00:01:46 <thinrichs> #startmeeting CongressTeamMeeting
00:01:47 <openstack> Meeting started Thu Dec 17 00:01:46 2015 UTC and is due to finish in 60 minutes.  The chair is thinrichs. Information about MeetBot at http://wiki.debian.org/MeetBot.
00:01:48 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
00:01:51 <openstack> The meeting name has been set to 'congressteammeeting'
00:02:28 <bryan_att> hi all
00:02:41 <thinrichs> bryan_att: glad you could join us today!  I'll put you on the agenda.
00:02:52 <bryan_att> thanks, I have a few questions
00:03:06 <thinrichs> Here's our agenda for the day:
00:03:12 <thinrichs> 1. distributed arch
00:03:18 <thinrichs> 2. python3 status
00:03:22 <thinrichs> 3. mid-cycle meetup
00:03:28 <thinrichs> 4. Tempest and gate
00:03:36 <thinrichs> 5. bryan_att and install
00:03:39 <thinrichs> Anything else?
00:05:10 <thinrichs> #topic distributed arch
00:05:19 <thinrichs> We now have the core of the new arch up in gerrit:
00:05:32 <thinrichs> #link https://review.openstack.org/#/c/247232/
00:05:42 <thinrichs> masahito left a comment about how to fix the test.
00:06:08 <thinrichs> pballand: did you want to make that change and merge it, or should we merge it now and let masahito fix it?
00:06:14 <pballand> I confirmed the fix masahito pointed out
00:06:32 <pballand> I am now putting all of the other pieces back in
00:06:40 <pballand> I can push the existing skeleton with the fix
00:06:55 <pballand> (it is already in my local repo)
00:07:43 <thinrichs> I jumped the gun a bit.  So that change is ready, and you've got another change you're working on to fold everything else back in.
00:08:12 <pballand> just pushed that fix; still prepping the bigger change with the control bus back in
00:08:33 <pballand> baby steps :)
00:08:36 <thinrichs> Run into any other issues yet?
00:09:00 <masahito> hi
00:09:03 <pballand> no - it appears to be working as one would expect - wish we had know about that sooner
00:09:21 <thinrichs> Moving forward, let's do this….
00:09:29 <thinrichs> We'll merge the patch you just pushed.
00:09:51 <thinrichs> Then let's make sure that we don't get blocked again.
00:10:16 <thinrichs> So if we hit a problem, push what's there to review and crowdsource the problem.
00:10:24 <pballand> sounds good
00:10:58 <thinrichs> Especially given that we have people working around the clock (i.e. in the U.S. and Asia), even pushing problems at the end of the day let's people make progress on them overnight.
00:11:30 <thinrichs> And once there's enough functionality merged that we can try to hook up datasources, API, policy engines, let us know so we can parallellize that effort.
00:12:12 <thinrichs> ekcs: you picked up the remaining distributed arch blueprint.  Any problems/questions/concerns?
00:12:51 <thinrichs> pballand, masahito: I'm glad to see we're making progress again.  Nice work!
00:13:36 <ekcs> thinrichs: As I was constructing test case to frame the problem, I came across this issue that I don’t quite understand yet.
00:14:18 <ekcs> #link https://review.openstack.org/#/c/258711/3
00:14:35 <ekcs> Other than that, things are progressing.
00:16:07 <thinrichs> ekcs: I just saw your email and responded.  See if that makes sense.
00:17:08 <thinrichs> For everyone else….
00:18:34 <thinrichs> I think there's a race condition on how many of the messages get delivered and processed.
00:18:37 <ekcs> thinrichs: kind of makes sense. but based on my prelim understanding of the retry_check_db_equal code, it will retry until all updates have been dequeued and processed.
00:19:25 <thinrichs> Hmmm…. maybe that's not the right answer.
00:20:01 <thinrichs> I thought there was a race condition where when 2 messages are published, you don't know whether they will both be processed or just 1 will be processed.
00:20:22 <thinrichs> But ekcs is right: the code that is checking a condition should wait until the condition becomes true, which should force both the messages to be processed.
