so you were able to reproduce it? Did the same thing happen with any other service?
00:20:10 <ekcs> I’m guessing the driver threw the wrong error or the right errors were not correctly caught by congress.
00:20:15 <ramineni_> ekcs: yes, im able to reproduce it easily, if you stop glance service and then start congress, congress doesnt start
00:20:23 <thinrichs> ramineni_: maybe in the constructor it's trying to contact Glance, and if that fails it throws an error.  It's possible we've only got try/catch around the poll() call and not around the constructor call.
00:21:25 <ramineni_> thinrichs: ya, thught of the same , actually it is happening in adding executable methods?
00:21:58 <ramineni_> thinrichs: its trying to contact glance then
00:22:01 <thinrichs> In any case, I'd think that datasource failures shouldn't cause Congress to fail to start
00:22:17 <ekcs> thinrichs: makes sense. and seems consistent with the error log.
00:22:23 <ramineni_> thinrichs: ya, will  fix for pike1
00:22:37 <ramineni_> ekcs: not sure happening with other servcies too, will check
00:23:02 <ekcs> thinrichs ramineni_ : yup agreed that datasource failures shouldn’t cause congress to fail to start. I suspect the same issue may be in several other drivers too.
00:23:03 <thinrichs> Inside of harness it looks like we're catching all exceptions.
00:23:52 <thinrichs> https://github.com/openstack/congress/blob/master/congress/harness.py#L155
00:24:07 <thinrichs> Wrong line.  Should be 161
00:24:29 <thinrichs> Nuts—it reraises the exception.
00:24:49 <thinrichs> in 163.  So we should just remove line 163  I'd think.
00:24:49 <ekcs> yup.
00:25:06 <ramineni_> thinrichs: yes, may be logging the exception and and return should do fine
00:25:06 <thinrichs> Who wants to make the 1-line change?
00:25:25 <thinrichs> Probably not return, since we're in the middle of the loop spinning up all the datasources
00:26:12 <ramineni_> thinrichs: right
00:26:50 <ekcs> great glad it’s a simple fix =)
00:27:45 <ekcs> anything else to discuss around pike 1?
00:28:58 <ekcs> ok let’s move on then.
00:29:27 <ekcs> #topic status
00:30:01 <ekcs> Let’s do some quick statuses. Feel free to share things we’ve been doing, looking at, thinking about, or whatever potentially of interest here =)
00:30:11 <ekcs> thinrichs: would you like to start us off?
00:30:18 <thinrichs> Sure
00:30:42 <thinrichs> Been underwater of late.
00:30:50 <thinrichs> Trying to get back to adding policies to the library.
00:31:06 <thinrichs> And catching back up with where things are going.
00:31:20 <thinrichs> Not a lot to say unfortunately
00:31:27 <ekcs> thinrichs: thanks!
00:31:35 <ekcs> anything interesting from your trip?
00:32:36 <thinrichs> Seeing some good interest in policy in different pockets of the kube community.
00:32:56 <thinrichs> Most keen interest is coming from the federation component of kube.
00:33:04 <thinrichs> I don't know of an OpenStack equivalent.
00:33:32 <thinrichs> Federation is a component that replicates workloads across multiple kube instances/clusters.
00:33:58 <thinrichs> In OS, we have federated identity.  But in kube they federate workload placement.
00:34:08 <thinrichs> Policy is attractive as a way of controlling which workloads get placed in which clusters
00:35:22 <thinrichs> If anyone ever decides to create a federation for OpenStack clusters, I'd say we lobby for Congress to be part of it.
00:35:47 <thinrichs> But otherwise not a lot of carryover into the OpenStack world.
00:36:24 <ekcs> hmm interesting… I don’t understand all of what you said but that’s helpful =)
00:36:32 <thinrichs> :)
00:36:41 <ekcs> oh also, would you like someone to put up the policy library APIs in a spec?
