17:03:43 <davidkranz> #startmeeting qa
17:03:44 <openstack> Meeting started Thu Apr  4 17:03:43 2013 UTC.  The chair is davidkranz. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:03:46 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:03:48 <openstack> The meeting name has been set to 'qa'
17:03:59 <davidkranz> #topic Summit
17:04:28 <davidkranz> I just saw an email to the qa list about a scenarios session.
17:04:39 <ravikumar_hp> unreviewed
17:04:41 <davidkranz> I don't know when that session was submitted.
17:04:51 <donaldngo_hp> hi
17:05:07 <davidkranz> I don't think it was there when we discussed the sessions.
17:05:11 <ravikumar_hp> i think 10days ago
17:05:14 <JoseRaxQE> hey
17:05:29 <jaypipes> davidkranz: yeah, I think it was a bit of a latecomer
17:05:41 <davidkranz> I am still not sure I understand what the real purpose is.
17:06:23 <dwalleck> My understanding was talking about more project integration tests (nova/cinder integration tests, etc)
17:07:09 <davidkranz> dwalleck: If they are currently deficient they should be improved but that is already a clear part of Tempest's mission.
17:07:22 <dwalleck> davidkranz: agreed
17:07:34 <jaypipes> ++
17:07:43 <davidkranz> So perhaps this should be discussed as part of the last session which is to discuss missing tests.
17:08:01 <davidkranz> ravikumar_hp: I think you submitted that one.
17:08:10 <ravikumar_hp> yes. test gaps
17:08:20 <jaypipes> davidkranz: either that, or we continue to discuss it in the mailing list thread and make it clear that just because something may be long-running doesn't mean it doesn't belong in tempest
17:08:40 <ravikumar_hp> to convert to blueprints that will address in Havava .
17:08:57 <davidkranz> ravikumar_hp: I think blueprints for missing tests is the right approach.
17:09:32 <ravikumar_hp> ok.
17:09:41 <davidkranz> jaypipes: I hope we hash out the whole "what tests should be in tempest" issue at the summit.
17:09:46 <ravikumar_hp> davidkranz: no session needed ?
17:10:02 <davidkranz> ravikumar_hp: I don't think so but you can bring this up in your gap session.
17:10:29 <davidkranz> ravikumar_hp: I just don't see it as a new concept that we need a separate session for.
17:10:37 <jaypipes> I agree.
17:10:51 <ravikumar_hp> ok
17:10:59 <davidkranz> ravikumar_hp: Great.
17:11:19 <davidkranz> I think it would be good for session proposers to put up etherpads before the summit.
17:11:37 <dwalleck> makes sense]
17:11:43 <davidkranz> Is there some "official" place to put such etherpads.
17:12:08 <dwalleck> isn't there an openstack etherpad?
17:12:40 <davidkranz> dwalleck: Yes, I meant the links so people can find them.
17:13:09 <davidkranz> dwalleck: to the individual session etherpads. Ideally they would be in the session descriptions online.
17:13:37 <davidkranz> We could ask ttx about that.
17:14:37 <davidkranz> dwalleck: Are you guys going to post the new stuff you talked about before the summit?
17:15:07 <dwalleck> davidkranz: we're working out getting it right now. Sam?
17:15:16 <RAX-Sam> Hey David. It's been posted to StackForge. I've finished the the zuul, build configuration, etc... and it's waiting on Approval
17:15:23 <RAX-Sam> https://review.openstack.org/#/c/25870/
17:15:46 <RAX-Sam> When that goes through, the repositories will be hosted on StackForge
17:16:21 <davidkranz> RAX-Sam: OK, it would be really helpful if we could take a look before the summit so the discussion can be better informed.
17:16:36 <RAX-Sam> Absolutely. Has been our goal. :-)
17:17:02 <davidkranz> RAX-Sam: If there is some kind of holdup in the official process perhaps you could point us at an informal link to the code.
17:17:36 <davidkranz> #topic General Discussion
17:17:39 <RAX-Sam> There has been some activity and review today. If it hasn't been merged by today I'll email some links around tomorrow
17:17:47 <davidkranz> RAX-Sam: Thanks!
17:18:06 <davidkranz> Any one have a topic to bring up?
17:18:22 <donaldngo_hp> https://blueprints.launchpad.net/tempest/+spec/update-expected-exception-tests is implemented can someone with previleges close it?
17:19:23 <afazekas> related to the scenario testing the policy.json mentioned by ykaul
17:19:49 <davidkranz> donaldngo_hp: Just did that.
17:20:12 <donaldngo_hp> thank you
17:20:18 <davidkranz> afazekas: Not sure what you are pointing at.