00:20:52 <ekcs> sorry for the terse info on the review page. I’ll add more explanation now.
00:20:54 <thinrichs> ekcs: I'd have the test print out the results of that query at every step.  It'll be annoying b/c it prints out a bunch, but it'll show you what it's getting.
00:21:09 <thinrichs> Moving on…
00:21:17 <thinrichs> #topic python3
00:21:34 <thinrichs> ekcs pushed us all the way to python3 compatibility
00:22:15 <thinrichs> I put a patch together to make python34 voting now, which means that code doesn't get merged if it's not python34 compatible.
00:22:17 <thinrichs> #link https://review.openstack.org/#/c/255621/4
00:23:09 <thinrichs> I've had a few issues with it.  If someone can do a sanity check so Andreas doesn't have to find another bug, that'd be helpful.
00:23:48 <thinrichs> #topic Mid-cycle meet-up
00:24:00 <thinrichs> I put all the info for the meet-up into the usual wiki...
00:24:02 <thinrichs> #link https://wiki.openstack.org/wiki/Sprints
00:24:08 <thinrichs> Here's the page…
00:24:18 <thinrichs> #link https://wiki.openstack.org/wiki/Sprints/CongressMitakaSprint
00:24:27 <thinrichs> We have a location, date, time, and agenda.
00:24:50 <thinrichs> Any questions?
00:26:07 <thinrichs> #topic Tempest and the gate
00:26:21 <thinrichs> We had a gate breakage b/c of a keystone test failure.
00:26:30 <thinrichs> ramineni: anything special about this one or your fix?
00:27:01 <ramineni> thinrichs: its again the failure we hit because of migration
00:27:13 <ramineni> thinrichs: migration of tempest
00:27:42 <thinrichs> So we should still expect those from time to time, right?
00:27:51 <ramineni> thinrichs: yes
00:28:09 <thinrichs> ramineni: thanks for staying on top of those!
00:28:30 <ramineni> thinrichs: stable/liberty different job patch got merged
00:28:51 <ramineni> thinrichs: but we still have a problem :(
00:29:14 <thinrichs> ramineni: remind me which one got merged
00:29:47 <ramineni> https://review.openstack.org/#/c/250598/
00:29:52 <thinrichs> Right—the one that uses a different devstack job to test kilo/liberty
00:30:01 <thinrichs> ramineni: what's the new problem?
00:30:07 <ramineni> yes
00:30:18 <ramineni> thinrichs: thought tempest plugin arch is mrged on liberty branch , but looks like it isnt part of stable liberty
00:30:52 <ramineni> thinrichs: i should change the job not to use, my bad should have checked before :(
00:31:28 <thinrichs> How is something merged into liberty but not on the stable branch?  The only branches we have are … master, stable/kilo, stable/liberty.
00:32:04 <ramineni> thinrichs: not sure , https://github.com/openstack/congress/tree/stable/liberty
00:32:31 <ramineni> thinrichs: there is no congress_tempest_tests here
00:33:09 <thinrichs> Maybe that got merged after we cut the liberty branch, and we didn't backport it.
00:33:39 <thinrichs> Well at least the fix is straightforward right?  It just means another review that goes through infra.
00:33:44 <ramineni> yep, i didnt notice it , i thought it merged long before , should be on liberty
00:33:50 <ramineni> yes
00:34:12 <ramineni> will raise one more patch today .. hopefully that should fix the issue on liberty gate failures
00:34:20 <thinrichs> ramineni: thanks for seeing this through.  Sounds like you're on top of it.
00:35:02 <thinrichs> bryan_att: you're next
00:35:05 <thinrichs> #topic Install
00:35:10 <bryan_att> ok
00:35:17 <thinrichs> bryan_att: could you give us some background and then ask your questions?
00:35:24 <bryan_att> sure
00:35:35 <bryan_att> I'm working on a manual/scripted install process based upon the congress readme and the Ansible installer playbook I got from VMWare a while back.
00:35:42 <bryan_att> https://wiki.opnfv.org/copper/academy/joid/congress
00:35:51 <bryan_att> I didn't get all the way (got most of the way) with Ansible so I fell back to scripting this in bash one step at a time.