00:36:42 <thinrichs> Sorry it wasn't clearer.  Tried to not use too much jargon
00:37:27 <thinrichs> ekcs: that's probably smart.  I seem to remember volunteering to write the spec, but maybe that's unreallistic
00:37:53 <thinrichs> I can help review of course.
00:38:01 <ekcs> ok great. anyone else want to volunteer for that?
00:39:50 <ekcs> ok let’s move on then =)
00:39:57 <ekcs> I can go next.
00:40:11 <ekcs> I’ve mostly been getting up to speed on ONAP. Attended some of Open Networking Summit this week.
00:40:22 <ekcs> And looking forward to this ONAP mini-summit tomorrow #link http://events.linuxfoundation.org/events/open-networking-summit/program/schedule
00:40:54 <ekcs> Also been doing digging around and reading about best practices to find things to add to our policy library.
00:40:57 <aimeeu> ekcs: you will know more than I do!
00:41:05 <aimeeu> about ONAP
00:41:09 <ekcs> that’s mostly it from me.
00:41:18 <ekcs> aimeeu: haha.
00:42:40 <ekcs> aimeeu: anything you’d like to share status wise?
00:43:32 <aimeeu> I started the generic datasource blueprint - the more I think, the more complicated it gets. :)   Still buried with OPNFV, ONAP, and OpenStack-Helm project work.
00:43:42 <aimeeu> that's it from me
00:44:35 <ekcs> Thanks!
00:45:06 <ekcs> just for some context around ONAP, it’s the merger of open ecomp and OPEN-O. In both cases there is significant role for policies. in the ecomp vision there will be some common policy language that maps to specific policy engines such as xacml, drools, and potentially congress in the future.
00:45:39 <ekcs> I’d also like to get familiar with the requirements for that common policy language and see what ideas congress can offer there.
00:46:08 <ekcs> anyway. masahito would you like to share anything?
00:46:16 <masahito> sure
00:46:31 <masahito> I'm still working on parallel execution.
00:47:16 <masahito> manual testing is done. then I'm trying to make unit tests for it.
00:47:23 <masahito> that's from my side.
00:47:37 <ekcs> awesome!
00:47:53 <ekcs> ramineni_: anything you’d like to share?
00:48:04 <ramineni_> ekcs: sure,
00:48:22 <ramineni_> ekcs: working to get dashboard project onboard, pretty much pacthes are in, testing and devstack integration in progress , may be once this is done ill start with monitoring ui part
00:48:49 <ekcs> awesomeness.
00:49:20 <thinrichs> Nice!
00:49:44 <ekcs> ok then.
00:49:52 <ekcs> #topic open discussions
00:49:57 <ekcs> anything else to talk about?
00:50:34 <masahito> ekcs: Do you plan to have forum session in Boston?
00:51:32 <ekcs> masahito: I did not submit any forum proposals. is that what you mean?
00:52:26 <masahito> Just a question whther we have a forum session or not.
00:54:29 <ekcs> masahito: I don’t think we have a congress specific one planned.
00:55:15 <masahito> ekcs: got it.
00:55:51 <aimeeu> the topics submitted for consideration in the Forum #link http://forumtopics.openstack.org/
00:57:49 <thinrichs> Got to run.  Thanks everyone!
00:57:59 <ekcs> bye thinrichs
00:58:27 <ekcs> thanks aimeeu ! Maybe we could look through it and comment on sessions of interest to us.
00:58:46 <ekcs> seems like there is community review?
00:59:50 <aimeeu> Just reading the emails - "a team of representatives from the User Committee, the Technical Committee, and Foundation staff will take the sessions proposed by the community and fill out the schedule."
01:00:06 <aimeeu> Schedule due out 10 April
01:00:16 <ekcs> ok got it. thanks aimeeu !
01:00:21 <ekcs> alright. out of time. thanks all and see you next time!
01:00:33 <aimeeu> thanks ekcs!
01:01:23 <ekcs> #endmeeting