08:00:51 <ollivier> #startmeeting Functest weekly meeting 30 Oct. 2018
08:00:51 <collabot`> Meeting started Tue Oct 30 08:00:51 2018 UTC.  The chair is ollivier. Information about MeetBot at http://wiki.debian.org/MeetBot.
08:00:51 <collabot`> Useful Commands: #action #agreed #help #info #idea #link #topic.
08:00:51 <collabot`> The meeting name has been set to 'functest_weekly_meeting_30_oct__2018'
08:01:02 <ollivier> #info Cédric Ollivier
08:01:06 <viktor_t> #info Viktor Tikkanen
08:01:07 <juhak> #info Juha Kosonen
08:02:02 <ollivier> delia_: could you print you name (#info) :)
08:02:52 <delia_> #info Delia Popescu
08:02:59 <ollivier> thank you
08:03:02 <delia_> :)
08:03:10 <ollivier> #topic Action point follow-up
08:03:26 <ollivier> #info juju_epc has been fixed
08:03:38 <ollivier> we now clone what seems to be the github mirror
08:03:47 <ollivier> #link https://github.com/RebacaInc/abot_charm/commit/c90cbae22c1f6b6f997a2048cb8ffac2d7c7f9e0
08:04:17 <ollivier> #info tempest_full and tempest_scenario pass vs Fuel master
08:04:30 <ollivier> All 4 active versions passed in Orange Openlab as well
08:04:41 <ollivier> (from Fraser to master)
08:04:57 <ollivier> Glance API v1, Cinder volumes backup and VNC tests were disabled
08:05:06 <ollivier> #link https://build.opnfv.org/ci/view/functest/job/functest-fuel-baremetal-daily-master/lastFailedBuild/console
08:05:27 <ollivier> Now only rally_full is not executed in gates. That's great!
08:05:44 <ollivier> (but we should still enhance gates to run rally_full)
08:06:19 <ollivier> #info vmtp is now protected vs unversioned OS_AUTH_URL
08:06:32 <ollivier> juhak: great catch! Thank you
08:06:41 <ollivier> #link https://gerrit.opnfv.org/gerrit/#/c/63913/
08:06:57 <ollivier> #info rally and rally-openstack have been updated in master and stable/hunter
08:07:04 <ollivier> #link https://gerrit.opnfv.org/gerrit/64113
08:07:12 <ollivier> #link https://gerrit.opnfv.org/gerrit/64115
08:07:24 <ollivier> Hunter and master are working well here.
08:08:24 <ollivier> #info our releng job was updated to write the custom tempest params per installer (novnc vs spice, ceph, etc...)
08:08:36 <ollivier> #link https://gerrit.opnfv.org/gerrit/#/c/64109/
08:10:10 <ollivier> I think we can safely create the OPNFV tags
08:10:14 <ollivier> That's all for the last actions. I suggest now to have a look to OPNFV installer results as we are closed to the release.
08:10:32 <ollivier> ok?
08:10:44 <juhak> fine
08:10:46 <delia_> ok
08:11:11 <ollivier> Let's start by Fuel it's currently the most active installer.
08:11:17 <ollivier> #topic Fuel results
08:11:29 <ollivier> #info 2 tests are failing (juju_epc and barbican) if os-nosdn-nofeature-ha
08:11:41 <ollivier> #link https://build.opnfv.org/ci/view/functest/job/functest-fuel-baremetal-daily-master/554/console
08:12:01 <ollivier> #link https://jira.opnfv.org/browse/FUEL-400
08:12:08 <delia_> barbican will pass after https://gerrit.opnfv.org/gerrit/#/c/63995/
08:12:40 <delia_> I saw juju_epc has different failure on different runs
08:12:51 <delia_> I updated the JIRA with 2 cases
08:13:42 <ollivier> yes. One is about firewall (quite trivial). the other I'm not convinced about internal connectivity in all your scenarios and pods.
08:14:22 <delia_> is it passing on other installers?
08:14:36 <delia_> I did not yet check
08:14:39 <ollivier> Yes here
08:14:54 <delia_> but apex or compass?
08:14:56 <ollivier> Since we switch to eurecom gitlab to github it should pass
08:15:15 <ollivier> last Apex run fails due the upstream changes but it passed before
08:15:29 <delia_> ok, I will look into it
08:16:27 <ollivier> From my understanding it works well.
