08:00:55 <ollivier> #startmeeting Functest weekly meeting 18 Apr. 2018
08:00:55 <collabot`> Meeting started Wed Apr 18 08:00:55 2018 UTC.  The chair is ollivier. Information about MeetBot at http://wiki.debian.org/MeetBot.
08:00:55 <collabot`> Useful Commands: #action #agreed #help #info #idea #link #topic.
08:00:55 <collabot`> The meeting name has been set to 'functest_weekly_meeting_18_apr__2018'
08:01:02 <ollivier> #info Cédric Ollivier
08:01:05 <juhak> #info Juha Kosonen
08:01:14 <ollivier> Thank you for the report!
08:01:23 <LindaWang> #info Linda Wang
08:01:24 <ollivier> delay!
08:03:04 <ollivier> Let's wait if Delia joins us. Jose, Viktor and Serena are not connected.
08:04:12 <ollivier> #topic Action point follow-up
08:04:24 <ollivier> #info all Kubernetes testcases are now described in User Guide
08:04:33 <ollivier> #link https://gerrit.opnfv.org/gerrit/55629
08:04:46 <ollivier> #info "Functional testing for OPNFV Kubernetes deployments" by Konrad is now online
08:04:56 <ollivier> #link http://testresults.opnfv.org/functest/ons2018/
08:05:06 <ollivier> #link https://gerrit.opnfv.org/gerrit/#/c/55629/
08:06:07 <ollivier> Linda: I don't think we do plan a second presentation. At least Cristina, Juha and I have seen it.
08:07:01 <ollivier> LindaWang: or maybe during the dedicated weekly meeting?
08:07:49 <ollivier> #info "ONS Summary" was sent last Monday
08:08:17 <LindaWang> ollivier: yes, since most functest members have been there
08:09:01 <ollivier> LindaWang: maybe Valentin will be interested too. I will send a proposal by email.
08:09:21 <LindaWang> ollivier: sure
08:09:39 <ollivier> #topic Functest Fraser
08:09:52 <ollivier> #info Fraser release is postponed (1 week)
08:10:08 <ollivier> #link https://lists.opnfv.org/pipermail/opnfv-tech-discuss/2018-April/020965.html
08:11:08 <ollivier> We can publish our artifacts next Friday if we want or wait the additional week.
08:11:22 <ollivier> I would say it depends on juju_epc (next topic)
08:12:11 <ollivier> #info several issues when verifying APEX via juju_epc are under investigation
08:12:27 <ollivier> juju-wait mustn't hang indefinitely (os-odl...)
08:12:38 <ollivier> juju_epc should stop if worloads are on failure
08:12:59 <ollivier> Could you please review the related change https://gerrit.opnfv.org/gerrit/#/c/55937/ ?
08:13:07 <LindaWang> ollivier: ok
08:13:22 <ollivier> It works well locally
08:13:32 <ollivier> next APEX runs will get more relevant logs
08:14:10 <ollivier> #info a new bug has been created https://jira.opnfv.org/browse/FUNCTEST-961 and the testcase should fail for all installers
08:14:50 <ollivier> It's on Rebaca side.
08:15:47 <ollivier> (abot-epc-basic can't start due to unmet dependencies (Oracle Java 8))
08:16:39 <ollivier> We could release Functest anyway simply because nothing could be done in our tree regarding that issue.
08:17:04 <ollivier> but wait and see :)
08:17:37 <LindaWang> it failed to download Oracle JDK 8
08:17:49 <LindaWang> basically because of network issue
08:17:58 <ollivier> 404
08:18:22 <ollivier> I think the file has been removed.
08:19:16 <ollivier> We could fix it today if it's just about urls.
08:19:34 <LindaWang> oh, what about this issue: https://hastebin.com/vequtupepe.py
08:20:13 <LindaWang> the step of 'juju destroy-controller' failed, but the juju_epc passed
08:20:28 <LindaWang> #link https://build.opnfv.org/ci/view/functest/job/functest-compass-baremetal-daily-master/92/console
08:20:44 <ollivier> Yes. Vnf clean are not considered in results
08:21:53 <LindaWang> ok, the same issue with vrouter
08:21:59 <ollivier> I will check it but it seems the former code and we couldn't get all logs.
