08:00:08 <morgan_orange> #startmeeting Functest weekly meeting February 2nd 08:00:08 <collabot> Meeting started Tue Feb 2 08:00:08 2016 UTC. The chair is morgan_orange. Information about MeetBot at http://wiki.debian.org/MeetBot. 08:00:08 <collabot> Useful Commands: #action #agreed #help #info #idea #link #topic. 08:00:08 <collabot> The meeting name has been set to 'functest_weekly_meeting_february_2nd' 08:00:14 <morgan_orange> #info Morgan Richomme 08:00:18 <viktor_nokia> #info Viktor Tikkanen 08:00:19 <juhak> #info Juha Kosonen 08:00:54 <morgan_orange> #info agenda https://wiki.opnfv.org/functest_meeting 08:01:12 <morgan_orange> #info topic Action point follow up 08:01:21 <morgan_orange> #link http://ircbot.wl.linuxfoundation.org/meetings/opnfv-testperf/2016/opnfv-testperf.2016-01-26-08.00.html 08:01:49 <jose_lausuch> hi 08:01:50 <morgan_orange> #info hardware configuration check done on Huawei, Orange POD2 and LF POD2 08:01:55 <jose_lausuch> #info Jose Lausuch 08:01:59 <morgan_orange> hi jose_lausuch and raghavendrachari 08:02:11 <raghavendrachari> Hi 08:02:17 <raghavendrachari> have a nice day .!! 08:02:49 <jose_lausuch> :) 08:02:49 <morgan_orange> as far as I understand we excluded a problem tempest worker / hardware config of the jumphost 08:02:52 <jose_lausuch> what is the topic sorry? 08:03:08 <morgan_orange> just reviewing the action points of last week 08:03:30 <morgan_orange> we checked the HW config of the jumphost to see if it was the cause of the worker issue detected by viktor_nokia 08:03:49 <jose_lausuch> yes 08:03:53 <jose_lausuch> did we reach any conclussion? 08:04:42 <morgan_orange> I think we got the same config on LF POD1 and LF POD2 and had issue on LF POD1 not on LF POD2 so ...my understanding that this is not the cause of the error 08:04:49 <morgan_orange> but I may be wrong... 08:05:05 <jose_lausuch> and huawey pod had similar HW 08:05:17 <jose_lausuch> a lot of CPUs in all the cases... 08:05:21 <jose_lausuch> on the jumphost I mean 08:05:53 <jose_lausuch> Huawei sorry 08:05:55 <morgan_orange> #info juhak refactoring of volume management on rally-cert done 08:06:08 <morgan_orange> #info morgan_orange adaptation of oDL suite for JOID done 08:06:17 <jose_lausuch> juhak: rally shows always problems with live migration 08:06:29 <jose_lausuch> shall we say live migration of VMs is not supported ? 08:06:45 <morgan_orange> #info robot scenario path changed in ODL repo, but scenario did not change 08:06:57 <morgan_orange> #topic Release B 08:06:58 <juhak> live migration requires shared storage 08:07:14 <jose_lausuch> but we have ceph in theory... 08:07:28 <morgan_orange> #info refactoring of traces of Tempest and Rally in Jenkins done => nice results 08:07:33 <jose_lausuch> or we don't use it for ephemeral... 08:07:49 <morgan_orange> #info question on rally issue on live migration => shall these scenario be excluded? 08:07:59 <juhak> should we exclude live migration cases - for the time being? 08:08:42 <jose_lausuch> I don't know how the tests are performed, but I imagine it uses ephemeral (local disk) which I think it is not ceph, it's just local , but I'm not 100% 08:08:43 <jose_lausuch> sure 08:08:53 <jose_lausuch> but we might exclude them 08:08:54 <morgan_orange> if we know why it is failing...it makes sense... 08:09:09 <jose_lausuch> and just show the tests that are in tempest (if any) to prove that it fails 08:09:19 <jose_lausuch> and explain why 08:09:43 <jose_lausuch> May-meimei: can you check on Compass pods how is that implemented? 