07:59:49 <ollivier> #startmeeting Functest weekly meeting 6 Nov. 2018
07:59:49 <collabot`> Meeting started Tue Nov  6 07:59:49 2018 UTC.  The chair is ollivier. Information about MeetBot at http://wiki.debian.org/MeetBot.
07:59:49 <collabot`> Useful Commands: #action #agreed #help #info #idea #link #topic.
07:59:49 <collabot`> The meeting name has been set to 'functest_weekly_meeting_6_nov__2018'
08:00:07 <delia_> #info Delia Popescu
08:00:13 <ollivier> #info Cédric Ollivier
08:00:20 <juhak> #info Juha Kosonen
08:01:00 <ollivier> #topic Action point follow-up
08:01:23 <ollivier> #info juju_epc were fixed again (Abot's package was buggy during a day)
08:02:02 <ollivier> It should be ok now. I saw one mme compilation failures in Fuel
08:02:23 <ollivier> #info first Functest gambia and hunter tags were published
08:02:31 <ollivier> #link https://gerrit.opnfv.org/gerrit/#/c/64273/
08:02:42 <ollivier> #info functest-alpine.sh (releng) now generates tempest_conf.yaml for all OPNFV installers
08:02:55 <ollivier> https://gerrit.opnfv.org/gerrit/#/c/64109/
08:03:27 <ollivier> I think this file should be installer centric. If it's about scenarios, we could still blacklist in tempest.
08:03:51 <ollivier> (next topic about OVN)
08:04:12 <ollivier> #info Apex rocky jobs now launch Functest hunter containers
08:04:21 <ollivier> All the other installers are sill running latest instead of hunter
08:04:50 <ollivier> It works well anyway thanks to backward compatibility.
08:06:32 <delia_> are there differences between hunter and latest?
08:07:22 <ollivier> yes. Hunter select the official Rocky package. Latest selects the last package or even clones master repositories (tempest)
08:07:36 <delia_> ok
08:08:00 <delia_> fuel is running queens so it should stay with latest..
08:08:21 <delia_> thanks for clarifying it for me
08:08:35 <ollivier> It's not so simple to conform with OpenStack master gates but latest works well vs Rocky and even Queens here.
08:10:07 <ollivier> no pb. Fuel should run gambia in its master jjbs from the time being.
08:10:26 <delia_> ok
08:10:36 <ollivier> and switch to hunter after the first hunter milestones.
08:10:59 <ollivier> Here are few bugs opened in installers:
08:11:00 <ollivier> https://jira.opnfv.org/browse/APEX-645
08:11:42 <ollivier> It has to be fixed in Apex then it will get the blue ball (it should be noted it's about swift)
08:11:48 <ollivier> juhak: Great catch.
08:12:16 <ollivier> Patrole fails due to L2GW misconfigurations (https://jira.opnfv.org/browse/APEX-646)
08:12:35 <ollivier> last 2 tempest_full tests in failure (master) (https://jira.opnfv.org/browse/APEX-647)
08:12:49 <ollivier> Connecion to 115.249.81.187:8080 failed (Connection timed out)  (https://jira.opnfv.org/browse/FUEL-400)
08:13:09 <ollivier> test_dualnet_multi_prefix_slaac fails in ovs scenario (https://jira.opnfv.org/browse/FUEL-402)
08:13:57 <ollivier> FUEL-400 was closed yesterday. Let's see the next runs.
08:14:29 <ollivier> Any comment about action points?
08:15:29 <ollivier> #topic Functest
08:15:38 <ollivier> #info Fuel os-nosdn-nofeature-ha master (Functest latest) got a blue ball
08:15:46 <delia_> I saw vmtp test from time to time gets timeout on waiting for server to come up
08:16:02 <delia_> i am thinking to increase boot_timeout
08:16:06 <delia_> for vmtp
08:17:11 <ollivier> It conforms with the classical timeout. https://git.opnfv.org/functest/tree/functest/opnfv_tests/openstack/vmtp/vmtp.py#n118
08:17:24 <ollivier> I'm fine to increase that. What's your proposal value?
08:18:35 <delia_> 300 from 180
08:18:43 <delia_> I think that is in shaker tets too
08:19:31 <delia_> yes, 300 is for shaker
08:20:07 <ollivier> Yes. I'm updating it right after the meeting. Thank you.
08:20:23 <delia_> thank you
08:20:39 <ollivier> It currently keeps default from singlevm.
08:20:46 <delia_> I know
08:21:22 <ollivier> #link https://build.opnfv.org/ci/view/functest/job/functest-fuel-baremetal-daily-master/571/console
08:21:32 <ollivier> heat_ims testing was not so good but passed the current criteria (at least 1 successful test).
08:22:16 <ollivier> #info Apex os-nosdn-nofeature-* (and callipso) will get a blue ball as soon as APEX-645 is closed (swift cross domains)
08:22:26 <ollivier> #link https://build.opnfv.org/ci/view/functest/job/functest-apex-baremetal-daily-gambia/26/console
08:22:33 <delia_> I am facing difficulties in testing tests that do not run on arm too
08:22:50 <delia_> no pod available at this point
08:23:08 <ollivier> we are installing new pods for Functest.
08:23:48 <delia_> great
08:23:52 <delia_> :))
08:24:05 <ollivier> I could give you an access to Orange Openlab but it only allows testing ONAP if I'm not wrong.
