09:05:16 <chandankumar> #startmeeting QA Office Hour
09:05:17 <openstack> Meeting started Thu Feb  1 09:05:16 2018 UTC and is due to finish in 60 minutes.  The chair is chandankumar. Information about MeetBot at http://wiki.debian.org/MeetBot.
09:05:18 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
09:05:20 <openstack> The meeting name has been set to 'qa_office_hour'
09:05:43 <chandankumar> gmann: check above ^^
09:05:54 <gmann> nice, this way we will all office hour logging here - http://eavesdrop.openstack.org/meetings/qa_office_hours/2018
09:06:03 <gmann> chandankumar: nice.
09:06:39 <chandankumar> gmann: should i proceed?
09:06:40 <gmann> we can start
09:06:43 <gmann> yup
09:06:50 <chandankumar> #link http://paste.openstack.org/show/658285/ -> Bugs to be traiged
09:07:09 <andreaf> I added a comment to https://bugs.launchpad.net/patrole/+bug/1745871
09:07:10 <openstack> Launchpad bug 1745871 in Patrole "RBAC tests for group type specs" [Wishlist,Confirmed] - Assigned to Felipe Monteiro (fm577c)
09:07:16 <andreaf> #link https://bugs.launchpad.net/patrole/+bug/1745871
09:07:32 <chandankumar> andreaf: thanks :-)
09:07:44 <andreaf> There's a patch linked there and its merged - but it doesn't say much it looks more like a working item than a bug
09:07:51 <chandankumar> #link https://bugs.launchpad.net/devstack/+bug/1743688
09:07:52 <openstack> Launchpad bug 1743688 in congress "Tempest unable to detect service availability properly, causing congress tests to fail" [Medium,In progress] - Assigned to Anusha (anusha-iiitm)
09:08:08 <gmann> andreaf: that bug is to add patroel tests right
09:08:44 <andreaf> gmann: I guess yeah
09:08:51 <gmann> andreaf: we have service client in tempest(patch you mentioned) but patrole test needs to write
09:09:46 <chandankumar> gmann: is working on the above bug
09:09:48 <andreaf> gmann: sure - in any case it's a patrole task more than a bug - let's move to the next one
09:10:09 <chandankumar> next one https://bugs.launchpad.net/devstack/+bug/1744096
09:10:11 <openstack> Launchpad bug 1744096 in devstack "CentOS install fails with 'python3: command not found'" [Undecided,In progress] - Assigned to Stephen Finucane (stephenfinucane)
09:10:44 <chandankumar> centos-7 does not provide python 3 we need to use software collections for that
09:11:03 <andreaf> gmann  https://bugs.launchpad.net/devstack/+bug/1743688 is not really a tempest bug right?
09:11:05 <openstack> Launchpad bug 1743688 in congress "Tempest unable to detect service availability properly, causing congress tests to fail" [Medium,In progress] - Assigned to Anusha (anusha-iiitm)
09:11:27 <chandankumar> https://review.openstack.org/#/c/535365/
09:11:37 <gmann> andreaf: yea updated the comment and ll fix plugin side
09:11:47 <andreaf> thanks
09:11:59 <gmann> devstack we need to update current list of service contain only tempest own service.
09:12:12 <gmann> but once plugin are fixed
09:14:09 <chandankumar> gmann: andreaf based on above bug this might be related https://bugs.launchpad.net/tempest/+bug/1746687
09:14:11 <openstack> Launchpad bug 1746687 in tempest "tempest plugins should be loaded though configuration" [Undecided,New]
09:15:08 <chandankumar> this one comes because i was working on kolla tempest images https://review.openstack.org/#/c/537807/ to create two containers image
09:15:21 <chandankumar> one having base tempest image and another with all plugins
09:16:21 <chandankumar> kolla suggested to keep one container with all tempest plugins in it and load tests based on config
09:16:30 <gmann> is it means we need to provide some tool or script to take all mentioned plugin and install ?