08:17:04 <ollivier> Apex faced with issues when downloading from 8080 last release (LF rules)
08:17:25 <ollivier> Was it from the LF pod?
08:17:42 <ollivier> No ericsson-pod1
08:18:05 <delia_> from what I see in test results in apex is failing also
08:18:30 <ollivier> Apex haven't run for few days and faced with the former issues on gitlab
08:19:00 <delia_> I got to juju_ecp run starting 17.10
08:20:08 <delia_> no, I have a pass on fraser 2018-10-08  lf-pod1 on apex
08:20:47 <delia_> anyway I will investigate the failures
08:20:53 <ollivier> Apex gambia jjb just started https://build.opnfv.org/ci/view/apex/job/apex-daily-gambia/
08:21:11 <ollivier> I bet juju_epc will be successful in 5 hours.
08:21:21 <delia_> ok
08:21:49 <ollivier> Globally Fuel os-nosdn-nofeature is good.
08:22:09 <ollivier> #info 1 rally_jobs hung if os-odl-nofeature-ha
08:22:15 <ollivier> it seems fine now
08:22:25 <ollivier> #link https://build.opnfv.org/ci/view/functest/job/functest-apex-baremetal-daily-master/351/console
08:22:36 <ollivier> #info all advanced testcases failed in last os-odl-nofeature-ha run
08:22:44 <ollivier> #link https://build.opnfv.org/ci/view/functest/job/functest-fuel-baremetal-daily-gambia/62/console
08:23:23 <delia_> there is a bug in some neutron package, might me related, I am not sure
08:24:03 <ollivier> Badly we can't compare with other ODL results (Compass and Apex are not well configured) but I think the issues are related to Fuel here.
08:24:19 <ollivier> #info rally_jobs, vmtp and shaker are failing in ovs scenarios (+1 tempest test )
08:24:26 <ollivier> it seems about Fuel (badly Apex and Compass don't provide results)
08:25:37 <ollivier> I have doubled checked vmtp and shaker. Normally flavor extra spec are managed by Functest and not the third-parties (shaker or vmtp)
08:25:58 <ollivier> and singlevm2 passed they should be ok.
08:26:37 <delia_> I mught me able to verify that
08:26:42 <ollivier> Let me know if I can help you about that part. I would like to see that 2 testcases passing in DPDK scenarios.
08:26:49 <delia_> if shaker image has the prop
08:27:16 <ollivier> I tested here with small instead of huge and it seems well configured.
08:27:26 <delia_> I don't have a dpdk build now, but I can set the DEPLOY_SCENARIO to dpdk and see if the prop are passed
08:27:43 <delia_> I think it used to pass
08:28:12 <ollivier> FLAVOR_EXTRA_SPECS="hw:mem_page_size:small"
08:28:24 <delia_> small?
08:28:29 <ollivier> yes I think
08:28:54 <delia_> https://build.opnfv.org/ci/job/functest-fuel-baremetal-daily-master/557/consoleFull
08:29:00 <delia_> ovs, ha shaker pass
08:29:47 <ollivier> #link https://docs.openstack.org/nova/latest/user/flavors.html
08:30:29 <ollivier> ok they should be ok.
08:31:14 <ollivier> I don't see why rally_jobs could fail here. It the same logic as for rally.
08:31:18 <delia_> FLAVOR_EXTRA_SPECS="hw:mem_page_size:large"
08:31:23 <delia_> this is what we are using
08:31:28 <delia_> in releng
08:32:05 <ollivier> Yes but here I'm testing FLAVOR_EXTRA_SPECS="hw:mem_page_size:small" as I don't setup a DPDK scenario.
08:32:25 <delia_> ah, ok, I can do that to
08:32:38 <ollivier> I checked that shaker and vmtp vms were well configured.
08:32:42 <ollivier> thank you
08:32:58 <ollivier> but Fuel has always faced with troubles when running vmtp :)
08:33:47 <delia_> one question, not really related
08:33:51 <delia_> http://testresults.opnfv.org/test/#/results
08:33:56 <juhak> in rally_jobs, creation of networks, ports etc starts failing after 30 successful ones
08:34:04 <juhak> ServiceUnavailable: Unable to create the network. No tenant network is available for allocation.
