08:58:01 <gmann> #startmeeting QA Office Hours
08:58:01 <openstack> Meeting started Thu Jan 18 08:58:01 2018 UTC and is due to finish in 60 minutes.  The chair is gmann. Information about MeetBot at http://wiki.debian.org/MeetBot.
08:58:02 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
08:58:04 <openstack> The meeting name has been set to 'qa_office_hours'
08:58:14 <gmann> who all here today
08:59:21 <gmann> andreaf: masayukig ping
09:01:13 <andreaf> o/
09:01:30 <andreaf> I'll just be a minute than I'll be here for office hours
09:01:35 <gmann> seems like 2 of us
09:01:43 <gmann> sure
09:02:44 <andreaf> gmann: good afternoon
09:03:01 <gmann> andreaf: good morning for you :)
09:03:03 <gmann> lets start
09:03:14 <gmann> #topic Bug triage
09:03:41 <gmann> i have not generated the bug report but we can check the new bug
09:03:52 <gmann> #link https://bugs.launchpad.net/tempest/+bugs?search=Search&field.status=New
09:04:02 <gmann> this is tempest one ^^
09:04:25 <gmann> #link https://bugs.launchpad.net/tempest/+bug/1741554
09:04:26 <openstack> Launchpad bug 1741554 in tempest "TypeError: _unshelve_server() takes no arguments" [Undecided,New]
09:05:51 <gmann> this seems coding error
09:06:20 <andreaf> ok... ERROR tempest.test [-] tearDownClass does not call the super's tearDownClass in these classes tempest.api.compute.servers.test_server_actions.ServerActionsTestJSON
09:06:46 <andreaf> that is a tempest test but it seems weird that it would not be raised elsewhere
09:07:01 <andreaf> I can have a look at that one since i implemented that check :)
09:07:14 <gmann> nice,
09:07:44 <gmann> even _unshelve_server() things which runs on  addOnException
09:08:26 <openstackgerrit> Merged openstack/tempest master: Skip testcases when shared nets are found in AutoAllocateNetworkTest  https://review.openstack.org/530831
09:09:45 <gmann> andreaf: you want to assign to yourself or comment there ?
09:11:12 <andreaf> added a comment
09:11:17 <openstackgerrit> Merged openstack/tempest master: Remove 'deactivate_image' config options  https://review.openstack.org/526885
09:11:33 <openstackgerrit> TommyLike proposed openstack/tempest master: Replace volume size with configured value  https://review.openstack.org/535201
09:12:06 <gmann> thanks
09:12:22 <gmann> next
09:12:24 <gmann> #link https://bugs.launchpad.net/tempest/+bug/1702843
09:12:25 <openstack> Launchpad bug 1702843 in tempest "test_boot_server_from_encrypted_volume_luks has insufficient skip decorator" [Undecided,New]
09:13:57 <gmann> we should use 'attach_encrypted_volume' config option here instead of checking cpeh and nfs in skip
09:14:40 <openstackgerrit> Merged openstack/tempest master: Relocate branch conditional for tempest-full job  https://review.openstack.org/531305
09:15:37 <andreaf> do we have that  config option already?
09:15:40 <gmann> but zhufl asked question on that about attach_encrypted_volume means booting server from encrypted volume also
09:16:23 <gmann> we have but need to check and confirm whether both feature 1. attach encrypted volume and 2. boot server from encrypted volume is same things
09:16:36 <gmann> means supported as together
09:17:02 <andreaf> ok I see
09:17:35 <gmann> but anyways current condition is not good, either we should use existing config option or add new one
09:17:48 <gmann> bug is valid that no way to skip this test
09:20:30 <gmann> added comment there and triaged
09:21:23 <gmann> next -
09:21:25 <gmann> #link https://bugs.launchpad.net/tempest/+bug/1743996
09:21:26 <openstack> Launchpad bug 1743996 in tempest "test_list_endpoints_for_token failed when heat is installed" [Undecided,New]
09:23:37 <andreaf> that looks like something in the heat plugin may be breaking things but not sure
09:24:43 <gmann> yea, heat_plugin can be enabled and its endpoints not found in keystone
09:24:50 <andreaf> looking at the test code, this means that there is a heat_plugin service available setting
09:24:56 <andreaf> which is not found in the catalogue
09:25:45 <andreaf> that's an interesting one, since the test may break if a plugin defines a service available which does not match a catalogue entry
09:25:59 <andreaf> and we don't really have a strong contract on that
09:26:00 <gmann> yea, i am not sure heat_plugin should be as separate service
09:26:51 <andreaf> gmann: well I guess heat just wants to have a way to skip all tests from in-tree plugin by setting that to false
09:27:02 <andreaf> gmann: which I guess is legitimate
09:27:16 <andreaf> gmann: I think the test should only use tempest services
09:27:35 <andreaf> gmann: so maybe we can limit that to tempest service
09:27:36 <gmann> andreaf: agree, we should check for tempest services only
09:27:38 <andreaf> services
09:27:43 <gmann> +1
09:28:12 <gmann> i know ceiolmeter add aodh_plugin as service_avaiable also which also can cause same issue
09:28:58 <andreaf> I'll make a patch for that
09:30:35 <gmann> andreaf: thanks added comment on bug
09:31:09 <gmann> next-
09:31:14 <gmann> link https://bugs.launchpad.net/tempest/+bug/1733498
09:31:15 <openstack> Launchpad bug 1733498 in tempest "image.v1.test_image_members failed for “The success code is different than the expected one“" [Undecided,New]
09:34:09 <gmann> api-ref says 204 status code - https://developer.openstack.org/api-ref/image/v1/#add-member-to-image
09:34:18 <gmann> which is what test is checking.