17:20:33 <afazekas> We should extend the default policy.json 's in-order to have a user type which is member of tenant but does not have permission to do anything
17:21:17 <afazekas> by the default policy many operation is doable by everyone
17:21:47 <afazekas> we do not test the code path when the operation rejected by the policy module
17:22:09 <davidkranz> afazekas: Oh, I see.
17:22:11 <afazekas> I think the policy module itself is tested by unit test
17:22:35 <dwalleck> I think the trick is that there's no guarantee in a given environment what the policy.json is
17:22:48 <davidkranz> afazekas: Doesn't trying to do an admin thing as non-admin cover that?
17:23:01 <afazekas> I do not thing so
17:23:26 <afazekas> some cases yes, but in many case it is not
17:23:32 <dwalleck> You could do generated tests from a policy.json, but that would definitely need a lot of work
17:23:41 <afazekas> yes
17:23:43 <davidkranz> afazekas: Which code path are you saying is not covered? It is potentially different for each project, no?
17:24:01 <afazekas> when the policy module rejects on operation
17:24:05 <mtreinish> davidkranz: here now sorry, lunch went long
17:24:15 <davidkranz> dwalleck: Isn't that unit-test territory?
17:24:22 <afazekas> we have test cases for attemting operations from differnt tenent, but it is differnt decision
17:24:47 <dwalleck> davidkranz: Not necessarily
17:25:05 <afazekas> davidkranz: the correctness of the policy module itself is a unit test territory imho
17:25:19 <afazekas> but the usage of the policy module is not
17:25:50 <davidkranz> afazekas: So this heads back to the issue of how complete a functional api test tempest is supposed to be.
17:26:02 <davidkranz> afazekas: There seem to be many opinions about this.
17:26:08 <afazekas> https://blueprints.launchpad.net/tempest/+spec/add-glance-policy-tests
17:26:44 <afazekas> davidkranz: yes to policy module allow us to define very crazy polices
17:27:31 <davidkranz> afazekas: Where would you like to see this discussion end up?
17:28:16 <afazekas> I would like to add to have two type of "regular" users
17:28:18 <dwalleck> I think the difference is "Does policy.json work?" vs. "Do I have the expected policy.json deployed and does it enforce the permissions I expect?"
17:28:47 <afazekas> the "nobody"  type, who does not have any permission, but can authenticate
17:29:19 <afazekas> and the "somebody" , who has the same capability what the default users have now
17:29:30 <afazekas> and the admin users ..
17:30:48 <davidkranz> afazekas: The blueprint you pointed to does this with admin/non-admin users.
17:31:09 <davidkranz> afazekas: I must be missing something because I don't understand the difference in code path with nobody/regular user.
17:31:11 <afazekas> davidkranz: yes, but compare it with the default devstack policy.json
17:32:04 <afazekas> http://www.fpaste.org/R9d3/
17:32:33 <afazekas> I do would like to use different policy than in the blueprint
17:32:45 <afazekas> "default": "role:sombody",
17:33:12 <afazekas> the "demo" user should have this right
17:34:55 <afazekas> the policy.json which is in the blueprint is copied from an openstack documantion
17:35:08 <afazekas> http://docs.openstack.org/developer/glance/policies.html
17:35:11 <davidkranz> afazekas: I see that.
17:35:23 <davidkranz> dwalleck: Do you have a comment about this?
17:36:59 <davidkranz> afazekas: How about you propose something specific.
17:37:17 <davidkranz> afazekas: I don't think there is any objection, just not clear about what needs to be changed.
17:37:23 <dwalleck> davidkranz: I see both sides. I'm thinking. If we have this specific config, then we're really just testing that specific configuration
17:37:46 <davidkranz> dwalleck: Yes, and for that particular project.
17:38:26 <davidkranz> But we could have devstack use a policy file that has more meat for testing.
17:38:42 <davidkranz> I think that is what afazekas wants.
17:38:46 <afazekas> I started a longer e-mail about the scenario stress and policy testing .., but I did  not finished it and probably I will rephrase it
17:39:03 <afazekas> davidkranz: yes
17:39:25 <davidkranz> afazekas: OK, sounds good. We can continue in email.
17:39:34 <afazekas> ok
17:40:00 <davidkranz> Anything else to bring up?
17:41:33 <davidkranz> OK, I guess we are done for today then.
17:41:36 <sdague> not from me, just looking forward for portland
17:42:07 <davidkranz> sdague: Me too.
17:42:51 <davidkranz> I hope some of us can take a look at the new RAX stuff.
17:43:06 <davidkranz> See you all next week.
17:43:12 <davidkranz> #endmeeting