08:09:49 <morgan_orange> #info possible now to have a synthetic view of the different results 08:09:57 <morgan_orange> #link http://testresults.opnfv.org/proto/index-pod.html 08:09:57 <jose_lausuch> ceph for cinder? ceph for ephemeral? 08:10:07 <morgan_orange> #link http://testresults.opnfv.org/proto/index-scenario.html 08:10:18 <morgan_orange> #link http://testresults.opnfv.org/proto/index2.html 08:10:23 <morgan_orange> on Orange POD it is Ceph 08:10:38 <jose_lausuch> joid? 08:10:55 <morgan_orange> so tests should run but since yesterday we got regression, no Tempest tests are run (error saying that the config file cannot be found) 08:11:02 <morgan_orange> jose_lausuch: yes joid /orange POD2 08:11:23 <jose_lausuch> morgan_orange: ok, thanks, I see that live migration doesn't fail on orange pod2, but cinder stuff failed.. 08:12:13 <morgan_orange> so if it is OK on some PODs, I suggest to keep the test and explains in the user guide... 08:12:17 <jose_lausuch> yes 08:12:20 <jose_lausuch> +1 08:12:26 <juhak> ok 08:12:36 <jose_lausuch> do we know why volume rally tests fail on that pod? 08:12:43 <jose_lausuch> test scenario CinderVolumes.create_and_attach_volume 08:12:48 <jose_lausuch> test scenario CinderVolumes.create_and_delete_volume 08:12:53 <jose_lausuch> test scenario CinderVolumes.list_volumes 08:12:57 <jose_lausuch> test scenario CinderVolumes.create_nested_snapshots_and_attach_volume3 08:13:01 <jose_lausuch> those are success 0% 08:13:04 <jose_lausuch> here https://build.opnfv.org/ci/view/functest/job/functest-joid-orange-pod2-daily-master/55/console 08:13:11 <jose_lausuch> to further investigate 08:13:14 <jose_lausuch> not now :) 08:13:36 <jose_lausuch> action someone with access to orange pod 2 08:13:38 <jose_lausuch> David? :) 08:13:58 <morgan_orange> I think David_Orange juhak viktor_nokia raghavendrachari and myself have access to this POD... 08:14:41 <raghavendrachari> yes am having access 08:14:43 <morgan_orange> #info regarding B release, still investigations mainly on vPing, vIMS 08:15:04 <juhak> yes, i can execute cinder cases manually in pod2 and check what's the problem 08:15:42 <jose_lausuch> juhak: ok, thanks. just things like that, cinder list, cinder create, snapshots, etc 08:15:52 <morgan_orange> #action juhak perform cinder test on Orange pod2 08:15:54 <jose_lausuch> attach 08:16:03 <morgan_orange> #info ODL suite on joid now looks like in apex (15OK /3 KO) I think that it is due to an update of the code, the tests did not follow => so I can change the test or just document it 08:16:09 <morgan_orange> #info ONOS tests OK 08:16:22 <morgan_orange> #info promise tests run 08:16:52 <morgan_orange> #info Tempest was relatively stable alert on regression on Orange POD2 (last 2 run => impossible to find the config) => I will have a look at it 08:17:05 <morgan_orange> #action morgan_orange Tempest on Orange POD2 where is the tempest.conf... 08:17:13 <jose_lausuch> morgan_orange: remember the tests that fail are the lats ones, delete stuff, that is what pbandzi did, not taken from the repo, maybe there is something to change 08:17:36 <morgan_orange> jose_lausuch: yes I will check that too 08:17:37 <jose_lausuch> yes, something fishy happens with tempest.conf 08:17:45 <jose_lausuch> I've seen it several times 08:17:47 <jose_lausuch> ok 08:18:01 <morgan_orange> #action morgan_orange ODL check the delete scenario from upstream repo (instead of tests in OPNFV repo) 08:18:22 <morgan_orange> what is strange is that it was very stable...I can do it manually to see what happens 08:19:02 <morgan_orange> #info Rally suite seems Ok for the release 08:19:25 <morgan_orange> I saw some errors on heat scenario but nothing critical for the release 08:19:42 <jose_lausuch> yep 08:19:47 <jose_lausuch> heat was failing in some cases too 08:19:49 <morgan_orange> for the vPing/SSH, I saw success results in Apex, Compass and Joid 08:19:58 <jose_lausuch> with ODL or ONOS? 08:20:05 <GeraldK_> are you IRC only? GTM tells me waiting for moderator. 