16:00:11 <morgan_orange> #startmeeting Functes/ODL test scope extension November 20th 16:00:11 <collabot`> Meeting started Fri Nov 20 16:00:11 2015 UTC. The chair is morgan_orange. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:11 <collabot`> Useful Commands: #action #agreed #help #info #idea #link #topic. 16:00:11 <collabot`> The meeting name has been set to 'functes_odl_test_scope_extension_november_20th' 16:00:12 <jose_lausuch> https://jira.opnfv.org/browse/ONOSFW-141 16:00:18 <morgan_orange> #info Morgan Richomme 16:00:34 <jose_lausuch> it has to be Qinglong to remove it from that functest srpint 16:00:40 <jose_lausuch> #info Jose Lausuch 16:00:57 <morgan_orange> jose_lausuch: I told him on the chan but did not get answers 16:01:27 <jose_lausuch> yes 16:01:30 <jose_lausuch> I will send email 16:01:46 <jose_lausuch> is that mei mei? 16:01:56 <morgan_orange> no Qinglong 16:01:58 <jose_lausuch> Qinglong 16:01:59 <jose_lausuch> yes 16:02:00 <jose_lausuch> ok 16:02:49 <mbeierl> You can also comment right on the issue itself and request them to remove it. You can see that the Sprint field is filled in with Functest Sprint 3. But an email is probably faster 16:03:27 <morgan_orange> jose_lausuch: if we are only 2, the meeting will be short.. :) 16:03:39 <jose_lausuch> yes, email sent :) 16:03:43 <jose_lausuch> thanks mbeierl 16:04:32 <mbeierl> jose_lausuch: anytime :) Just happened to be lurking in the channel and noticed the convo ... 16:04:48 <jose_lausuch> well, maybe it's because we didnt send invitation and people forgot :) 16:05:00 <morgan_orange> yes 16:05:38 <morgan_orange> i copied Peter's mail in the etherpad 16:05:43 <morgan_orange> https://etherpad.opnfv.org/p/functiontest 16:06:01 <morgan_orange> #info the goal is to discuss on the extensions of ODL test scope in Functest 16:06:05 <jose_lausuch> Mark is our JIRA expert from now on 16:06:21 <jose_lausuch> ok 16:06:23 <morgan_orange> #info for the moment we have basic tests (create/update/delete port/network/subnet) 16:06:38 <morgan_orange> #info ODL team has created lots of integration tests 16:06:52 <morgan_orange> #info the idea is to include the most significant one into Functest 16:07:06 <morgan_orange> #link https://wiki.opnfv.org/odl_new_test_suites 16:07:21 <morgan_orange> #info pbandzi started looking at the topic 16:07:22 <jose_lausuch> ah, you created a new wiki :) 16:07:35 <morgan_orange> #link https://etherpad.opnfv.org/p/functiontest 16:07:44 <morgan_orange> euh no, I just copied Ana's page some weeks ago 16:07:55 <morgan_orange> let's review Peter's mail 16:08:16 <jose_lausuch> ok 16:08:25 <amaged__> morgan_orange: sorry if this sounds obvious, is there a criteria on what should be included in functest vs upstream, when it comes to odl tests ? 16:08:45 <morgan_orange> not really, we did not discuss it 16:09:01 <morgan_orange> my view is that we should focus on tets that make sense from an E2E point of view 16:09:12 <morgan_orange> our goal is not to tests all the internal compoent s of ODL 16:09:22 <morgan_orange> but ODL as part of the whole solution 16:09:28 <jose_lausuch> thats right 16:09:44 <amaged__> morgan_orange: that the core of my challenge, as u know, so thanks 16:09:47 <jose_lausuch> but if we got any contributor from ODL in functest, they could even try whatever they want 16:09:51 <morgan_orange> #info amaged__ 16:10:15 <morgan_orange> #info no real criteria (so far) to move a test from ODL to Functest in OPNFV 16:10:15 <jose_lausuch> so that they also take advantage of testing their stuff in an opnfv-openstack deployment 16:10:41 <morgan_orange> #info but no need to import internal tests, we shall considered ODL as part of the wholution and priviledge End to End tests 16:11:11 <morgan_orange> we may need anyway to prepare an env including mininet 16:11:15 <morgan_orange> robot is already there 16:11:28 <morgan_orange> #info mininet may be required (not installed currently when running ODL tests) 16:11:37 <morgan_orange> #info Robot already used to run the basic tests 16:12:01 <jose_lausuch> ok 16:12:16 <morgan_orange> how do you understand his sentence 16:12:20 <jose_lausuch> question was also: on jumphost or on computes ? 