08:22:15 <ollivier> no. vrouter is about a wrong cloudify plugin
08:22:36 <ollivier> #info vyos_vrouter doesn't clean all its resources due to an unsupported flag:
08:22:46 <ollivier> #link https://build.opnfv.org/ci/job/functest-apex-baremetal-daily-fraser/14/console
08:22:55 <ollivier> #link https://jira.opnfv.org/browse/FUNCTEST-960
08:23:04 <ollivier> #link https://github.com/oolorg/opnfv-vnf-vyos-blueprint/pull/2
08:23:49 <ollivier> Valentin's proposal was merged this night and it's fixed as expected.
08:24:55 <ollivier> https://build.opnfv.org/ci/view/functest/job/functest-daisy-baremetal-daily-fraser/11/console
08:24:55 <LindaWang> good to know
08:26:02 <ollivier> #info several docs could be updated
08:26:12 <ollivier> #link https://jira.opnfv.org/browse/FUNCTEST-951
08:26:20 <ollivier> #link https://jira.opnfv.org/browse/FUNCTEST-958
08:26:28 <ollivier> #link https://jira.opnfv.org/browse/FUNCTEST-959
08:26:53 <OPNFV-Gerrit-Bot> Merged functest: Check app workload in juju_epc  https://gerrit.opnfv.org/gerrit/55935
08:27:05 <ollivier> Be free to work on 958 and 959. Konrad is working on https://jira.opnfv.org/browse/FUNCTEST-951
08:27:23 <ollivier> Thank you
08:29:33 <ollivier> Simply assign them to yourself if you could help about updating docs.
08:29:39 <ollivier> #info refstack test lists hugly differ between all installers (under investigation)
08:30:29 <ollivier> LindaWang: Have you checked why they are skipped?
08:30:59 <LindaWang> ollivier: Have not figured it out :(
08:31:10 <LindaWang> ollivier: i only have compass deplotyment locally
08:31:38 <ollivier> LindaWang: no pb. This can be finished later.
08:31:51 <ollivier> #topic Functest Gambia
08:32:02 <ollivier> #info all requirements are synced over the OPNFV projects (OpenStack Queens)
08:32:13 <ollivier> #info our testcases have been updated and seem working fine
08:32:40 <ollivier> Only compass is running and the results seem equal.
08:32:52 <ollivier> #link https://build.opnfv.org/ci/view/functest/job/functest-compass-baremetal-daily-master/92/console
08:33:00 <LindaWang> but compass is still based on Pike
08:34:02 <ollivier> Agree. But it's already a good check (OpenStack clients support backward compatibilty).
08:35:55 <ollivier> At least first Gambia milestones are done (healthcheck is passing)
08:35:56 <LindaWang> yes,  it makes sense
08:36:07 <ollivier> #info testr was replaced by stestr (refstack is currenly disabled)
08:36:15 <ollivier> #link https://review.openstack.org/561777
08:37:02 <ollivier> Thank you for this upstream work.
08:37:14 <LindaWang> np, i have tested locally
08:38:39 <ollivier> #info rally-openstack and neutron-tempest-plugin are installed in our containers
08:39:11 <LindaWang> rally -openstack package is from pypi directly?
08:39:16 <ollivier> juhak: I think your work is not part of 1.0.0. Do we have to clone the repo?
08:39:22 <ollivier> yes
08:39:36 <LindaWang> do we need to install rally package from pypi too?
08:40:03 <juhak> ollivier: it's partly in 1.0.0
08:40:14 <juhak> ollivier: work still continues
08:40:38 <LindaWang> ollivier:  i mean  rally repo is still cloned in our Dockerfile  https://github.com/opnfv/functest/blob/master/docker/smoke/Dockerfile#L24
08:40:55 <ollivier> juhak: Maybe it makes sense to clone the repo instead. Then you could activate Gnocchi checks.
08:41:52 <ollivier> LindaWang: Yes. We discussed about that point few meetings ago and Juha adviced that stable branches are good enough.