08:20:19 <jose_lausuch> GeraldK_: yes, IRC only :) 08:20:38 <morgan_orange> for the vPing/SSH apex and joid with odl_l2 08:20:48 <morgan_orange> for compass also I think 08:20:49 <GeraldK_> okay. just wondered as the invitation had a GTM link 08:20:55 <juhak> <jose_lausuch>: I'll check those heat cases too on pod2 08:21:08 <jose_lausuch> morgan_orange: floating ips didnt work in Apex+ONOS 08:21:21 <morgan_orange> #action juhak checkheat cases on Orange POD2 08:21:40 <jose_lausuch> neither in Apex+ODL-l3 08:21:41 <morgan_orange> jose_lausuch: but it works once with apex now? 08:21:52 <morgan_orange> apex/odl_l2 08:22:11 <morgan_orange> so https://wiki.opnfv.org/vping_brahmaputra_page?&#vping is up to date? 08:22:11 <jose_lausuch> Apex+ODL-L2 08:22:12 <jose_lausuch> yes 08:22:25 <morgan_orange> comapss I saw something successfull with onos 08:22:38 <morgan_orange> very long 266s but it was ok 08:23:32 <jose_lausuch> morgan_orange: compass+ODL-l2 failed too https://build.opnfv.org/ci/view/functest/job/functest-compass-huawei-us-deploy-bare-1-daily-brahmaputra/57/consoleFull 08:23:36 <jose_lausuch> Cannot establish connection to IP '192.168.10.101'. Aborting 08:23:45 <morgan_orange> yes the last success was compass/onos 08:24:09 <morgan_orange> Valentin_Orange: status on vIMS? 08:24:09 <jose_lausuch> yes https://build.opnfv.org/ci/view/functest/job/functest-compass-huawei-us-deploy-bare-1-daily-brahmaputra/56/consoleFull 08:24:37 <morgan_orange> #info vPing some scenario now working apex/odl_l2, compass/onos, joid/odl_l2.... 08:24:40 <jose_lausuch> #action jose_lausuch push to DB if vPing test timeout 08:24:52 <morgan_orange> #info more run needed to update and confirm the status 08:26:10 <morgan_orange> #info as you know theoretically we have slot for troubleshooting 08:26:31 <Valentin_Orange> morgan_orange: ok on apex/odl_l2 08:26:49 <Valentin_Orange> morgan_orange: ok on fuel/odl_l2 08:27:08 <Valentin_Orange> morgan_orange: KO on joid/odl_l2 .. 08:28:17 <morgan_orange> #link https://wiki.opnfv.org/vims_brahmaputra_page 08:28:29 <morgan_orange> #info progress on vIMS test case 08:28:49 <morgan_orange> #info vIMS OK on fuel and apex odl_l2, investigation in progress on joid and compass 08:29:08 <morgan_orange> #info reminder troubleshooting slots for Functest : LF POD1 (Apex) 18-20 UTC / LF POD2 (Fuel) 5-16 UTC, POD5 (Joid) 19-18 UTC, POD8 (Compass??) 13-15UTC, POD6 (joid), 19-20 UTC... 08:29:25 <morgan_orange> jose_lausuch: did you get access to POD5 and POD6? 08:29:40 <morgan_orange> just to know if it is now easier than LF POD access... 08:29:44 <jose_lausuch> morgan_orange: not yet, still waiting, the JIRA issue was changed to another person 08:29:49 <morgan_orange> :) 08:30:13 <morgan_orange> I think we can report that access to the POD is still an issue...and yesterday no POD was abvailable for testing... 