16:12:26 <morgan_orange> In OPNFV it is task of deployer to install and configure ODL, together with Openstack. And we already have docker container configured for runnning robot-tests. So that means we cannot just take jjb from opendaylight and use them in OPNFV. 16:12:58 <jose_lausuch> the last sentence is obvious to me 16:13:11 <jose_lausuch> I understand we have have to automate ourselves the relevant test for us 16:13:27 <jose_lausuch> automate=integrate 16:13:30 <amaged__> Since we can do a lot with ODL, do we want to put priority to a group of functions ? or just leave it open to the submitters to add what they want 16:13:30 <morgan_orange> ok 16:14:48 <morgan_orange> good question, i amagined that regarding the timing, it will be more our decision...but you are right it would be interesting to let it open, with a mechanism for submitters who would like to integrate their results into OPNFV 16:14:55 <morgan_orange> for B release i think it is too late 16:15:10 <morgan_orange> I see more a selection of additional testcases made by us (peter, you,..) 16:15:36 <morgan_orange> but for the future it could be interesting to find a way to simplify the import 16:15:51 <jose_lausuch> yes, I agree 16:15:58 <morgan_orange> #info question amaged__: do we want to put priority to a group of functions ? or just leave it open to the submitters to add what they want 16:15:59 <jose_lausuch> its getting late to do a lot of things now 16:16:38 <morgan_orange> #info too short to define procedure to easily import tests from ODL in an open way, but we should keep it mind for next steps 16:17:14 <morgan_orange> jose_lausuch: you question jumphost versus compute was related to mininet, right? 16:17:22 <jose_lausuch> yes morgan_orange 16:17:30 <amaged__> one more question, i hope its not disturbing the conversation, now that we covered tests related to upstream, Do we want to avoid tests that could be relevant to other projects (inside OPNFV) or they are welcomed too, for example : service function chaining project (SFC) tests would be welcomed or left to SFC project ? 16:17:37 <jose_lausuch> but I wouldnt involve jumphost at all 16:17:51 <jose_lausuch> everything we need either in docker file or computes 16:17:58 <jose_lausuch> I dont know what makes more sense 16:18:12 <jose_lausuch> but not in the jumphost 16:18:16 <jose_lausuch> to avoid again dependencies 16:18:33 <morgan_orange> #info question mininet on jumphost or compute 16:18:44 <jose_lausuch> amaged__: SFC is already covered by Yardstick and SFC team 16:18:48 <morgan_orange> #info everything we need either in docker file or compute 16:18:54 <morgan_orange> #info not in jumphost 16:18:58 <morgan_orange> agree 16:19:13 <morgan_orange> #info let's vote next tuesday :) 16:19:15 <jose_lausuch> but other test could be welcomed 16:19:28 <morgan_orange> #question test coverage Do we want to avoid tests that could be relevant to other projects (inside OPNFV) or they are welcomed too, for example : service function chaining project (SFC) tests would be welcomed or left to SFC project ? 