09:17:06 <chandankumar> I am not able to convince them to keep two containers for two use cases
09:18:08 <gmann> but why not you install all plugin and then create image
09:18:21 <gmann> like devstack does via TEMPEST_PLUGINS
09:18:41 <chandankumar> gmann: in devstack we specify which plugin to use in respective ci jobs
09:19:17 <gmann> i mean as part of image build in kolla, do not just pick tempest + script take requried plugin and install them
09:19:56 <andreaf> chandankumar: I'm not sure I understand why the docker image has to pre-install all plugins and then the CI job select which to use
09:20:20 <andreaf> chandankumar: that's the wrong way of using this, which is what is causing the issue
09:21:00 <chandankumar> andreaf: so you mean get a base tempest image then based on demand install plugins
09:21:02 <andreaf> chandankumar: but I don't know how the CI job works - it would be good to get a more specific issue perhaps
09:21:32 <andreaf> chandankumar: well yeah it depends what you need to test I can imagine many possible approaches
09:22:18 <andreaf> chandankumar: you can have required-projects on the job definition to pull in the plugin repos you need and then in the docker image mount that folder and install everything that is found on that mount point
09:23:00 <andreaf> chandankumar: but my ideas might be off - I don't know what the CI job wants to achieve
09:24:06 <chandankumar> my usecase was to 1. use tempest image if an user wants to run tempest tests only 2. if you need more one then one more service tempest tests, use all tempest plugins containers and run from there
09:26:37 <chandankumar> that's why i was proposing there two containers, but kolla team suggsted to bundle all plugins in a container so that user does not need to install any things and load tests based on config
09:27:42 <chandankumar> but loading tests is independent on config
09:29:17 <masayukig[m]> can you use whitelist/blacklist for that?
09:30:16 <chandankumar> masayukig[m]: yes that can be used while running tests
09:30:20 <andreaf> chandankumar: what's the workflow for a kolla user, how is a deployment defined? Is it possible for the deployment tool to build the right tempest image based on the services which are actually used?
09:31:32 <andreaf> chandankumar: I guess this is the docker container https://github.com/openstack/kolla/blob/master/docker/tempest/Dockerfile.j2
09:31:54 <chandankumar> current workflow which i am doing : on a containerized deployment, pull the tempest image, then install required plugin then run the required stuff
09:32:13 <gmann> but user want precreated images seems
09:32:44 <chandankumar> andreaf: yes that image
09:33:21 <gmann> i think all plugins image will be too haeavy
09:33:52 <chandankumar> andreaf: i donot think that is possible to build the right tempest image based on services which are actually build
09:34:12 <chandankumar> on a containerized deployment, images are prebuilt in a registery and fetched from there
09:34:15 <andreaf> chandankumar, gmann: so if you want to avoid requiring internet access you can pre-cache all plugins on the file system - they do not need to be installed
09:36:16 <gmann> andreaf: and install them based on env detected services ?
09:37:22 <andreaf> gmann: yeah well at some point the image needs to be configured - plugins could be installed at configuration time
09:37:24 <chandankumar> pre-caching means we need to keep all their dependencies also there and so that based on discoverd services they can be installed
09:37:37 <andreaf> gmann, chandankumar: the problem may be plugin dependencies though
09:37:44 <gmann> yea..
09:40:23 <andreaf> gmann, chandankumar:  so if you really want to have one image that contains tests for all possible services installable via kolla then yeah install all plugins and handle which tests to run via service_available flags
09:40:48 <chandankumar> i think we need a script to install plugins on demand based on service availabilty other way is to pull all the dependencies at one place for the required and then install the plugins during configuration
09:41:56 <gmann> chandankumar: that might be too complex from user point of view
09:42:20 <gmann> how big is 'all plugin installed' image?
09:42:52 <andreaf> but that's a testing image, it's something for CI pipelines I guess?
09:43:16 <andreaf> so people using that image is kind of technical users right?
09:44:03 <andreaf> and what is wrong with the service_available approach? what about using a whitelist to filter tests otherwise?