08:30:26 <morgan_orange> we are supposed to be in testing phase with stable System Under Test.... 08:30:35 <morgan_orange> no run of Fuel since 2 or 3 days.. 08:30:51 <morgan_orange> The graphs are crual but they do not lie... 08:30:53 <jose_lausuch> still failed deployments sometimes :) 08:31:26 <morgan_orange> #info so to summarize on Functest for B release 08:32:10 <morgan_orange> #info vPing => need more scenario to check, recent progress in apex/odl_l2 and compass/onos but all combinations still not there 08:32:14 <leo_wang> vIMS is so not stable 08:32:22 <leo_wang> all kinds of problems 08:32:53 <morgan_orange> #info vPing_userdata need more scenario to confirm existing results (all old_l2 scenario OK) 08:33:32 <morgan_orange> #info ODL => check the 3 custom tests but OK on the different labs (just a test to check norhbound API of the Controller at the end) 08:33:50 <morgan_orange> #info ONOS => ok on all installers (need more run to confirm stability) 08:34:30 <jose_lausuch> good 08:34:45 <morgan_orange> #info vIMS OK on apex/fuel still issues with compass/joid, complex test cases (more close to realistic needs of Telco) 08:34:56 <morgan_orange> leo_wang: how could we investigate more efficiently 08:35:06 <morgan_orange> would it be possible to grant access to Valentin_Orange? 08:35:15 <morgan_orange> we are lacking direct access to a compass lab 08:35:33 <morgan_orange> Intel POD8 has been assigned but can we use it? is there any Compass solution with a scneario running in this lab? 08:35:42 <leo_wang> the problem varies every time 08:36:11 <morgan_orange> a priori not with apex and fuel 08:36:16 <leo_wang> even failed to download cfy blueprint 08:36:23 <morgan_orange> network issue 08:37:01 <leo_wang> and create a venv may fail too 08:37:01 <morgan_orange> ok continue discuss it with Valentin_Orange offline but it would be great to have an access to the lab to help you 08:37:42 <leo_wang> :( 08:37:51 <leo_wang> ok 08:38:01 <morgan_orange> #info promise OK 08:38:22 <morgan_orange> #info ovno a bit late but theoretically it is ready from a Functest/CI perspective 08:38:44 <morgan_orange> #info sdnvpn idem ovno wait for a run of the scenario to see if the tests are properly triggered 08:39:05 <morgan_orange> I think we should ask for a strategy on scenario on CI 08:39:22 <jose_lausuch> what do you mean? 08:39:29 <morgan_orange> CI slots have been defined..these slots must be used to run the different scenario 08:39:42 <morgan_orange> It is strange to have 1 run / day last week end for each installer... 08:39:49 <jose_lausuch> yes 08:39:54 <jose_lausuch> that is not acceptable 08:39:57 <morgan_orange> during the CI slot we should have at least 2 runs with 2 different scenario 08:40:26 <morgan_orange> #action morgan_orange propose scenario strategy for CI to get more deployment and then more tests 08:40:54 <morgan_orange> #info labs still not really ready for troubleshooting (no CI labs available yesterady) 08:41:37 <morgan_orange> #info doc has been updated, ChrisPriceAB split it to provide a global doc, everbody can still review it and patch it especially for the troubleshooting section 08:41:48 <morgan_orange> we have lots of things to say... :) 08:41:55 <morgan_orange> anythging else on B release? 08:42:24 <ChrisPriceAB> Once the docs are on master I should be able to put them in context. Then we may need some editing. But otherwise on track. 08:42:30 <morgan_orange> #action leo_wang provide access to Compass labs to Valentin_Orange for vIMS troubleshooting on compass 08:42:52 <jose_lausuch> ChrisPriceAB: aren't they on master already? what do you mean? 08:43:24 <ChrisPriceAB> They are, just saying I will have a central doc up soon. Then whateer is on master will be pulled in. (Patches will not) 08:43:42 <jose_lausuch> ChrisPriceAB: ok, thanks ! 