08:01:03 #startmeeting Functest weekly meeting July 12th 08:01:03 Meeting started Tue Jul 12 08:01:03 2016 UTC. The chair is morgan_orange. Information about MeetBot at http://wiki.debian.org/MeetBot. 08:01:03 Useful Commands: #action #agreed #help #info #idea #link #topic. 08:01:03 The meeting name has been set to 'functest_weekly_meeting_july_12th' 08:01:08 #topic call role 08:01:54 #info Morgan Richomme 08:01:54 #info ollivier 08:02:04 #chair jose_lausuch ollivier 08:02:04 Current chairs: jose_lausuch morgan_orange ollivier 08:02:42 #info agenda: https://wiki.opnfv.org/display/functest/Functest+Meeting 08:02:49 #info SerenaFeng 08:02:58 #info Jose Lausuch 08:03:01 #topic action point follow up 08:03:07 nice to use this chan :) 08:03:11 #link http://ircbot.wl.linuxfoundation.org/meetings/opnfv-testperf/2016/opnfv-testperf.2016-07-05-08.00.html 08:03:36 #info AP1 => request sent to helpdesk to get our gerrit robot on this chan, no feedback so far 08:03:55 #info but jenkins/flake8 now operationnal for Functest 08:04:32 #info AP2: document opnfv-functest documented from the functest main page, I should also still change it in the different documents 08:04:57 #info AP3: ollivier pythonization of ODL almost done 08:05:22 #no other action points 08:05:24 shall we fix the .sh problem or we wait for pythonization? 08:06:00 I think we should fix .sh anyway. 08:06:07 ya 08:06:15 we should avoid CI blocking, so I would +1 for the fix and/or disable ODL testing in CI 08:06:21 trozet reported the bug to jira 08:06:32 yep we will discuss all that in next sections 08:06:44 Are you speaking about the admin rights of *.sh? 08:06:49 #topic Colorado release Status 08:06:50 yes 08:06:58 or whatever the issue is 08:07:20 #info feedback provided to all the installers (except apex / mail ready but not sent yesterday) 08:07:36 /home/opnfv/repos/functest//testcases/Controllers/ODL/start_tests.sh: Permission denied 08:07:39 in // tim created JIRA related to the differnt issue on CI pipeline detected with apex 08:07:56 #link https://wiki.opnfv.org/display/apex/Apex+Scenario+Status 08:08:03 so let's start with apex 08:08:28 #info apex scenario: issues with ODL, security_scan and DVR support (odl_l3 scenario) 08:08:29 jose_lausuch: I am checking it in my container. 08:09:24 ok 08:09:35 in mine, it has X permissions 08:09:40 jose_lausuch: It seems fine here (I don't remember modifying it by hand) 08:10:34 ollivier: we have access to the LF POD1 (Apex) so if needed we need to find a windows slot but we shall be able to test it directly on the target POD 08:11:41 my connection to LFPOD1 is not stable, I get kicked out in 5 sec 08:11:53 we need also to see with luke, security scan are now working but issues with data collection 08:12:06 jose_lausuch: ok.... 08:12:26 regarding the other installers 08:12:51 #info fuel feedback most of the scenarios almost OK http://testresults.opnfv.org/reporting/functest/release/master/index-status-fuel.html 08:13:26 #info fuel last issues: bgpvpn (integration in progress) 08:13:42 we may note that the onos-sfc is already deployed 08:13:54 not sure onos-sfc suite is already triggered 08:13:57 is the test working? 08:13:58 mmm 08:14:07 morgan_orange: Ok we have to check it quickly :) 08:14:27 morgan_orange: where do you see that scenario triggered? 08:15:14 http://testresults.opnfv.org/test/api/v1/results?case=tempest_smoke_serial&last=10&scenario=os-onos-sfc-ha 08:15:28 tested on lf-pod2 10/7 08:15:30 ok here https://build.opnfv.org/ci/view/functest/job/functest-fuel-baremetal-daily-master/154/console 08:15:53 not triggered, because vping fails 08:16:07 #action morgan_orange asks Anthony if onos-sfc tests can be enable as there is one fuel scenario ready to support it 08:16:52 https://build.opnfv.org/ci/view/functest/job/functest-fuel-baremetal-daily-master/148/console 08:17:19 seems that the tests were already included in onos suite 08:17:25 2016-07-10 11:33:10,878 - sfc_fun_log - INFO - The packet has reached VM2 Instance 08:17:26 2016-07-10 11:33:10,879 - sfc - INFO - SFC function Working 08:17:38 I should discuss how results are collected 08:18:04 it seems that the sfc suite remplaced the old one 08:18:27 http://testresults.opnfv.org/test/api/v1/results?case=onos&last=10&scenario=os-onos-sfc-ha 08:19:27 #info feedback on compass, except ocl scenario (still under integration), other scenarios have issues with rally sanity - but last runs seems OK so wait and see .. 08:19:59 something to fix 08:20:34 #info feedback on joid: Narinder indicated a DHCP issue under investigation that should be a root cause of several issues 08:20:50 #info Luke Hinds 08:21:16 question: we have our automatic reporting, Tim created a wiki page to reference the JIRA associated with the scenario, I think we should reference the Jira for the other installer here 08:21:46 in the wiki page I did not see the Jira created by Viktor and Juha (results on tempest and Rally not reaching the success criteria) 08:22:13 #action morgan_orange create wiki pages to reference installer related JIRA 08:22:29 you mean this? https://wiki.opnfv.org/display/apex/Apex+Scenario+Status ? 