16:00:05 <tiffanie> #startmeeting openstack_ansible_meeting
16:00:06 <openstack> Meeting started Tue Oct  1 16:00:05 2019 UTC and is due to finish in 60 minutes.  The chair is tiffanie. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:07 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:00:09 <openstack> The meeting name has been set to 'openstack_ansible_meeting'
16:00:12 <tiffanie> #topic office hours
16:00:14 <noonedeadpunk> o/
16:00:19 <cjloader> o/
16:00:51 <noonedeadpunk> cjloader: how is your ironic patch doing?
16:01:17 <cjloader> i'm testing on hardware now
16:01:43 <cjloader> to ensure that it is not breaking ironic
16:02:16 <noonedeadpunk> ok, great
16:02:39 <jrosser> o/
16:06:08 <guilhermesp> o/
16:08:54 <noonedeadpunk> so I'm a bit stuck with murano right now. Tempest tests that never ends... But during looking inside I've just found that it's nova instance that stuck in build state. Hadn't dig into this yet.
16:09:40 <noonedeadpunk> jrosser: I see you had some progress with mind-to-mgmt ?
16:10:44 <guilhermesp> i guess the last chance was a couple of days ago https://review.opendev.org/#/q/topic:bind-to-mgmt+(status:open+OR+status:merged)
16:11:10 <guilhermesp> but I see almost a greenfield here https://review.opendev.org/#/c/670051/ :)
16:11:29 <noonedeadpunk> yep, and debian is kinda worth rechecking
16:11:45 <noonedeadpunk> so that's what I was asking:)
16:12:22 <guilhermesp> so let me recheck that
16:15:10 <jrosser> guilhermesp: yes we are very very close
16:15:34 <jrosser> i wonder if we should just merge all the "Add global override for service bind address" patches
16:15:58 <jrosser> then even if we don't finish the detail of galera for train we already have a set of overrides that work for multinode
16:15:59 <noonedeadpunk> why not?
16:16:19 <jrosser> i mean we should merge them, i think
16:16:25 <guilhermesp> i just rechecked some that has some isolated jobs failing
16:16:46 <noonedeadpunk> Just remove -W and I'm pretty ready for reviews
16:17:00 <jrosser> possibly some minor py3 bugs revealed by these as some roles we've not really touched since switching to py3
16:17:31 <jrosser> so os_glance is failing this now always openstack-ansible-nfs_glance
16:17:43 <jrosser> which maybe we should just get rid of?
16:18:08 <guilhermesp> btw, looking at https://review.opendev.org/#/c/683080/ do we still need this job? https://github.com/openstack/openstack-ansible-os_glance/blob/ae681ded4e6bc033c75f72b7a429a5b04dd0a676/zuul.d/jobs.yaml#L16
16:18:20 <guilhermesp> yeah jrosser
16:19:47 <jrosser> if anyone really wants that we should convert it to a scenario in the integrated repo
16:20:19 <guilhermesp> agreed
16:20:55 <noonedeadpunk> jrosser: it's failing due to the same tempest thing
16:21:09 <jrosser> ok, so i removed all the -W from the bind-to-mgmt patches
16:21:15 <noonedeadpunk> So, we can just fix tempest and leave it as is as for now
16:21:17 <jrosser> noonedeadpunk: ah interesting, how did you find that?
16:21:23 <jrosser> i could not find the tempest log easily
16:22:20 <jrosser> another outstanding thing is i've not been able to land any of the ansible 2.8 patches to the test repo
16:22:34 <jrosser> they fail in obscure ways looks like the connection plugin
16:22:47 <jrosser> which we don't see at all in the integrated repo
16:23:14 <jrosser> for example https://review.opendev.org/#/c/659658/
16:24:43 <openstackgerrit> Jonathan Rosser proposed openstack/openstack-ansible-os_masakari master: Add global override for service bind address  https://review.opendev.org/683138
16:25:40 <noonedeadpunk> I've scrolled through ara error and found that the same keystone test fails
16:33:58 <openstackgerrit> Jonathan Rosser proposed openstack/openstack-ansible-os_glance master: Add global override for service bind address  https://review.opendev.org/683080
16:34:07 <jrosser> noonedeadpunk: ^ that should test the glance / tempest thing
16:36:04 <noonedeadpunk> actually https://review.opendev.org/#/c/685974/ seems to fail:(
16:36:53 <noonedeadpunk> ok, but that's completely another thing I'd say....
16:37:06 <openstackgerrit> Merged openstack/openstack-ansible master: Collect logs from auditd in CI jobs  https://review.opendev.org/681241
16:37:13 <chandankumar> jrosser: noonedeadpunk what was the tempest issue?
