07:36:07 #startmeeting qtip 07:36:07 Meeting started Mon Dec 12 07:36:07 2016 UTC. The chair is yujunz-zte. Information about MeetBot at http://wiki.debian.org/MeetBot. 07:36:07 Useful Commands: #action #agreed #help #info #idea #link #topic. 07:36:07 The meeting name has been set to 'qtip' 07:36:17 #topic roll call 07:36:23 #info Yujun Zhang 07:36:59 #info akhilbatra 07:37:27 Agenda today 07:37:30 #link https://etherpad.opnfv.org/p/qtip-meetings-2016-12-12 07:38:06 Not many topics today 07:38:59 #topic action followup 07:39:15 #info AP1 all review CLI intern project plan on https://goo.gl/cWf9cR 07:39:29 Any comments? zhihui akhilbatra 07:41:51 If you have any, please feel free to add in google docs or discuss in mailing list 07:42:08 #info AP2 yujunz create a poll for meeting slot with framadate.org or doodle.com 07:42:43 It's done. And the meeting will be rescheduled to UTC0730 every Monday 07:42:59 #link http://doodle.com/poll/z7ey2u8cg4qa629x 07:43:28 #topic architecture evolution updates 07:43:37 I have made a draft and open for comments 07:43:47 #link https://wiki.opnfv.org/display/qtip/Architecture 07:45:26 Any immediate questions? 07:46:27 what mean Multi Nodes mode? 07:47:07 Multi nodes is for future. A benchmark which runs across multiple nodes 07:47:37 For example, performance of a geography distributed virtual data center 07:48:11 And you want to evaluate the interconnection speed between different physical DC 07:48:36 Something like that, but not limited to it. 07:49:27 Single node will work for most cases and is our target in Danube. 07:50:03 Multi Node is something to be considered when you make modules. Keep in mind of this possible extension. 07:50:10 I see. 07:50:51 #topic unit test files and class naming 07:51:02 #link https://jira.opnfv.org/browse/QTIP-182 07:51:31 #link https://gerrit.opnfv.org/gerrit/#/c/25759/ 07:51:48 akhilbatra is working on it 07:52:36 Serena has concerns on selective running. What about you zhihui ? 07:53:00 I agree with serena. 07:53:26 The benefit 07:53:41 is small to change the name 07:53:56 what is the issue with selective run? 07:54:46 There she comes 07:55:20 akhilbatra has just asked about "what is the issue with selective run?" 07:55:37 #link https://gerrit.opnfv.org/gerrit/#/c/25759/ SerenaFeng 07:56:36 I suppose tox is not only for selective running, but also used for creating virtualenvs automatically and manage multiple version of Pythons 07:57:22 pytest use test* to decide whether it is a test file or test function 07:58:06 if we use other prefix for indexing, we need to config prtest.ini 07:58:42 #link http://doc.pytest.org/en/latest/example/pythoncollection.html#changing-naming-conventions 07:59:03 I think there would be a solution, akhilbatra could you dig further on this issue? 07:59:21 The bottom line is keep all test cases discoverable 07:59:42 Currently from the CI build log, only part of the test cases are collected by tox/pytest 07:59:44 There are three ways we can customize prefix/postfix for test naming 08:01:05 1 - configuring on pytest.ini or tox.ini 08:01:05 2 - configuring in setup.cfg itself 08:01:05 3 - deriving custom named class from unitest.TestCase. 08:01:05 pytest will not ignore any class deriving from above class 08:01:29 We can choose the most suitable 08:02:36 Good. Could you propose one, since you have already look into it, akhilbatra? 08:02:52 I'm not asking for it now, you can amend your patch set after meeting 08:03:31 Hi Taseer 08:03:42 We started the meeting on UTC0730 08:05:11 Continuing on the TestXxx vs CheckXxx, once it works, I think we should rename all instead of just part of them 08:06:38 We can continue discussion on JIRA and gerrit 08:06:46 #topic active sprint followup 08:06:50 Hi 08:07:01 #link https://jira.opnfv.org/secure/RapidBoard.jspa?projectKey=QTIP&rapidView=135 08:07:19 ok i will put all three suggestions on JIRA 08:07:20 ok i will put all three suggestions on JIRA 08:07:46 akhilbatra please propose ONE that you think that is best 08:08:16 And of course, why you think so. 08:08:50 Many tasks are in progress and some are still unassigned. 