09:36:02 <gmann> glance seems to return same
09:36:03 <gmann> #link https://github.com/openstack/glance/blob/08e28cd921920cf8f5401c839376b65c19a97880/glance/api/v1/members.py#L171
09:36:52 <gmann> do we run these test in gate?
09:37:11 <andreaf> I thought we removed all image.v1 tests
09:37:38 <andreaf> uhm they are still there
09:37:52 <andreaf> but I don't think we run them in the gate?
09:38:38 <gmann> andreaf: yea, we skip them
09:39:37 <gmann> i am not sure how he is getting 200 status code.
09:41:44 <gmann> i will check by running that API if glance return 200
09:41:56 <gmann> and accordingly triage the bug.
09:41:58 <andreaf> I cannot see any record of a run in o-h either
09:42:24 <gmann> yea, we should remove tests anyways.
09:42:59 <gmann> but with putting that on ML first and glance team also agree. i remember they said yes and thats why we skip them
09:43:08 <gmann> i can check that.
09:43:39 <andreaf> actually it is in o-h
09:43:53 <gmann> is it
09:44:09 <andreaf> http://status.openstack.org/openstack-health/#/tests/tempest?limit=1000&searchTest=tempest.api.image.v1.test_image_members.ImageMembersTest&offset=2000
09:44:18 <andreaf> we used to run it successfully
09:44:25 <andreaf> now we skip it though
09:44:56 <gmann> yea, i did not find the any code change from glance side since mitaka
09:45:10 <andreaf> no we still run it somewhere I guess, perhaps on periodic
09:45:25 <andreaf> o-h does not really help me to find out where the runs come from..
09:45:41 <andreaf> anyways it passes
09:45:48 <gmann> yea
09:45:50 <andreaf> http://status.openstack.org/openstack-health/#/test/tempest.api.image.v1.test_image_members.ImageMembersTest.test_add_image_member?duration=P3M
09:46:18 <andreaf> so something is wrong / different in the setup that lead to that bug
09:46:27 <andreaf> perhaps more logs / configs would help
09:46:34 <andreaf> I would mark that as incomplete
09:46:52 <gmann> yea, first let me check on current env of pike
09:47:16 <andreaf> if I can go back to https://bugs.launchpad.net/tempest/+bug/1743996 for a sec
09:47:17 <openstack> Launchpad bug 1743996 in tempest "test_list_endpoints_for_token failed when heat is installed" [Low,Triaged] - Assigned to Andrea Frittoli (andrea-frittoli)
09:47:27 <gmann> sure
09:47:42 <andreaf> that test checks that what's returned by keystone matches what's configured on tempest side
09:48:11 <andreaf> but first of all that only checks that tempest configuration is correct which does not seem like a test worth having to me
09:48:44 <andreaf> and also even for tempest services there's no guarantee that the catalog name will match the service_available flag
09:48:56 <andreaf> so tbh I would just remove the test for good
09:49:51 <gmann> currently it is covering this API -tokens/%s/endpoints
09:50:29 <andreaf> gmann: yeah but a good test should create an endpoint and check if it is returned by the list I think
09:51:10 <gmann> but if there is no endpoints in keystone for services tempest configured then tests will fail anyways right
09:51:31 <gmann> till we fallback to endpoints matches by other param like region etc
09:52:28 <andreaf> gmann: what we use to filter is the catalog type, so we could test with that instead of the service_available flag
09:52:29 <gmann> how about checking only identity endpoints
09:53:08 <andreaf> gmann: we don't check identity today because there is not service_available flag for identity :D
09:53:09 <gmann> catalog_type seems more correct
09:53:54 <gmann> ah right :)
09:54:22 <gmann> anyways this test is not being run by defcore so m not strongly oppose to remove this
09:55:03 * gmann 3 min left
09:56:01 <gmann> andreaf: anyways please propose and we can discuss on review to fix it or remove
09:56:03 <gmann> let's move to open topic
09:56:07 <gmann> #topic open
09:56:22 <gmann> andreaf: we discussed in last meeting about creating etherpad for Rocky PTG
09:56:37 <gmann> would you like to do that?
09:57:28 <andreaf> gmann: yeah I will do asap
09:57:35 <gmann> andreaf: thanks.
09:57:52 <gmann> andreaf: let's close the office hour if nothing else from your side
09:59:16 <andreaf> ok
09:59:19 <andreaf> gmann: https://etherpad.openstack.org/p/qa-rocky-ptg
09:59:26 <andreaf> I'll send an email
09:59:43 <gmann> andreaf: nice, thanks
09:59:48 <gmann> thanks for joining
09:59:50 <gmann> #endmeeting