08:22:34 yes 08:23:16 shall we create a page or ask each installer to do it, assuming that on our side we have our automatic reporting... 08:23:47 mmmm 08:23:59 our automatic reporting tells how functest behaves 08:24:09 but we dont have to check every isntaller' deployment status 08:24:17 I think its something for each installer to do 08:24:20 but its a nice wiki 08:24:26 maybe david mcbride could propose that 08:24:31 but I dont think its for us to do 08:24:39 ok agree 08:24:49 just reference the links in our Colorado page 08:25:11 other topic: I also noticed a bug (I need to create a Jira) on rally_sanity => status is always PASS even the criteria is not correct 08:25:43 ok 08:25:48 can you assign it to juhak? 08:26:01 SerenaFeng: as Juha is in vacations may I assign this future Jira to you => need to change values in the DB and fix it in the code (I assume it is because the success rate is given as a string so I do not knwo what the test is doingĂ  08:26:01 or is he on vacation? 08:26:17 ok 08:26:47 but I will not do it until next week, it that ok? 08:26:56 #Action morgan cretae a jira assigned to SerenaFeng and details issues on rally_sanity result 08:27:24 I will attend OpenStackChina Summit from tommorow to the end of this week 08:27:24 SerenaFeng: yep no problem, we will keep on being over optimistic until next week, then with the change on the DB we will come back to reality 08:27:45 ok 08:27:45 no rush it is just reporting and we have everything in DB 08:28:08 other subtopic for Colorado, valentin got lots of question on the famous vIMS test case 08:28:25 boucherv_orange: a little status on the different requests so far 08:29:25 boucherv_orange: shall we create Jira to track the requests? 08:30:19 #topic security_scan 08:30:37 #info discussion planned during the TSC meeting today to validate the creation of a dedicated repo 08:31:35 2 things to discuss: 1) JIRA (problem with data collection => as scans are now OK) 2) success criteria 08:31:36 I did give this some time to think over, do you still want to try to define a binary pass / fail grade? 08:31:39 regarding 1) 08:32:24 I have no strong opinion, as you said it should be more seen as recommendation than test for scenario validation 08:32:48 so a binary PASS/FAIL would make sense later when we got more maturity and the system in place for all the installers 08:33:03 we can put it in place but not consider it for scenario validation 08:33:25 it is however interesting to report it so installer may make some evolution to get better results 08:33:34 What I was thinking is that we could pull out the overall score into functest's dashboard? 08:34:02 so my suggestion would be a reporting like Tempest (duration, nb of failed, details on failed cases) but just for information => not used for validation of the scenario 08:34:04 e.g. 48% success, 50% fail, 2% N/A 08:34:23 and then link to the reports (the full html with eye candy) 08:35:13 yes we could push the results to the DB as a Tempest result => http://testresults.opnfv.org/test/api/v1/results?case=tempest_smoke_serial&last=2&installer=apex 08:35:22 with also the link to the artifact report 08:35:33 and based on that re use a jinjer2 framework like vIMS or Tempest 08:35:46 http://testresults.opnfv.org/reporting/functest/release/master/index-tempest-apex.html 08:36:07 if you are OK with that luke, you can add the push to DB, I can do all the reporting/jinjer2 stuff 08:36:22 and add a security tab in the reporting page 08:36:35 http://testresults.opnfv.org/reporting/ in addition of tempest vIMS and Status 08:37:31 I don't mind that, looks like there are already a few testcases doing a DB push that I could lift the related functions from, and use in my code 08:37:49 it is also not a top priority (as not part of scenario validation)..we must focus on CI integration 08:37:59 ok I will create some minor JIRA then 08:38:12 ok, well that would suit me well, as I will be hands full with extending to different installers 08:38:47 sounds good though 08:38:55 regarding the JIRA, I hope I will find some time to also test (hope my connection will be more stable...) 08:39:50 #topic AoB 08:40:01 last of all on sec_scan 08:40:09 SerenaFeng: do you think you could have time to meet people from rally at OpenStack Summit China? 08:40:15 going to start looking into the report download failure today 08:40:24 #undo 08:40:24 Removing item from minutes: 08:40:37 will chat with Tim later on (although he might be on leave) 08:40:43 I try to catch some stuff for the minutes 08:40:49 If need, I will try to contact them 08:40:54 #info security_scan now producing nice reports 08:41:07 so you can hashtag an action against me that I am on it 08:41:12 #info still an issue with integration in CI pipeline when collecting the restuks 08:41:33 #info secruity_scan not used for scenario validation (must be generalized to all the installers first) 08:42:03 #info however as teh info are usefull => adopt tempest approach to collect success rate, errors and create a Security reporting (like Tempest of vIms) 08:42:29 #info minor priority / generalization to othe installers