08:44:25 <morgan_orange> ok viktor_nokia juhak jose_lausuch Valentin_Orange ChrisPriceAB raghavendrachari leo_wang May-meimei anything else to add for B release... 08:44:55 <morgan_orange> #topic plugfest 08:44:56 <jose_lausuch> 1 small thing 08:45:00 <morgan_orange> #undo 08:45:00 <collabot> Removing item from minutes: <MeetBot.ircmeeting.items.Topic object at 0x1b71950> 08:45:16 <jose_lausuch> it seems everyone liked the new way of cleaning openstack defaults 08:45:25 <morgan_orange> oh yes... 08:45:30 <juhak> well done 08:45:33 <jose_lausuch> I'll merge https://gerrit.opnfv.org/gerrit/#/c/8927/1 08:45:41 <morgan_orange> ok 08:45:46 <viktor_nokia> :) :) 08:46:06 <morgan_orange> if all the installers would have provide a consistent config, it would not have been needed but...we can see that it is useful... 08:46:20 <jose_lausuch> maybe for the next release we can have 2 options 08:46:38 <jose_lausuch> cleanup everything, or just cleanup the things that were there before the tests 08:46:41 <morgan_orange> let's discuss in the last 10 minutes discussions on next release... 08:46:45 <morgan_orange> #topic plugfest 08:46:50 <jose_lausuch> sure 08:47:03 <morgan_orange> #info as you know a plug fest is on track. it will be organized in Colorado in May (Cable Labs) 08:47:09 <morgan_orange> question 1: who will attend? 08:47:24 <morgan_orange> I mean from Functest contributor;.. 08:47:37 * jose_lausuch doesn't... 08:47:50 <morgan_orange> I will try to follow the discussion on dovetail on this topic but not easy to find time 08:48:04 <morgan_orange> I am not sure we are mandatory for such plugfest but our tests will be... 08:48:09 <juhak> Nokia is planning to participate PlugFest. don't know yet if Viktor or me will be there 08:48:52 <morgan_orange> I think Orange will probably also participate but not clear who, all the more as we are supposed to organize a plugfest in Q3 in France 08:49:27 <morgan_orange> so is userguide + container + adapation of the graphs enough as Functest contribution to this plugfest? 08:49:37 <morgan_orange> or do you see stuff we should have to do from no to May 08:49:45 <morgan_orange> specifically for such event? 08:50:22 <morgan_orange> #info Nokia and Orange will probably participate (need confirmation, people not designated at the moment) 08:50:47 <morgan_orange> #info question what can we bring as input for such event 08:50:50 <jose_lausuch> I think that event is more focus on NFV point of view right? 08:50:58 <jose_lausuch> something maybe more important for yardstick guys 08:51:40 <morgan_orange> yep even if running our basic suites gives already a good feedback...and provide some elements of comparison... 08:52:07 <morgan_orange> but OK the idea is not to answer today, I will keep a small point also next weeks to discuss that and I will try to follow a little bit more dovetail 08:52:11 <ChrisPriceAB> I think there will be some YardStick people there and Fuel. 08:52:19 <morgan_orange> #topic Release C 08:52:36 <ChrisPriceAB> But functest has a bunch of troubleshooting skils that would be applicable so don't not go... :D 08:52:59 <morgan_orange> #link https://etherpad.opnfv.org/p/functest_release_C 08:53:23 <jose_lausuch> ChrisPriceAB: :) to what level troubleshooting fits in that event? 