13:30:51 <gmann> #startmeeting qa 13:30:53 <openstack> Meeting started Tue May 26 13:30:51 2020 UTC and is due to finish in 60 minutes. The chair is gmann. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:30:54 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:30:56 <openstack> The meeting name has been set to 'qa' 13:31:00 <gmann> who all are here today ? 13:31:24 <kopecmartin> hello 13:31:57 <paras333> hi 13:32:27 <gmann> let's start 13:32:34 <gmann> #link https://wiki.openstack.org/wiki/Meetings/QATeamMeeting#Agenda_for_next_Office_hours 13:32:36 <gmann> today agenda ^^ 13:33:04 <gmann> #topic Announcement and Action Item (Optional) 13:33:16 <gmann> no announcement form my side or AI from previous meeting 13:33:47 <gmann> skipping the ussuri tracking, we need to update the agenda after PTG and start Victoria cycle tracking 13:33:50 <gmann> #topic OpenStack Events Updates and Planning 13:34:09 <gmann> next week is Virtual PTG 13:34:11 <gmann> #lin https://etherpad.opendev.org/p/qa-victoria-ptg 13:34:13 <gmann> #link https://etherpad.opendev.org/p/qa-victoria-ptg 13:34:17 <openstackgerrit> Sam Kumar proposed openstack/tempest master: Sort roles in keystone token https://review.opendev.org/729803 13:35:38 <gmann> we have schedule up for topic wise also but we will wait for the conference channel. it might be zoom i think 13:36:07 <gmann> anything we need to prepare/plan for PTG which we missed ? 13:36:12 <gmann> or any feedback 13:37:39 <gmann> ok, let's move 13:37:40 <gmann> #topic Sub Teams highlights 13:37:42 <gmann> Tempest 13:37:58 <gmann> there were two bugs for plugins and tempest testing 13:38:20 <gmann> #link http://lists.openstack.org/pipermail/openstack-discuss/2020-May/014601.html 13:38:47 <gmann> sorry 13:38:49 <gmann> #link http://lists.openstack.org/pipermail/openstack-discuss/2020-May/015011.html 13:38:52 <gmann> this one ^^ 13:39:43 <gmann> fix on devstack side work fine but i need to change the tempest role run-tempest also to use the stable constraint when old tempest is used even stein or train gate 13:40:05 <gmann> i did not find time yesterday, may be i will look tomorrow 13:40:31 <gmann> this is hitting 3rd party CI as none of upstream job on stein and train gate use older tempest 13:41:13 <gmann> anything else to discuss on Tempest side ? 13:42:25 <tosky> gmann: will that affect other users of the run-tempest role in branched repositories 13:42:27 <tosky> ? 13:42:30 <tosky> (like grenade) 13:43:26 <gmann> tosky: it should not, I have not found the way but what i will be doing is to check if old tempest is being used irrespective of branch so that stein, train, ussuri using master tempest has no change 13:44:40 <gmann> but i need to check how i can get value of TEMPEST_BRANCH var in run-tempest 13:45:04 <gmann> which is what set by jobs when using the old tempest 13:46:37 <gmann> but yes, i will test that properly so that it does not disturb the current usage 13:47:28 <gmann> Patrole 13:47:42 <gmann> nothing to share from my side on this. 13:48:06 <gmann> i have one update to share on hacking 13:48:39 <gmann> i am reverting the py35 job and updated the req constraint for py35 which needs to be done on hacking side 13:48:41 <gmann> #link https://review.opendev.org/#/c/730686/ 13:48:54 <gmann> this working fine. need review on this. masayukig ^^ 13:49:03 <gmann> same i have done on ksa side. 13:50:20 <gmann> let's move 13:50:21 <gmann> #topic Community goal tracking 13:50:36 <gmann> #link https://etherpad.opendev.org/p/qa-community-wide-goals-tracking 13:51:02 <gmann> we still waiting for 2nd goal for V cycle and zuulv3 migration is all good for QA but need audit for all repo 13:51:38 <gmann> #topic Bug Triage 13:51:48 <gmann> #link https://etherpad.opendev.org/p/qa-bug-triage-ussuri 13:51:56 <gmann> kopecmartin: any news on bugs 13:52:12 <kopecmartin> tempest bug numbers look still good, slowly decreasing \o/ 13:52:30 <kopecmartin> one question regarding the roles used in tempest, see this bug 13:52:40 <kopecmartin> #link https://bugs.launchpad.net/tempest/+bug/1330132 13:52:40 <openstack> Launchpad bug 1330132 in tempest "Creation of Member role is no longer required" [Undecided,Confirmed] 13:53:10 <kopecmartin> i found it confusing .. which is the current one? i think that Member role was replaced by _member_ one, right? 