00:35:59 <bryan_att> I will bring it back to Ansible, then MAAS/JuJu, Fuel/Puppet, RDO, etc so I get the widest OPNFV installer support, over time.
00:36:09 <bryan_att> At the least I will have a post-OPNFV-installer that works with OpenStack as deployed by these different infra mgmt tools.
00:36:17 <bryan_att> For now, I have a repeatable install process, and have passed various adhoc tests plus the "tox -epy27" tests.
00:36:28 <bryan_att> First Question I have, is how the congress tempest tests relate to the the "tox -epy27" tests.
00:36:55 <thinrichs> bryan_att: they are completely unrelated
00:37:05 <thinrichs> the tox -epy27 runs a bunch of unit tests
00:37:24 <thinrichs> Those are all just basic python tests that we run all the time
00:37:36 <bryan_att> OK, I need to know how to integrate the tempest tests into our OPNFV CI process, so a pointer to that in the docs will be appreciated.
00:38:02 <thinrichs> The tempest tests assume there's an entire openstack cloud running and then checks that congress is behaving properly when interacting with the other projects.
00:38:21 <bryan_att> yes, that's our assumption as well
00:38:34 <thinrichs> You would integrate the Congress tempest tests the same way you'd integrate any other OpenStack project's tempest tests.
00:38:35 <bryan_att> the tests run once the openstack and all components are installed
00:38:42 <thinrichs> bryan_att: yep
00:39:18 <thinrichs> ramineni: do you know if we have docs that detail how to run the tempest tests?
00:39:29 <bryan_att> ok, just to let you know that's an objective. I will look for how to do it. OPNFV People running the base system tests should already know.
00:40:10 <ramineni> we have readme on running tempest tests , im not sure if that what is you are expecting
00:40:28 <bryan_att> I will look at the readme. A link is appreciated.
00:40:32 <thinrichs> Try this...
00:40:34 <thinrichs> #link https://github.com/openstack/congress/tree/master/congress_tempest_tests/tests
00:40:34 <ramineni> https://github.com/openstack/congress/blob/master/congress_tempest_tests/tests/README.rst
00:40:41 <bryan_att> thanks
00:40:51 <bryan_att> Next, some other questions / suggestions are being accumulated. How/when to bring these to the Congress team is an open question.
00:41:24 <bryan_att> I know you are all very busy. But want to get started on these to get more engaged in the project.
00:41:27 <thinrichs> We always like feedback!
00:41:42 <thinrichs> I'd recommend an email, probably to the mailing list, if you're comfortable with that.
00:41:42 <bryan_att> I'll give you a quick overview now, and we can followup.
00:41:49 <thinrichs> Great!
00:41:51 <bryan_att> #1 is that Horizon is not showing the Policy tab.
00:42:03 <bryan_att> No idea why.
00:42:18 <thinrichs> jwy: you here?
00:42:33 <thinrichs> jwy is our expert on Horizon
00:42:49 <bryan_att> ok, a followup or irc sidebar would be great
00:43:12 <bryan_att> as background: The use cases I am focusing on are shown in the OPNFV Copper document at http://artifacts.opnfv.org/copper/docs/design/index.html
00:43:17 <thinrichs> seems she's not here
00:43:26 <bryan_att> (these may seem off-base and hypothetical)
00:43:37 <bryan_att> But I'm struggling to find a lot of use cases I can achieve through the datasources currently supported by Congress.
00:43:45 <bryan_att> I may suggest additional datasource table columns, if I can find the type of data/relations supporting the use cases, in the OpenStack component data structures.
00:43:54 <ramineni> bryan_att: have u checked , ui files of congress are copied to horizon dashboard directory
00:44:22 <bryan_att> does that assume that congress is running on the same server as horizon?
00:45:02 <bryan_att> I have congress running in a LXC container on the same host as horizon, but horizon is also running in a container.
00:45:29 <bryan_att> the deployment for JOID is shown at https://wiki.opnfv.org/copper/academy
00:45:37 <bryan_att> The first diagram there.