08:09:10 Feel free to claim the tasks if you are interested in them. 08:09:52 The tasks in active sprint has higher priority than the ones in backlog 08:10:02 serena had a concern that configuring tox.ini will put extra work on developer's side. here by developer, we mean who will use qtip right? 08:10:02 custom unitest are created by us or the developer will also make such unittest? 08:11:03 developer != user 08:11:24 developers are us, who write codes and unit tests 08:11:54 We don't have dedicated test developers, so those who implement a new feature is responsible to write companion unit test 08:12:36 Users of qtip don't run unit tests. They run benchmarks provided by QTIP 08:14:23 ok. so having custom naming convention at one config file would be best. so that we need not derive every test from same base class again. 08:14:23 which config file ( setup.cfg, tox.ini or pytest.ini) should we use, i am not sure. As i dont know other stuffs which are using these config files. 08:14:36 I think once we have a correct configuration, there will be no extra work for developer. So I personally positive for this change 08:15:10 setup.cfg is used by pbr. tox.ini is used by tox, pytest.ini is used by pytest. 08:15:29 But it is possible that tox also reads pytest.ini or pytest also reads tox.ini 08:15:58 pyest read tox.ini for certain 08:15:59 So the best thing would be put it in one file make it work for all. Or make a reference to avoid duplication. 08:16:05 *pytest 08:16:22 Does that work for you SerenaFeng zhihui? 08:16:41 Once it is configured correctly, I think no extra work is required on pytest 08:17:18 if there's no extra work, I am positive 08:17:34 akhilbatra, if your patch will not block the existed unit test, we can turn to check*** 08:18:14 so final patch should acccept both check*** and test***? 08:18:43 I suppose we change all to Check*** 08:18:57 Otherwise you will get warning on TestPlan() 08:19:00 I think only check** is ok, 08:19:16 +1 for only check** 08:19:41 This will avoid future confusing on terminology 08:20:19 OK, let's move on. 08:20:36 #topic CI status 08:20:38 #link https://build.opnfv.org/ci/view/qtip/ 08:20:43 zhihui has cleaned up obsolete jobs, so it looks much clear now 08:21:11 It seems we have three green/blue jobs and just one failure 08:21:28 Do you have any information to share on the failed job, zhihui? 08:21:47 Yes, 08:22:14 The deployment job on zte-pod2 is failed for a long time. 08:22:36 1 month and 2 days exactly 08:22:52 So there is not an openstack for qtip to run test. 08:23:27 For VM benchmarking? 08:23:37 And I am sorry I don't know when the deployment job is back to normal. 08:23:57 because it depends on fuel4nfv daily version. 08:24:27 Not only for vm benchmarking 08:24:53 So, it is an error of the environment, not just QTIP task 08:25:06 the ip of host is fetched by the installer. 08:25:39 not just qtip task 08:26:03 Link https://build.opnfv.org/ci/view/fuel/ 08:26:05 I see, then it is not very urgent. At least we have pod3 08:26:49 #topic new tasks in JIRA 08:26:58 #link https://jira.opnfv.org/issues/?filter=11198 08:28:36 If you are working on some new tasks, please add it to current sprint. 08:28:42 So that the team can be aware of who is working on what 08:28:51 To avoid overlapping 08:30:50 #topic AoB 08:31:02 Any other topics you want to talk about? 08:31:47 Good. 08:31:56 I think that's all for today. Thank you all 08:31:58 #endmeeting