and fix in CI 08:42:39 #topic AoB 08:43:17 Some of my collegues works in rally will be there, maybe I can contact them 08:43:30 #info SerenaFeng participating to OpenStack Summit China => could be interesting to discuss with Rally community as we are using rally for Tempest/Rally and could use it more for additional scenario 08:44:18 ollivier: started working really more upstream with ODL (we are wating for their merge on the last patch) 08:44:34 about what information you want me to disscuss with them? 08:44:50 I think we should investigate to do the same with rally (I know that Juha and Viktor already reported bugs but it could be interesting to discuss also the feature evolutions) 08:45:50 the idea is to get a better understanding how it works, where can we discuss feature evolution, what they could expect from us as end to end tester in OPNFV ecosystem 08:46:16 what kind of contribution we could reasonably make 08:46:18 ... 08:46:24 another topic for AoB 08:46:39 we start getting questions from feature projects aiming to connect the compute/control node 08:47:01 okey, I will have a try to disscuss with them 08:47:01 basically for feature projects tehre are several ways to be onboarded of teh SUT 08:47:12 the plugin/charm/addon/ansible playbook in the installer 08:47:22 postinstallation scripts 08:47:53 some projects seem to priviledge the second approach but today we do not have an abstraction layer for such connectivity and have to do it manually installer per installer 08:48:10 let's propose it for D-release 08:48:18 it shouldnt be that difficult 08:48:21 for D release, we probably have to think to that, see if there is already something for apex in SNAPS framework 08:48:30 4 if-else statements :) 08:48:33 it should not be very difficult 08:48:48 the question is is it good to encourage this approach versus a more installer integrated approach 08:49:08 it is more a question for genesis 08:49:16 the ideal case would be to have an installer API 08:49:20 common for all installers 08:49:24 but yes, thats genesis 08:49:37 but yet another "requirement" 08:49:47 which might stay in JIRA forever 08:50:23 OK maybe we can also discuss that with the test community because whatever the nature of the tests, such API/library would be useful 08:50:51 yes 08:50:56 #info requests from feature project to be able to connect to the different SUT nodes 08:51:12 #info so far no standard way => namual processing different from one installer to another 08:51:34 #info discussions on an installer integrated approach versus a post-installation through test umbrella projects 08:52:03 #info could be a genesis topic 08:52:05 #action morgan_orange jose_lausuch discuss this topic with testing community 08:52:11 AoB: enjoy this video https://www.youtube.com/watch?v=Zzb0xLfnn9Y 08:52:14 ok 08:52:50 #info summer organization: I will be on vacations on Wednesday until the 8th of August 08:53:15 #info Jose will take the lead until beginning of August 08:53:27 we need someone to lead for the first week of August 08:53:31 starting tomorrow? 08:53:39 yes.... 08:53:42 ok 08:54:08 who volunteers for august? 08:54:19 envy, no summer vacation in China 08:54:28 SerenaFeng: would you be OK to represent Functest? for the first week of August 08:55:09 the weekly meeting? 08:55:34 yes weekly meeting 08:55:39 release meeting (3Pm UTC) are not convenient for China but you may report to David Mc Bride by mail (idem for the weekly test meeting -2 PM UTC) 08:56:38 okey 08:56:44 #info SerenaFeng will lead Functest on the first week of August 08:56:55 any other info you want to share this week? 08:57:08 not from my side 08:57:23 I have implement auto update of testapi using ansible-playbook 08:57:34 and make the git review 08:57:48 #info auto update of testapi using ansible-playbook impemented by SerenaFeng 08:58:02 is it enable on testresults.opnfv.org? 08:58:16 until now, I think test-api evoluation is completely finish for now 08:58:32 yes 08:59:00 the PUT pethod is tehre, I did not use it yet - we probably need to code some function in functestUtils getTrustIndicator, modifyTrustIndicator 08:59:10 you can run the script in local vm and update testapi of testresults.opnfv.ofg 08:59:10 but great work on the test API 08:59:46 do we need to modify the date in the database? 08:59:53 hi morgan , amwaitign for the doc ? 09:00:36 SerenaFeng: which date? 09:00:38 The update_mongodb.py already support the modification, only need to add the changes to changes_in_mongodb.py 09:00:47 trust_indicator 09:00:56 to support the new data model 09:01:23 yes you want to do or I do, as you want 09:01:38 from trust_indicator: 0 to trust_indicator:{current: 0, history: []} 09:02:13 I add a jira to implement it 09:02:22 raghavendrachari: the patch on devguide has been submitted the doc is here http://artifacts.opnfv.org/functest/docs/devguide/index.html 09:02:33 ok 09:03:36 #action SerenaFeng update test api on testresults.opnfv.org with last version of datamodel 09:04:58 ok that is all for this week 09:05:27 have a good week, I will be back on the 8th of August, enjoy the Alicante style 09:05:48 #endmeeting