17:01:44 <thinrichs> #startmeeting CongressTeamMeeting 17:01:45 <openstack> Meeting started Tue Jun 9 17:01:44 2015 UTC and is due to finish in 60 minutes. The chair is thinrichs. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:01:46 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 17:01:49 <openstack> The meeting name has been set to 'congressteammeeting' 17:02:04 <thinrichs> masahito_, pballand: hi 17:02:49 <thinrichs> Let's start with status updates, as usual. 17:02:53 <thinrichs> #topic status 17:03:03 <thinrichs> pballand: want to get us started? 17:03:32 <pballand> sure 17:03:56 <pballand> not much to report from last week - just made a mechanical change for oslo-policy 17:04:25 <pballand> this week I’m looking at adding an API validator to close a but on on errors with invalid API input 17:04:31 <pballand> that’s it for me 17:05:04 <thinrichs> pballand: the oslo_policy change is that we need to reference oslo_policy instead of oslo.policy, correct? 17:05:34 <pballand> it removes the copy of the oslo policy code from our repo, and uses the (newly created) shared package 17:05:55 <pballand> minor namespace and code changes were also introduced since we last copied the code from oslo 17:06:37 <thinrichs> So then in our requirements.txt we have oslo.policy, but throughout the code we use oslo_policy. Is that right? That seems weird. 17:07:13 <thinrichs> The underscore versus the period, that is. 17:07:48 <pballand> one’s a PIP package name, the other is a python namespace 17:08:17 <thinrichs> Got it. 17:08:27 <pballand> you can talk with the oslo team about their naming conventions, but I don’t know there is much we can do about it 17:09:36 <thinrichs> The API validator sounds cool. How does the validator know what data to validate? In code, or does it pull in types as API meta-data? 17:10:21 <pballand> I’m planning on adding JSON Schemas for each resource, and validating against that 17:10:42 <thinrichs> Nice! 17:11:23 <pballand> there was a blueprint for that at one point, but I know there is a bug so I was going to do the work in reference to the bug 17:11:36 <pballand> #link https://bugs.launchpad.net/congress/+bug/1367888 17:11:38 <openstack> Launchpad bug 1367888 in congress "Malformed curl command causes internal error" [Medium,Triaged] - Assigned to Peter Balland (pballand) 17:12:05 <thinrichs> Here's the blueprint. 17:12:06 <thinrichs> https://blueprints.launchpad.net/congress/+spec/api-validation 17:12:22 <thinrichs> I think the bug was a higher-level concern. 17:12:47 <thinrichs> The bug was that the API framework didn't catch an error during json-loading. 17:13:05 <thinrichs> So that would be before the API-validation would even have a chance to run. 17:13:52 <thinrichs> You could always reference both the bug and the blueprint in the commit message. 17:14:02 <pballand> ok, I believe the fix for that specific concern is already in 17:14:09 <pballand> just need some test code 17:14:32 <thinrichs> Sounds good. 17:15:06 <thinrichs> Anything else? 17:15:54 <pballand> actually, I just checked, json parsing errors are not caught, so it falls through to the default (500) handler 17:16:00 <pballand> will take care of that 17:16:03 <pballand> nothing else from me 17:16:06 <thinrichs> Cool. 17:16:37 <thinrichs> masahito_: want to tell us what you've been working on? 17:16:57 <masahito_> yes 17:17:49 <masahito_> I've been working on adding test fors api/*.py 17:18:01 <masahito_> https://bugs.launchpad.net/congress/+bug/1435616 17:18:02 <openstack> Launchpad bug 1435616 in congress "missing test coverage for datasource api code" [Medium,New] - Assigned to Masahito Muroi (muroi-masahito) 17:19:11 <masahito_> I couldn't have enough time, so I continue to make a patch for it. 17:19:12 <thinrichs> That'd be great to get those tests moved out of test_congress and into the api/ folder. 17:19:47 <thinrichs> It was great what you did for the table API. 17:19:53 <masahito_> I'll upload this patch in this week. 17:20:11 <masahito_> thank you. 17:20:36 <thinrichs> masahito_: Is there anything you need from the rest of us? 17:21:03 <masahito_> yes. 17:22:30 <thinrichs> masahito_: what do you need from us? 17:23:22 <masahito_> After reading codes, I found out that there is a file which is not used as api in this folder. 17:25:21 <masahito_> I wonder if I need to add a test for it or not. 17:25:27 <thinrichs> Which file 17:25:29 <thinrichs> ? 17:25:53 <masahito_> It is congress/api/datasource_config_model.py 17:27:06 <masahito_> harness.py creates a service from the file but router.py doesn't set it in a handler. 17:28:45 <thinrichs> Hmmm… 17:29:04 <thinrichs> This is weird. I think I remember asking for the config for a datasource, which I assume is handled by the datasource_config_model. 17:29:48 <alexsyip> I remmber that also. Do you remmeber what the URL was ? 17:30:52 <thinrichs> I think I was using the client. 17:31:07 <thinrichs> Looking... 17:31:18 <alexsyip> There is a tempest test that uses it. 17:31:35 <alexsyip> The tempest tests are not running right now, but they did before. 17:33:43 <thinrichs> I can't seem to track this down. I did notice that there's the Driver config that we can check from the client. 17:33:48 <thinrichs> So maybe that's what I am remembering. 17:34:15 <alexsyip> nevermind, it looks like the HA tempest test does not check the config. 