08:53:37 <morgan_orange> #info for the moment it is a white page, I see already some possible actions/improvements and I have questions 08:54:15 <ChrisPriceAB> plugging in new components, running our software on new hardware. There is likely trouble in those waters... (back on topic jose_lausuch) 08:54:17 <jose_lausuch> morgan_orange: I will fill some info, some ideas 08:54:39 <jose_lausuch> ChrisPriceAB: ok, I see, thanks 08:54:40 <morgan_orange> for ODL, I do not know if we should keep it inside or externalize it (as ONOS) 08:55:06 <morgan_orange> I think it is time to try to give back to rally and to contribute directly on rally to define Telco KPI 08:55:15 <morgan_orange> question who goes to Austin for the OpenStack summit... 08:55:25 <morgan_orange> I think it would be easier to do that in Barcelona... 08:55:28 <jose_lausuch> morgan_orange: I will add some titles (blocks) for each component, so people can add ideas in an ordered way 08:55:48 <jose_lausuch> Barcelona is closer ;) 08:55:48 <morgan_orange> yes I think we should consider 08:56:01 <morgan_orange> not in time.. 08:56:10 <morgan_orange> - existing testcases (evolution, ...) 08:56:20 <morgan_orange> - nex test cases (sync with Intel and new VNF) 08:56:33 <morgan_orange> - use of orchestrator? 08:56:34 <jose_lausuch> yep 08:56:42 <jose_lausuch> something to be discussed of course 08:56:44 <morgan_orange> - better analytics of our results (we are building a mine) 08:57:04 <morgan_orange> - move to current home API to something like ELK ? 08:57:12 <jose_lausuch> ELK? 08:57:22 <morgan_orange> Elastic Search / Logstash / Kibana 08:57:22 <jose_lausuch> ah ok 08:57:28 <jose_lausuch> yep 08:57:46 <morgan_orange> discuss when we should onboard or externalize companion test cases 08:57:47 <jose_lausuch> shall we everyone take the action to write ideas? 08:58:09 <morgan_orange> #info etherpad open for Functest Release C 08:58:18 <morgan_orange> #info we should consider: 08:58:36 <morgan_orange> #info - existing testcases (evolution, in&out) 08:58:45 <morgan_orange> #info next test cases (sync with Intel and new VNF) 08:58:54 <morgan_orange> #info role of orchestrator for testing 08:59:08 <morgan_orange> #info -upstream contribution( E.g. Rally) 08:59:17 <morgan_orange> #info - analytics? 08:59:35 <morgan_orange> #info moving from home made API + Mongo to ELK? 08:59:52 <morgan_orange> #action evrybody complete this etherpad 09:00:02 <morgan_orange> #link https://etherpad.opnfv.org/p/functest_release_C 09:00:27 <morgan_orange> we will have more time after C release, but we will probably not see in real before Berlin, which will be too late for new feature definition 09:00:52 <morgan_orange> we need to have a clear view of our vizion for March/April => idea is to define the figure as we did for brahmaputra 09:01:15 <morgan_orange> https://wiki.opnfv.org/functextnexttaks 09:01:27 <morgan_orange> The figure has not been changed since last July 09:01:43 <morgan_orange> and it is was not fully stupid... 09:01:53 <morgan_orange> #topic AoB 09:01:57 <morgan_orange> we are 1 minute late 09:02:12 <morgan_orange> anything you want to share before ending the meeting? 09:02:22 <jose_lausuch> nop 09:02:27 <morgan_orange> I want to share one => vote for jose to quartely awards... 09:02:29 <jose_lausuch> everything ok 09:02:54 <jose_lausuch> hahaha its not needed, you are too nice :) 09:02:59 <morgan_orange> we can vote also to viktor_nokia: juhak Valentin_Orange for their contribution 09:03:13 <morgan_orange> and also May-meimei of course 09:03:41 <jose_lausuch> I'm already the Rock'n Tester... ChrisPriceAB^ 09:04:14 <morgan_orange> ok so have a good week, lots of work ahead...meaning that such system needs still some testing... 09:04:27 <morgan_orange> #endmeeting