09:44:22 <andreaf> err... nose? https://github.com/openstack/kolla/blob/master/docker/tempest/Dockerfile.j2#L30
09:44:59 <gmann> heh, is it work with nose i doubt :)
09:45:07 <chandankumar> andreaf: users should be both, i am working on running tempest using containers in tripleo-ci
09:45:39 <gmann> andreaf: chandankumar yea i agree on same approach. install all and run with service_available basis or whitelist/backlist
09:45:43 <andreaf> tempest doesn't work with nose
09:45:47 <gmann> yea
09:45:57 <chandankumar> andreaf: the images requires cleanup
09:46:39 <gmann> ok, can we move next. time is already 50 min over
09:46:57 <chandankumar> gmann: andreaf sure
09:47:18 <gmann> link https://bugs.launchpad.net/tempest/+bug/1745322
09:47:19 <openstack> Launchpad bug 1745322 in tempest "stackviz folder does not show up in logs on zuulv3 native jobs" [Undecided,New]
09:47:24 <gmann> andreaf: ^^ this is fixed right
09:47:40 <andreaf> yeah
09:48:00 <andreaf> I forgot I had a bug for that :P
09:48:04 <gmann> :)
09:48:40 <andreaf> updated
09:48:55 <chandankumar> andreaf: thanks :-)
09:49:04 <chandankumar> moving to next
09:49:06 <chandankumar> https://bugs.launchpad.net/tempest/+bug/1745307
09:49:07 <openstack> Launchpad bug 1745307 in tempest "Group related cases would fail if driver has group spec check" [Undecided,In progress] - Assigned to YuZhang (ivysdu)
09:50:01 <chandankumar> https://review.openstack.org/537784 patch is up
09:51:02 <chandankumar> neutron dvr ci all tempest tests passed
09:52:17 <chandankumar> gmann: andreaf http://logs.openstack.org/84/537784/4/check/neutron-tempest-dvr/62bd05c/job-output.txt.gz#_2018-02-01_05_55_56_594439 here it is failed
09:52:21 <chandankumar> may be recheck will fix it
09:52:37 <gmann> chandankumar: ok, ll review it tomorrow
09:52:59 <chandankumar> gmann: andreaf can we set the priority on this one https://bugs.launchpad.net/tempest/+bug/1745307
09:53:00 <openstack> Launchpad bug 1745307 in tempest "Group related cases would fail if driver has group spec check" [Undecided,In progress] - Assigned to YuZhang (ivysdu)
09:53:57 <chandankumar> similar bug https://bugs.launchpad.net/tempest/+bug/1745307
09:54:12 <chandankumar> sorry
09:54:27 <chandankumar> next bug https://bugs.launchpad.net/tempest/+bug/1660612
09:54:29 <openstack> Launchpad bug 1660612 in neutron "Tempest full jobs time out on execution" [High,Confirmed]
09:55:00 <chandankumar> gmann: andreaf ^^ above bug needs attension
09:56:12 <chandankumar> related fix https://review.openstack.org/#/c/536598/
09:58:06 <andreaf> that fix is not ok... it changes existing tox envs which has an impact on existing jobs
09:59:14 <gmann> yea, it make it all scenario to run parallel
09:59:26 <gmann> which cause another issue
10:00:21 <gmann> but tempest full deos not run  scenario tests right
10:00:22 <chandankumar> andreaf: gmann please comment on the bz/review
10:00:28 <gmann> sure.
10:00:31 <andreaf> done
10:00:54 <andreaf> sorry folks I need to go thanks for the office hours today!
10:01:02 <chandankumar> andreaf: thanks :-)
10:01:05 <chandankumar> gmann: last one https://bugs.launchpad.net/devstack/+bug/1746671
10:01:06 <openstack> Launchpad bug 1746671 in devstack "Plugins checking out to master branch" [Undecided,New]
10:01:19 <gmann> time up seems.
10:01:44 * gmann need to hurry home as snow prediction and too dark here
10:01:46 <chandankumar> ack
10:01:51 <chandankumar> #endmeeting