17:34:46 <masahito_> sorry, my connection was lost. 17:35:13 <thinrichs> masahito_: nothing happened other than alexsyip and I thinking we had used the APi call that you said wasn't hooked up. 17:35:34 <thinrichs> masahito_: but then I found a different config API call that was hooked up and thought that I may have been remembering using that one. 17:35:58 <thinrichs> masahito_: alexsyip said he may have been thinking of something else too. 17:36:09 <thinrichs> masahito_: maybe you could check if there's a way to query the datasource (not the driver) config, and if not hook up this datasource_config_model to the API properly. 17:36:45 <masahito_> ok 17:37:58 <thinrichs> While we're on the topic of APIs, I wanted to add a note to the spec about the API changes we were considering: /services/ versus /data-sources/ and /policy-engines/. 17:38:10 <thinrichs> https://review.openstack.org/#/c/183959/1/specs/liberty/policy-engine-api.rst 17:38:30 <thinrichs> At the summit I remember us deciding on the /services/ option, but I don't remember why. 17:38:39 <thinrichs> Does anyone remember? 17:39:00 <thinrichs> It'd be good to add a note to the spec saying what we decided and WHY. 17:39:43 <thinrichs> pballand: you were leading that discussion. Do you remember? 17:40:43 <pballand> I’l need to dig up my notes 17:40:56 <pballand> but I recall a big reason was to fit with delegation 17:41:54 <pballand> it also fits more cleanly with the idea that policy-derived tables and plugin-derived tables are used the same way 17:43:03 <thinrichs> Agreed. 17:43:15 <thinrichs> I don't think that was the nail in the coffin though. 17:43:28 <thinrichs> Anyway, if anyone remembers, let's try to add a note to the spec. 17:43:51 <thinrichs> Oops—running a little short on time here. Let's move on to alexsyip. 17:44:01 <thinrichs> alexsyip: want to give us a status update? 17:44:51 <alexsyip> There was a patch to hook up our tempest tests that merged yesterday. 17:45:07 <alexsyip> I was experimenting with that change. It does not seem to be working correctly. 17:45:36 <thinrichs> I noticed it too. The error I saw was not being able to clone nova 17:45:37 <thinrichs> http://logs.openstack.org/13/186113/8/experimental/gate-congress-dsvm-api/33bb0c2/ 17:45:52 <alexsyip> I had been working on a test for duplicate rules in the DB, but the tempest tests have been failing locally for me. 17:46:16 <thinrichs> All the tempest tests or just a couple? 17:46:24 <alexsyip> just the test_ha tests. 17:46:34 <alexsyip> there was some kind of keystone authentication problem. 17:46:49 <alexsyip> But only for write API calls. The datasource tempest tests still pass. 17:47:20 <alexsyip> I suspect it’s a problem with my setup, which is why I wanted the jenkins test to see if they also fail. 17:49:21 <thinrichs> Ok. Let us know what you find out. 17:49:48 <thinrichs> I'll spend a few cycles trying to figure out why jenkins isn't running the tests properly. 17:50:18 <thinrichs> I'll do a quick status update too…. 17:50:35 <thinrichs> I'm trying to get the sequence of patches for VM-placement delegation merged. 17:50:54 <thinrichs> The global requirements file changed the package I was using for LP solvers, which caused the test failures. 17:51:07 <thinrichs> I'll get that squared away and uploaded soon. 17:51:16 <thinrichs> alexsyip and masahito_: thanks for the reviews. 17:51:46 <thinrichs> I also completed the task of creating a Tablename object to encapsulate a bunch of table-name manipulation code. Mostly mechanical. 17:52:39 <thinrichs> I was also following a couple of threads about Keystone policy to understand what the keystone team is planning and what Nova is actually requesting in terms of policy. 17:52:45 <thinrichs> Here are a couple of links to the archive. 17:52:46 <thinrichs> http://lists.openstack.org/pipermail/openstack-dev/2015-June/065598.html 17:52:46 <thinrichs> http://lists.openstack.org/pipermail/openstack-dev/2015-June/065598.html 17:53:34 <thinrichs> I've been looking into whether Keystone would be another good experiment for delegation. 17:54:20 <thinrichs> Since 5 minutes left, I'll open it up for discussion. 17:54:24 <thinrichs> #topic Discussion 17:54:30 <masahito_> I have 17:54:48 <masahito_> Will we have mid cycle meet up? 17:55:01 <thinrichs> masahito_: that's a great question. 17:55:18 <masahito_> I need to reserve my schedule and budget if we have. 17:55:24 <thinrichs> masahito_: are you interested? I think we'd need to decide on the goals of the meet-up. 17:56:03 <masahito_> yes 17:56:33 <thinrichs> Were you thinking a code-sprint (i.e. everyone gets together to hammer out some specific feature or two)? 17:56:40 <thinrichs> Or were you thinking more of a design-session? 17:57:21 <masahito_> I thought meet up is a code-sprint. 17:58:04 <thinrichs> In the past, we did more of a policy conference. 17:58:16 <thinrichs> But a code-sprint sounds good too. 17:58:30 <thinrichs> Did you have a feature or features in mind? 17:59:01 <thinrichs> Time check: 2 min 17:59:12 <thinrichs> 1 min 17:59:30 <masahito_> I want congress get scaling out feature 18:00:00 <thinrichs> I could definitely see us spending a day or two hammering on performance/scale-out. 18:00:14 <thinrichs> Let's finish this discussion on the ML, to guage interest. 18:00:21 <thinrichs> We're out of time for this week. 18:00:24 <thinrichs> Thanks everyone! 18:00:30 <masahito_> Should we go to congress room to discuss it? 18:00:38 <thinrichs> masahito_: sounds good. 18:00:46 <thinrichs> #endmeeting