08:42:04 <juhak> ollivier: would make sense, I could start to integrate Gnocchi cases to functest
08:42:22 <LindaWang> ok
08:42:49 <ollivier> Then we could clone master without git id and we will freeze it later. Or do we update the git commit id all along the dev cycle?
08:43:58 <ollivier> The first model is good right now. But it could lead to issues. I will update the Dockerfile to clone the repo (master).
08:44:26 <juhak> ok
08:44:45 <ollivier> #info parser PTL is aware that we won't maintain a dedicated Parser container for G-release
08:45:31 <ollivier> He has to fix his code. 1 exception per release.
08:45:58 <ollivier> (E wrong OpenStack release, F python namespace conflicts...)
08:46:30 <ollivier> #topic AoB
08:46:34 <ollivier> #info 01/05 and 08/05 are days off (at least) in France
08:47:06 <juhak> 01/05 in Finland
08:47:27 <ollivier> and I think I will be partially off in May (classical french holidays).
08:48:22 <ollivier> Ok 01/05 can be simply cancel. I think we are already in a good shape for G-release we can cancel several meetings and discuss offline if needed :)
08:48:55 <ollivier> I'm suprised that Fraser will be release on 01/05...
08:49:56 <ollivier> #info Could we enable vnfs in virtual deployment? https://git.opnfv.org/releng/tree/jjb/functest/functest-alpine.sh#n175
08:50:40 <LindaWang> ollivier: as far as i know, only one virtual pod of Huawei in Jenkins would support VNFs
08:51:06 <ollivier> due to the resources needed?
08:51:14 <LindaWang> YES
08:52:05 <ollivier> ok. Let's see if we do enable it later for functional gating.
08:52:22 <ollivier> another topic?
08:53:31 <LindaWang> Could we talk about the tests supported for Apex/xci delivery and tratiditional one?
08:54:21 <LindaWang> what should we do? Update rally and tempest version is enough?
08:54:28 <ollivier> LindaWang: yes. Great topic
08:55:09 <ollivier> For the time being it's unclear and David is asking TestPerf instead Functest regarding that point.
08:55:50 <ollivier> I think the issue isnot the testcases themselves but if Functest has to follow master too.
08:56:10 <LindaWang> Functest should support both tracks
08:56:54 <ollivier> I warned about that few month ago (see first XCI mail thread)
08:56:54 <LindaWang> Functest could also follow the master too?
08:57:32 <ollivier> That has to be discussed in Functest before.
08:58:16 <LindaWang> then any result?
08:58:24 <ollivier> I highlight how backward compatibility works in OpenStack. Functest (master) can test Queens and Pike but nothing is enforced for master.
08:59:17 <ollivier> No. I haven't seen any result from XCI for a while ad we will work with Tim after the release.
08:59:24 <LindaWang> The key is how functest can test Queens and Pike?
08:59:50 <ollivier> Openstack clients, rally tempest enforces that.
09:00:03 <ollivier> The only problem could be snaps
09:00:14 <LindaWang> yes,
09:00:33 <LindaWang> regarding snaps, I think snaps_smoke could be removed maybe
09:00:44 <ollivier> we have to consider openstacksdk for basic operations if we need to track master.
09:01:07 <ollivier> I don't see the point to remove snaps_smoke?
09:01:46 <LindaWang> since tempest and refstack_defcore have already exercised most of openstack APIs
09:02:30 <LindaWang> Openstack/Shade  is also a good choice
09:02:58 <LindaWang> Yardstick has get started with openstack/shade to deal with openstack utils
09:03:16 <ollivier> I think the target is openstacksdk
09:03:58 <ollivier> But from the time being nothing is clear and we will discuss together before switching to OpenStack master.
09:04:24 <ollivier> I switched to Queens quickly to get results before installer blackout.
09:05:03 <LindaWang> ok, Shade is a wrapper of openstacksdk
09:05:19 <ollivier> We may have to rewrite healthcheck and to consider tempest as gating but we need first result.
09:06:26 <ollivier> But as soon as Fraser is released, we will focus on that problem (meeting included)
09:07:21 <ollivier> BYW I was not part of the early discussions about the new release model (before LA)
09:07:29 <ollivier> #endmeeting