08:34:16 <delia_> is it possible , like in the future, to filter based on test Case
08:34:57 <delia_> yes, I saw that, sometimes it has like no resp from keystone
08:35:21 <ollivier> juhak: thank you. Then rally_jobs does its job :)
08:35:35 <delia_> it should have nothing to do with the scenario
08:36:17 <ollivier> delia_: this interface is not so great. I have my own python script to get the latest from all installers (+scenarios)
08:36:46 <ollivier> Regarding all that issues, I'm still convinced that Functest is ok.
08:36:50 <delia_> ollivier: can you, please, share? :)
08:37:15 <delia_> it would be very useful
08:37:29 <ollivier> Sure. I used them to write the former global results in wiki page.
08:38:05 <ollivier> #action Cédric share python scripts to extract results from DB
08:38:08 <delia_> thenk you
08:39:17 <ollivier> Regarding arm, many jobs raised timeouts. they have mostly passed healthcheck but we lots all logs
08:40:13 <ollivier> the last one is not so bad https://build.opnfv.org/ci/view/functest/job/functest-fuel-armband-baremetal-daily-gambia/21/console (rally_jobs failed)
08:41:55 <delia_> we're on it :)
08:42:43 <ollivier> Great! Regarding release criteria, Fuel seems ok. Regarding Functest criteria, there are few remaining bugs in amd64  (Fw rules and possible connectivity issues) and we are waiting for arm64 logs.
08:43:32 <ollivier> #topic Compass
08:43:41 <ollivier> #info k8-nosdn-nofeature-ha master was verified
08:43:51 <ollivier> #link https://build.opnfv.org/ci/view/functest/job/functest-compass-baremetal-daily-master/369/console
08:44:05 <ollivier> #info it seems that only os-nosdn-bar-ha (master) is able to pass healthcheck but it raised the global timeout
08:44:13 <ollivier> snaps_smoke spent 3 hours
08:44:20 <ollivier> #link https://build.opnfv.org/ci/view/functest/job/functest-compass-baremetal-daily-master/368/console
08:44:27 <ollivier> no run has passed healthcheck in gambia
08:44:31 <ollivier> (baremetal)
08:45:18 <ollivier> #topic Apex
08:45:31 <ollivier> #info last gambia runs are failing due to side effects created by https://gerrit.opnfv.org/gerrit/#/c/63909/
08:46:37 <ollivier> A new gambia run just started. Let's see. https://build.opnfv.org/ci/view/apex/job/apex-daily-gambia/17/
08:46:52 <ollivier> next nosdn runs could pass all testcases (only juju_epc and tempest_full failed due to Glance API v1 and Cinder backup tests)
08:47:04 <ollivier> Or tempest.api.object_storage.test_crossdomain.CrossdomainTest.test_get_crossdomain_policy may fail
08:47:31 <ollivier> I think only Apex is deploying Swift. I will try to test it internally because it may be a tempest misconfiguraiton.
08:48:00 <ollivier> juhak: viktor_t: do you have a POD where swift is deployed?
08:48:24 <juhak> just a sec
08:49:43 <ollivier> #link https://build.opnfv.org/ci/view/functest/job/functest-apex-baremetal-daily-gambia/10/console
08:49:53 <ollivier> #info ODL testing is still misconfigured
08:50:02 <ollivier> #link https://gerrit.opnfv.org/gerrit/#/c/63909/
08:50:05 <juhak> ollivier: no, we don't have swift deployed
08:50:32 <ollivier> juhak: ok.
08:51:36 <ollivier> Regarding all that results, I think Functest is working well. We should check CrossdomainTest.test_get_crossdomain_policy and have a close to look to vmtp and shaker if DPDK.
08:52:00 <ollivier> I think we can safely create all tags.
08:52:26 <ollivier> I will create them tomorrow. (I will be in weekend tomorrow's evening)
08:52:45 <ollivier> 1/11 is day off here
08:54:22 <ollivier> Great work done by all contributors!
08:54:48 <ollivier> #topic AoB
08:55:24 <ollivier> I sent you yesterday my first results vs Raspberry PI 1 & 3. I try now to optimize a little bit
08:56:16 <ollivier> let me know if you're interested by the procedures.
08:56:36 <ollivier> I will execute tempest_full today...
08:56:56 <ollivier> Anything else?
08:57:11 <delia_> not from me
08:57:17 <juhak> no
08:58:11 <ollivier> ok. Thank you. Great work! Have a good week.
08:58:18 <ollivier> #endmeeting