13:53:25 <kopecmartin> but _member_ role is deprecated too, isn't it? 13:53:49 <tosky> by member 13:53:57 <openstackgerrit> Takashi Kajinami proposed openstack/devstack master: WIP: Allow overriding path for upper-constraints.txt https://review.opendev.org/730827 13:54:04 <tosky> Member->member, if I'm not mistaken 13:54:13 <kopecmartin> ahaa, so it's Member->_member_->member 13:54:48 <openstackgerrit> Takashi Kajinami proposed openstack/devstack master: WIP: Allow overriding path for upper-constraints.txt https://review.opendev.org/730827 13:54:54 <kopecmartin> thanks, this clears the things a little 13:55:06 <kopecmartin> one more bug i'd like to bring up 13:55:08 <gmann> tempest use 'member' 13:55:16 <kopecmartin> #link https://bugs.launchpad.net/tempest/+bug/1866082 13:55:16 <openstack> Launchpad bug 1866082 in tempest "Volume encryption tests handling on multibackend environments" [Low,Confirmed] 13:55:16 <gmann> not 'Memeber' right 13:55:57 <kopecmartin> tosky's looking for some suggestions how to proceed there, gmann any ideas? 13:56:40 <gmann> kopecmartin: https://review.opendev.org/#/q/topic:fix-member+(status:open+OR+status:merged) 13:56:42 <gmann> #link https://review.opendev.org/#/q/topic:fix-member+(status:open+OR+status:merged) 13:57:47 <kopecmartin> gmann: thanks 13:57:50 <kopecmartin> i see 13:58:03 <gmann> but you mentioned Tempest still use 'Member' ? 13:58:19 <kopecmartin> i think so, i'll check again 13:58:48 <gmann> kopecmartin: i think we do, we need to rename them also i think. 13:59:50 <kopecmartin> I'll check it and send a review 14:00:22 <gmann> kopecmartin: tosky so encryption is backend specific then feature flag is best way to go instead of dynamically detecting it from backends 14:01:05 <gmann> but is it core feature all backends should have or optional ? 14:01:17 <tosky> gmann: optional 14:01:34 <gmann> tosky: i think feature flag is right things for this. 14:01:36 <tosky> and yes, all the possible solutions I discussed are about feature flags 14:01:43 <tosky> just not sure which way is the best 14:02:22 <tosky> whether use the existing one, whether rely on the first value of it having encryption enabled, whether add another one, etc 14:03:34 <gmann> existing one you mean this ? - attach_encrypted_volume 14:03:43 <gmann> https://github.com/openstack/tempest/blob/ad25f2577b1118ab47db972bb11dd9ac29c34159/tempest/config.py#L552 14:03:47 <tosky> volume.storage_protocol 14:04:22 <tosky> the entire proposal is to handle the case when attach_encrypted_volume is enabled but some of the backends don't support it 14:04:22 <gmann> backend list and all are too complex and that might be stale after backends implement this feature 14:04:53 <gmann> but is it possible that we can attach the encrypted volume but cannot create ? 14:06:20 <tosky> it can't be handled at all if there is no support; I hit it for NFS 14:07:07 <gmann> i mean if attaching and creating the encrypted volume is all together backend support or not then we can use existing config option 14:07:28 <gmann> like we can rename 'attach_encrypted_volume' to 'encrypted_volume' and use it for attaching or creating 14:08:03 <gmann> if backend can have these two capabilities separately then we need new config 14:08:08 <tosky> anyway, that doesn't change the fact that 14:08:21 <tosky> - when multiple backends are defined, some may not work with feature X 14:08:51 <tosky> - a way to make sure that those tests are not executed on the backend that doesn't support it is needed 14:09:02 <tosky> hence the list of possible solutions on the ticket 14:09:52 <tosky> I think it may be a bit more time than what this meeting allows to go through the list 14:10:04 <gmann> but when multi backends are defined then Tempest do not know where the volume will be scheduled 14:11:15 <gmann> so for Tempest it is like fully supported for all of the backend your env have if any of backend does not support then it is disable and tempest skip the test 14:11:34 <gmann> having the scheduling knowledge in tempest is not possible right 14:12:30 <tosky> but we already have ways to drive the tempest behavior for certain features and also for specific volume backend 14:12:49 <gmann> but that is based on feature flag right 14:13:18 <tosky> that's the central assumption: none of the possible solutions I proposed in the ticket is based on something which is not a configuration variable 14:13:35 <tosky> there is no discovery, and I consider stuff like volume.storage_protocol as feature flag 14:14:21 <tosky> if you run tempest against a cloud which has multiple cinder backends and only some of them support encryption, right now you need to disable all the other backends 14:15:26 <gmann> 'disable all the other backends' you mean in Tempest ? 14:16:12 <gmann> anyways let's continue the discussion on this after office hour, in case anyone else has something to discuss 14:16:23 <gmann> any other critical bug or need disucssion 14:17:12 <gmann> #topic Critical Reviews 14:17:18 <gmann> any critical bug ? 14:18:26 <gmann> #topic Open Discussion 14:18:34 <gmann> anything else to disucss ? 14:18:37 <gmann> discuss 14:20:11 <gmann> ok let's close the office hour. 14:20:39 <gmann> we will cancel the next office hour due to PTG, i will announce it on ML also 14:20:44 <gmann> thanks everyone for joining 14:20:48 <gmann> #endmeeting