16:19:48 <morgan_orange> #info we created companion project 16:19:52 <morgan_orange> #info SFC / yardstick 16:20:00 <morgan_orange> #info GBP / Functest 16:20:22 <morgan_orange> #info it is true that we should develop something to track it otherwise risk to be lost or to overlap 16:20:41 <morgan_orange> for the moment the test "companion" projects are limited (Functest or yardstick 16:20:59 <morgan_orange> not always easy to get rationale why it is in functest why in yardstick 16:21:09 <morgan_orange> as we have a synchro meeting we do not have overlap 16:21:15 <morgan_orange> but to caricature 16:21:24 <morgan_orange> if you perform a ping and check it works +> functest 16:21:31 <jose_lausuch> agree 16:21:38 <jose_lausuch> its not well defined, but thats another topic 16:21:40 <morgan_orange> if you perform a ping, measure time, generate traffic, ..=> you are in yardstick 16:21:58 <morgan_orange> we could probably find a better way to formalize in C release 16:22:06 <morgan_orange> for the moment we will not cover everything anyway 16:22:06 <jose_lausuch> yep 16:22:45 <morgan_orange> so for B release we planned to integrate GBP if possible + additional functional tests to be filtered from https://jenkins.opendaylight.org/releng/view/CSIT-Jobs/ 16:23:02 <morgan_orange> amaged__: could you help us to define a list of relevant testcase 16:23:32 <morgan_orange> regarding the risk of overlap it is limited, yardstick is dealing with SFC and BGP VPN 16:23:53 <morgan_orange> not sure there other ODL related tests 16:24:05 <amaged__> morgan_orange: yes, i would like to do that 16:24:17 <morgan_orange> we could also onboard some BGP VPN tests in functest if necessary but we could discuss with Ana 16:24:33 <jose_lausuch> ok 16:24:36 <jose_lausuch> thanks 16:24:39 <jose_lausuch> guys, I need to leave 16:24:42 <morgan_orange> #action amaged__ help peter to find significant test cases for B release to be added into Functest 16:24:50 <morgan_orange> ok jose_lausuch have a good week end 16:24:57 <jose_lausuch> have a nice weekend too 16:25:01 <jose_lausuch> :) 16:25:02 <jose_lausuch> bye 16:25:14 <amaged__> bye 16:25:26 <mbeierl> ciao! 16:25:30 <morgan_orange> so amaged__ mean work will consists in studying how to integrate mininet in our test env (could be also useful for yardstick..) 16:25:38 <morgan_orange> and select testcases 16:25:46 <amaged__> morgan_orange: I will, any where in specific to track/add these relevant test cases ? 16:25:56 <morgan_orange> that are the question 2 , 3 and 4 of Peter 16:26:24 <morgan_orange> #info https://wiki.opnfv.org/odl_new_test_suites wiki page used to track /add testcases 16:26:29 <morgan_orange> #info or directly in Jira backlog 16:26:59 <amaged__> ok, got it, thx 16:27:10 <morgan_orange> #we do not have constraints regarding APi freeze, mainly integration work. However some dev may be neede to setup the env properly (introduction on mininet) 16:27:35 <morgan_orange> you have to synchonize with Peter...but you know him well :) 16:28:09 <amaged__> i'll do that and let you know of the outcome of our discussion 16:28:12 <morgan_orange> we discussed a little bit question 3, we will vote on Tuesday - where we should put mininet 16:28:39 <morgan_orange> as he mentioned on 4, some refactoring will be needed on his existing script 16:28:47 <morgan_orange> OK so if it is clear 16:28:56 <morgan_orange> je pense qu'on peut passer en français 16:28:57 <morgan_orange> et conclure 16:29:34 <amaged__> n'est pas a problem 16:29:46 <morgan_orange> j'ai bien compris 16:29:58 <morgan_orange> du coup réunion de sycnrho mardi prochain 8h UTC 16:30:26 <morgan_orange> #info next meeting Functest weekly, we will keep some time for ODL discussions 16:30:29 <morgan_orange> #endmeeting