00:46:05 <bryan_att> That's one idea I had, i.e. that the install instructions assume the same host.
00:46:11 <thinrichs> Hmmmm… not sure about the container issues.  But I'd imagine that if you copy the files into Horizon's container it should work.
00:46:18 <masahito> bryan_att: No. Horizon related files are keeped in congress repo. contrib/horizon/*. we need to copy it under horizon repo if you want to use.
00:47:05 <bryan_att> OK, I just need to know what to copy from the congress install to horizon and where to copy it to.
00:47:07 <ramineni> bryan_att: i think horizon should be able to automatically identify the plugins installed in env, but yes copying files should work
00:47:22 <ramineni> bryan_att: i did that for testing my bug :)
00:47:35 <thinrichs> ramineni: do plugins work across containers?  I'd be surprised.
00:47:42 <bryan_att> OK, I wasn't sure how Horizon was discovering the plugins.
00:47:51 <thinrichs> We're running a bit short on time.
00:47:56 <ramineni> im also not sure about containers , how would it wrk
00:48:06 <bryan_att> OK, I can send the rest to the list.
00:48:09 <thinrichs> bryan_att: do you have a long list?
00:48:33 <bryan_att> Getting there... I can dump it here
00:48:51 <thinrichs> We have time to discuss more, but I want to leave some time for open discussion.
00:49:01 <bryan_att> ok
00:49:08 <bryan_att> Some items include: How to create tables for use in policies; How to do string operations, e.g. simple compares with wildcards; How to add metadata/tags to various openstack databases for use by congress policies.
00:49:10 <thinrichs> A dump sounds good, so we have a sense of the breadth of issues.
00:49:24 <ramineni> bryan_att: i can share exact location where to copy seeing my dev , please ping me on IRC later
00:49:41 <bryan_att> OK, will do
00:49:43 <bryan_att> In testing, I seemed to get the congress database off into the weeds after a while:
00:49:49 <bryan_att> - some tables get created that I am not expecting, or that I can't delete (or don't get deleted) through the API e.g. when the related policy is created
00:49:55 <bryan_att> - "rule already exists" error when it doesn't or at least is not retrieved for any existing policy I created
00:50:01 <bryan_att> etc
00:50:31 <bryan_att> That's about all I have written down
00:50:46 <bryan_att> a few more I am scratching my head over though
00:50:55 <thinrichs> bryan_att: that's great feedback!
00:51:21 <bryan_att> ok, that's about all for me\
00:51:28 <thinrichs> bryan_att: if you could put that into an email, we can create a buglist and get to work fixing them.
00:51:36 <bryan_att> will do
00:51:53 <thinrichs> bryan_att: don't forget to include the version of congress you're using: kilo/liberty/master
00:51:57 <thinrichs> bryan_att: thanks!
00:52:05 <thinrichs> #topic open discussion
00:52:12 <bryan_att> I'm using stable/liberty
00:52:31 <thinrichs> bryan_att: that's the right one to be using.
00:52:57 <thinrichs> Open topics?  We haven't done status updates for a while, so that's a good topic too.
00:54:25 <masahito> My update is here.
00:54:59 <masahito> I pushed push-type-datasource-driver https://review.openstack.org/#/c/256303/
00:55:43 <thinrichs> masahito: that's some cool stuff!
00:55:51 <thinrichs> I took a look, let's have others take a look too.
00:56:18 <masahito> It's a first implementation. so feel free to comment it to improve the feature.
00:56:30 <thinrichs> That's a completely new way to get data into the system, so I'd imagine we'll want several people to think about it.
00:57:05 <thinrichs> It's a great first cut!
00:58:04 <masahito> Yes. And I noticed monasca team seems to want the driver for monasca. I'll revise it as much as possible by mid-cycle.
00:59:41 <thinrichs> All right.  We're out of time.
00:59:44 <thinrichs> Thanks all!
00:59:53 <thinrichs> We'll meet next week, but probably not the week after that.
00:59:56 <thinrichs> #endmeeting