08:00:23 <jose_lausuch> #startmeeting Functest weekly meeting April 18th 2017 08:00:23 <collabot> Meeting started Tue Apr 18 08:00:23 2017 UTC. The chair is jose_lausuch. Information about MeetBot at http://wiki.debian.org/MeetBot. 08:00:23 <collabot> Useful Commands: #action #agreed #help #info #idea #link #topic. 08:00:23 <collabot> The meeting name has been set to 'functest_weekly_meeting_april_18th_2017' 08:00:26 <jose_lausuch> hello everyone 08:01:50 <LindaWang> hi jose_lausuch good morning! 08:03:03 <jose_lausuch> hi LindaWang 08:03:07 <jose_lausuch> let's see who is in today 08:04:21 <jose_lausuch> HelenYao: SerenaFeng: ? 08:04:32 <SerenaFeng> hi 08:04:49 <SerenaFeng> Good morning 08:04:57 <jose_lausuch> good afternoon for you :) 08:06:52 <jose_lausuch> #topic role call 08:06:54 <HelenYao> hi 08:06:56 <jose_lausuch> #info Jose Lausuch 08:07:02 <SerenaFeng> #info SerenaFeng 08:07:08 <HelenYao> #info Helen Yao 08:07:22 <LindaWang> #info Linda Wang 08:07:23 <ollivier> #info Cédric 08:07:49 <jose_lausuch> #topic Action point follow-up 08:07:55 <jose_lausuch> #info AP: jose_lausuch propose demo for the hackfest about the available tooling in releng 08:08:23 <jose_lausuch> #info the page is not ediable any more, I will ask Ray to add it to Friday 08:08:28 <jose_lausuch> #info AP: jose_lausuch review https://gerrit.opnfv.org/gerrit/#/c/32083/ and associate patches (32083 impacting all the feature projects..need to be merged to avoid too many rebasing) DONE 08:08:42 <jose_lausuch> #info AP: all complete Functest postmortem (what was ++, +, - or --) 08:08:53 <jose_lausuch> we will have a topic about it 08:08:59 <jose_lausuch> #info AP: ollivier add example of good commit message 08:09:15 <jose_lausuch> https://wiki.opnfv.org/display/functest/Review+checkpoints 08:09:26 <jose_lausuch> #info DONE 08:09:41 <jose_lausuch> #info AP: jose_lausuch propose 2 new gating jobs: pylint and unix permissions. Follow up with Aric 08:09:47 <jose_lausuch> #topic Review Danube postmortem 08:09:50 <jose_lausuch> #link https://etherpad.opnfv.org/p/FunctestDanubePostMortem 08:09:59 <jose_lausuch> let's go for the ++ 08:10:20 <jose_lausuch> python abstraction, unit tests and dockerziation of test API 08:10:25 <jose_lausuch> I think they are great achievements 08:11:14 <jose_lausuch> I would like to add: Easy to integrate 3rd party test cases 08:11:29 <jose_lausuch> what do you think? 08:11:31 <SerenaFeng> agree 08:12:08 <SerenaFeng> that is a great work for Danube 08:12:47 <jose_lausuch> yes 08:13:01 <HelenYao> applause for us 08:13:01 <jose_lausuch> anything else ? 08:13:03 <ollivier> half achieved for D. E 08:13:25 <jose_lausuch> great work for Danube and improved in Euphrates 08:13:43 <jose_lausuch> anything else for the ++ ? 08:13:53 <ollivier> or started in D. 08:14:21 <HelenYao> support for openstack keystone v3? 08:14:28 <jose_lausuch> good point 08:14:35 <jose_lausuch> feel free to put it in the etherpad 08:15:36 <OPNFV-Gerrit-Bot> Cedric Ollivier proposed functest: Remove former cmd instance attribute https://gerrit.opnfv.org/gerrit/33445 08:15:37 <OPNFV-Gerrit-Bot> Cedric Ollivier proposed functest: Add docstings in feature.py https://gerrit.opnfv.org/gerrit/33447 08:15:54 <jose_lausuch> HelenYao: ? 08:15:59 <jose_lausuch> HelenYao: shall I put it? 08:16:11 <HelenYao> jose_lausuch: pls help to add it:) 08:16:21 <jose_lausuch> anyone can add things 08:16:24 <jose_lausuch> https://etherpad.opnfv.org/p/FunctestDanubePostMortem 08:17:20 <jose_lausuch> HelenYao: done 08:17:37 <jose_lausuch> don't you have access to that etherpad? 08:18:14 <HelenYao> jose_lausuch: i can see the page 08:19:18 <jose_lausuch> yes, it's easy to add or modify things :) 08:19:20 <jose_lausuch> ok 08:19:23 <jose_lausuch> for the + 08:19:37 <jose_lausuch> adoption of reporting test api/db by storperf and vsperft 08:19:40 <jose_lausuch> good point 08:19:48 <jose_lausuch> refactoring (name consistency) 08:19:59 <jose_lausuch> that is also related to the abstraction clases 08:20:04 <jose_lausuch> the refactor we have done in functest in general 08:20:09 <jose_lausuch> quick integration of refstack 08:20:20 <jose_lausuch> work with upstream (OpenStack, ODL) 08:20:33 <jose_lausuch> anything else you want to add? 08:22:27 <SerenaFeng> nope 08:22:31 <jose_lausuch> ok 08:22:34 <HelenYao> no 08:22:36 <jose_lausuch> now, the interesting things 08:22:39 <jose_lausuch> -- 08:22:43 <jose_lausuch> - promotion mechanism not ready 08:22:51 <jose_lausuch> that's something we still need to work on 08:23:20 <jose_lausuch> which is this: https://wiki.opnfv.org/display/functest/CI+evolution+for+Functest+in+Danube 08:23:44 <jose_lausuch> basically, for scenarios that normally pass daily/smoke, we should run weekly and more extensive test cases 08:24:18 <jose_lausuch> - logger and env/const : lots of change, still not fully clean 08:24:26 <jose_lausuch> we should have a discussion on this 08:24:30 <jose_lausuch> what do we do with constants 08:24:34 <jose_lausuch> releng constants or functest constants 08:24:40 <ollivier> ?? 08:25:17 <ollivier> we have already defined too many constants... Please defined in releng only the return values they are expected 08:25:39 <jose_lausuch> sorry, I was referring to the return values in releng 08:25:43 <jose_lausuch> we have our own constants, right 08:26:35 <ollivier> so 0 and 1 as Jenkins manages only 2 exit values 08:28:08 <jose_lausuch> yes, we wanted to have more values but then we need to install some plugins in jenkins to make it "orange" 08:28:15 <jose_lausuch> I'm not sure if we still want to do that... 08:28:56 <ollivier> Sure. We are mixing run errors and test results (we could write it in --) 08:29:38 <ollivier> (and we don't care about push_to_db) 08:29:51 <jose_lausuch> ollivier: can you add that to the etherpad? 08:30:41 <ollivier> again? If you want. 08:31:24 <jose_lausuch> I mean here https://etherpad.opnfv.org/p/FunctestDanubePostMortem 08:31:26 <jose_lausuch> in the -- 08:31:38 <SerenaFeng> we don't care about push_to_db ? 08:31:43 <SerenaFeng> what does it mean? 08:31:43 <ollivier> run_tests.opy 08:31:53 <SerenaFeng> I think push_to_db is very important 08:33:40 <SerenaFeng> and I am sorry if I misunderstand your point 08:33:55 <jose_lausuch> me too, what do you mean by "we don't care about push_to_db " ? 08:34:11 <LindaWang> Currently, if one testcase passed but push_to_db failed, it is not obvious to detect. 08:34:33 <ollivier> see run_tests.py 08:34:50 <ollivier> I'm searching the review 08:35:29 <ollivier> l 157 08:36:24 <MatthewLi> #info Jun Li 08:36:26 <jose_lausuch> ok 08:36:54 <jose_lausuch> - unit tests shall not be managed by an interns but reflex new feature/change => new unit test sto be adopted more widely 08:36:57 <jose_lausuch> who wrote this? 08:37:57 <ollivier> I share this point of view. I should'nt be written by only one person after features have been developped. 08:38:46 <ollivier> We have written it in our guidelines too (add unit tests when bugfixes...). 08:39:16 <jose_lausuch> yes, but we started from 0 unit test 08:39:19 <jose_lausuch> the intern did a good job 08:39:31 <jose_lausuch> now, for all the changes we do in the framework we have to add also the coverage 08:39:32 <jose_lausuch> taht's clear 08:41:27 <ollivier> I agree about the job done. Thank you for that. But unit tests should not be done at the end and should not be delegated. 08:41:39 <ollivier> to only one person as it's a common task. 08:41:45 <jose_lausuch> ollivier: agree 08:41:50 <jose_lausuch> for the -- 08:41:57 <jose_lausuch> we have no master jobs on apex 08:42:02 <jose_lausuch> and no weekly jobs for VNFs 08:42:11 <jose_lausuch> which are 2 major issues we had in Danube 08:42:29 <ollivier> ++ for Ashish 08:43:43 <jose_lausuch> any other comments? 08:43:45 <HelenYao> will we have verification job to run selected tests to make sure the code will not break the build whenever a patch is merged? 08:43:50 <jose_lausuch> we can move to current issues 08:44:04 <jose_lausuch> HelenYao: hopefully 08:44:04 <HelenYao> also, will we ask the installer to use stable build instead of latest build? 08:44:45 <ollivier> Please see https://gerrit.opnfv.org/gerrit/#/c/24801/ about push_to_db status 08:44:56 <jose_lausuch> HelenYao: what do you mean? 08:44:57 <HelenYao> I would propose to run integration test on stable builds(stable installer/functest/other testing project) 08:45:22 <HelenYao> this will cause less chaos and help expose the problem in an early phase 08:46:18 <jose_lausuch> yes, the installers should run healthcheck for their patches 08:46:23 <jose_lausuch> only apex has done that 08:46:26 <jose_lausuch> it should be for all 08:46:57 <ollivier> #link https://gerrit.opnfv.org/gerrit/#/c/24801/ 08:47:19 <jose_lausuch> #topic current issues 08:47:39 <jose_lausuch> any issues you want to raise now? 08:47:43 <jose_lausuch> gerrit reviews? 