08:24:21 <ollivier> I will double check.
08:24:38 <delia_> ok, thank you
08:25:27 <ollivier> #info Apex os-ovn-nofeature highlighted possible bugs in Functest and Snaps
08:25:49 <ollivier> rally_jobs couldn't create/list trunks https://artifacts.opnfv.org/logs/functest/lf-pod1/gambia/2018-11-05_11-43-15/rally_jobs/neutron.html#/NeutronTrunks.create_and_list_trunk_subports
08:26:50 <ollivier> tempest is able to test trunks
08:26:51 <ollivier> https://artifacts.opnfv.org/logs/functest/lf-pod1/gambia/2018-11-05_11-43-15/neutron_trunk/tempest-report.html
08:27:24 <juhak> that's interesting...
08:27:25 <ollivier> juhak: could you please have a look to that results? It seems that ovn defines its own rally tasks (I quickly checked yesterday).
08:27:38 <juhak> yes, need to check
08:27:42 <ollivier> thank you
08:27:58 <ollivier> few tempest, neutron-api-plugin and patrole tests have to be blacklisted (agent testing)
08:28:56 <ollivier> I think we could blacklist them in blacklist.txt https://git.opnfv.org/functest/tree/functest/opnfv_tests/openstack/tempest/custom_tests/blacklist.txt
08:29:49 <ollivier> If I'm not wrong, blacklisting could depend on scenarios.
08:31:10 <ollivier> https://git.opnfv.org/functest/tree/functest/opnfv_tests/openstack/tempest/tempest.py#n213
08:32:30 <ollivier> ok?
08:32:41 <juhak> ok
08:32:57 <ollivier> https://artifacts.opnfv.org/logs/functest/lf-pod1/gambia/2018-11-05_11-43-15/neutron-tempest-plugin-api/tempest-report.html
08:33:07 <ollivier> snaps_smoke failed due to the MTU hardcoded (1450>1442)
08:33:34 <ollivier> https://gerrit.opnfv.org/gerrit/#/c/64515/ (merged)
08:34:26 <ollivier> We need to update snaps commit id (no snaps gambia branch). I would prefer running snaps_smoke before the review because it could include few patches in a row.
08:34:43 <ollivier> juju could not find an external network in availability zone
08:36:00 <ollivier> From my understanding, nosdn and ovn scenarios are mostly good. From the time being I'm not convinced by the results of ODL and OVS DPDK.
08:36:14 <ollivier> They both fail on advanced testcases.
08:37:09 <ollivier> Let's see if vmtp could pass after updating timeouts. But vnf wouldn't run anyway.
08:37:10 <delia_> yes, I agree. I sonly the case for fuel on other installers too?
08:38:38 <ollivier> It's the same for Apex.
08:39:11 <ollivier> I haven't seen any Compass ODL run (I may be wrong) since I fixed the env vars.
08:40:14 <ollivier> I think we should enable VNFs in all cases. From the time being, they only run if nofeature but they would have passed in callipso. They are also disabled in DPDK scenarios
08:40:41 <ollivier> I don't think Apex support (or deploy) OVS DPDK.
08:41:00 <ollivier> and Compass DPDK failed in singlevm.
08:41:38 <ollivier> Do you agree? I think we should run the same tests in all scenarios.
08:42:39 <delia_> yes
08:42:51 <juhak> fine by me
08:43:08 <ollivier> great.
08:43:27 <ollivier> #info a first jjb has been proposed to verify Functest on functest-virtual
08:44:00 <ollivier> It could be executed on any computer if a functest-virtual label is defined.
08:44:29 <ollivier> https://gerrit.opnfv.org/gerrit/#/c/64461/
08:44:59 <ollivier> It just a first patch to ensure the process is right. We also now have a POD.
08:46:32 <ollivier> It leverages on, Apex as we require OpenStack master deployment to ensure the whole process. I'm fine to leverage on multiple installers as long as they are stable and firstly verified by Functest.
08:47:56 <ollivier> any additional comment about Functest?
08:49:13 <delia_> not from me
08:49:46 <juhak> the status looks good to me
08:50:34 <ollivier> yes fully agree. Few enhacement about last ovn runs but that's great. (tempest_full included)
08:50:55 <ollivier> #topic AoB
08:51:16 <ollivier> It  seems that the release will be voted today. (I was postponed from one week).
08:52:02 <ollivier> I'm quite concerned about ODL and OVS DPDK scenarios and YardStick else I'm fine.
08:52:55 <ollivier> The scenario list on wiki page mix OpenStack Rocky and Queens deployments. But it seems a mistake.
08:53:10 <ollivier> #info testing from Raspberry PI is becoming quite decent after removing logs (rally)
08:54:13 <ollivier> From the time being, I simply muted debug logs and rally seems ok. Creating tmpfs mount points for Functest and Rally outputs don't improve the duration.
08:54:55 <ollivier> I can write the procedure on wiki if you're interested to run tests on your own.
08:55:54 <juhak> yes please :)
08:56:34 <ollivier> sure. I will write a first draft quickly and complete end of week :)
08:56:47 <juhak> thank you
08:57:43 <ollivier> That's all here. Anything other business?
08:58:25 <delia_> nope
08:58:41 <juhak> not from my side
08:59:34 <ollivier> Then thank you very much. Have a nice week.
08:59:40 <ollivier> #endmeeting