16:37:43 <jrosser> chandankumar: https://review.opendev.org/#/c/685974/
16:40:46 <jrosser> so one of those is failing on this https://github.com/openstack/tempest/blob/master/tempest/api/identity/admin/v3/test_trusts.py#L34-L35
16:41:52 <jrosser> maybe becasue of this? https://github.com/openstack/openstack-ansible-os_tempest/blob/master/templates/tempest.conf.j2#L69-L70
16:43:04 <noonedeadpunk> let's check
16:43:05 <jrosser> then the second one is here https://github.com/openstack/tempest/blob/master/tempest/api/identity/base.py#L315-L317
16:44:43 <chandankumar> checking
16:44:43 <jrosser> needs application_credentials=true as well in [identity-feature-enabled]
16:46:12 <openstackgerrit> Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_tempest master: Fix tempest source install condition  https://review.opendev.org/685974
16:46:49 <chandankumar> as per this https://github.com/openstack/tempest/blob/master/tempest/config.py#L251 -> it should be enabled by default
16:47:40 <jrosser> chandankumar: default=False?
16:47:46 <noonedeadpunk> but default is false
16:47:47 <noonedeadpunk> ++
16:48:00 <jrosser> or you mean by default we should enable it :)
16:48:16 <chandankumar> we need to enable by default to true
16:48:21 <chandankumar> I am checking centos one
16:50:13 <chandankumar> jrosser: noonedeadpunk in centos it is skipped https://openstack.fortnebula.com:13808/v1/AUTH_e8fd161dc34c421a979a9e6421f823e9/zuul_opendev_logs_53d/685974/2/check/openstack-ansible-deploy-aio_distro_metal-centos-7/53d5c75/logs/openstack/aio1-utility/stestr_results.html
16:50:31 <chandankumar> setUpClass (tempest.api.identity.admin.v3.test_trusts
16:51:23 <chandankumar> https://eb7fd4fca907b0349dac-fddd31f569f44d98bd401bfd88253d97.ssl.cf1.rackcdn.com/685974/2/check/openstack-ansible-deploy-aio_metal-debian-stable/6801486/logs/openstack/aio1-utility/stestr_results.html -> failed but why
16:51:47 * chandankumar looks deeper
16:52:04 <noonedeadpunk> that's rly interesting question... Maybe new condition doesn't work as expected?
16:52:51 <noonedeadpunk> I mean which changed with that patch...
16:54:08 <jrosser> so we've now moved to master git version of tempest rather than pip
16:54:35 <noonedeadpunk> yep, that might be the case as wel.. But why centos do not fail then..
16:54:48 <noonedeadpunk> but skips as expected
16:55:01 <jrosser> oh well, we've not bumped the master sha have we
16:55:25 <jrosser> so now we could have the opposite problem that tempest is too new for the versions of the service we run
16:56:17 <noonedeadpunk> I'd say we did https://opendev.org/openstack/openstack-ansible/src/commit/e2c7eb271cc79e8077d2ac6e13f614053d2585f8/playbooks/defaults/repo_packages/openstack_testing.yml
16:56:44 <noonedeadpunk> but yeah, we didn't test it actually
16:56:51 <jrosser> right - but this https://review.opendev.org/#/c/683821/
16:57:04 <jrosser> ^ thats failing becasue tempest pip is too old for the service sha
16:57:23 <jrosser> this has the feeling of a circular problem :(
16:57:25 <chandankumar> from this https://eb7fd4fca907b0349dac-fddd31f569f44d98bd401bfd88253d97.ssl.cf1.rackcdn.com/685974/2/check/openstack-ansible-deploy-aio_metal-debian-stable/6801486/logs/ara-report/
16:57:37 <chandankumar> https://github.com/openstack/tempest/commit/8d1b36559c499c588c2a2644cd80cbcfeeda4291 got installed
16:58:21 <noonedeadpunk> but jrosser means that we probably need newer one
16:58:29 <noonedeadpunk> mich makes kinda cross-dependency
16:58:42 <noonedeadpunk> If I understand correctly
16:58:44 <chandankumar> noonedeadpunk: may be due this https://github.com/openstack/openstack-ansible-os_tempest/blob/master/defaults/main.yml#L244 it is coming?
16:58:59 <jrosser> so if we rewind a bit, this is where it all started https://review.opendev.org/#/c/682460/
16:59:50 <jrosser> there was a fix in keystone, we needed to bump the sha to get that
17:00:05 <jrosser> but then * was broken because in turn that needed a newer tempest
17:01:37 <tiffanie> #endmeeting