08:47:49 <jose_lausuch> things that are pending? 08:48:09 <HelenYao> https://gerrit.opnfv.org/gerrit/#/c/33299/ 08:48:18 <HelenYao> i need jjb expert to help with this patch 08:48:38 <HelenYao> i did not update the kvm jjb but it always complains kvm issue 08:48:43 <jose_lausuch> HelenYao: ah yes, we need to do the log thing 08:48:58 <jose_lausuch> HelenYao: I'll try to help and ask others if needed 08:49:17 <HelenYao> jose_lausuch: that's great. thx 08:49:22 <jose_lausuch> so 08:49:39 <jose_lausuch> the issue is when a jenkins job times out, we don't run the log collector 08:49:46 <jose_lausuch> so we can't see the logs pushed to artifcats 08:49:47 <jose_lausuch> right? 08:49:53 <HelenYao> yes 08:50:11 <jose_lausuch> ok 08:50:28 <HelenYao> other testing projects are having the same issue 08:50:43 <HelenYao> if the log collector is ready, it would be very helpful for debugging 08:51:28 <jose_lausuch> yes 08:51:29 <jose_lausuch> ok 08:51:34 <jose_lausuch> let's make that work this week 08:51:40 <HelenYao> neat 08:51:50 <jose_lausuch> #action HelenYao jose_lausuch log collector when jjob times out https://gerrit.opnfv.org/gerrit/#/c/33299/ 08:52:20 <HelenYao> do we have time to discuss the e plan? 08:52:50 <HelenYao> I would propose to make Functest callable through rest api 08:53:06 <jose_lausuch> HelenYao: you can expose your idea 08:53:19 <jose_lausuch> HelenYao: and we will take it into account next week 08:53:45 <HelenYao> great. you guys have great time next week:) 08:53:45 <SerenaFeng> r u sure next week? 08:54:07 <jose_lausuch> next week is the plugfest 08:54:11 <SerenaFeng> I think we would be in PlugFest, have no time to attend weekly :) 08:54:44 <jose_lausuch> SerenaFeng: yes, I meant in the plugfest :) 08:54:55 <SerenaFeng> ok :) 08:55:47 <SerenaFeng> And about the TestAPI authentication, do you think it is necessary to implement a prototype using a simple version? 08:55:52 <jose_lausuch> HelenYao: can you write your proposal somewhere in the wiki? 08:55:56 <jose_lausuch> in the Euphrates page 08:56:01 <jose_lausuch> in the comments for example 08:56:04 <jose_lausuch> or a subsection 08:56:28 <HelenYao> I left a comment in the wiki days ago 08:56:34 <jose_lausuch> SerenaFeng: yes, we could implement it. But it's not clear yet if we will use refstack authentication or ours 08:56:42 <jose_lausuch> but if we implement it I think it's good to have 08:57:02 <SerenaFeng> yes, 08:57:25 <MatthewLi> yep, wait for this now actually :) 08:57:33 <SerenaFeng> so I think it make sense to wait for the decision 08:57:50 <jose_lausuch> SerenaFeng: yes. but anyway, not bad idea 08:57:56 <jose_lausuch> but no rush 08:58:03 <SerenaFeng> if we implement a simple version and it will be discard later, there's no point 08:58:24 <SerenaFeng> ok, I suggest to discuss it during PlugFest 08:58:49 <jose_lausuch> yes, after the plugfest we should have a decission 08:58:54 <jose_lausuch> any other pending issues? 08:58:58 <SerenaFeng> MatthewLi I think you can use the current version, although no authentication 08:59:25 <MatthewLi> I think should implement it, then we can test it during plugfest even if it's simple version 08:59:57 <MatthewLi> now, have nothing, can't do anything, even though a discussion basis from my side 09:00:32 <SerenaFeng> to-be-discarded version better not to waste time doing it, my two cents :) 09:01:06 <MatthewLi> that's also my 2 cents :) 09:01:41 <jose_lausuch> SerenaFeng: yes, let's wait for a decission 09:01:42 <jose_lausuch> what I mean 09:01:57 <jose_lausuch> if the decission is to go for Refstack, it would also be good to have it in our api 09:02:11 <SerenaFeng> agree 09:02:25 <SerenaFeng> it should be implemented, the point in what way :) 09:03:00 <jose_lausuch> yes 09:03:03 <jose_lausuch> ok 09:03:12 <jose_lausuch> let's close the meeting here 09:03:17 <jose_lausuch> and see you need week 09:03:22 <jose_lausuch> HelenYao: pity you can't make it 09:03:24 <jose_lausuch> :) 09:03:27 <jose_lausuch> bye 09:03:29 <jose_lausuch> #endmeeting