07:00:41 <ricolin> #startmeeting multi_arch
07:00:42 <openstack> Meeting started Tue Aug 18 07:00:41 2020 UTC and is due to finish in 60 minutes.  The chair is ricolin. Information about MeetBot at http://wiki.debian.org/MeetBot.
07:00:43 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
07:00:45 <openstack> The meeting name has been set to 'multi_arch'
07:00:49 <ricolin> #topic roll call
07:00:56 <ricolin> o/
07:03:26 <kevinz> o/
07:03:35 <ricolin> kevinz, hi
07:03:42 <kevinz> hi ricolin
07:05:29 <ricolin> kevinz, Thanks for join, I would like to discuss about devstack job with you and ianw
07:05:32 <ricolin> #link https://review.opendev.org/#/c/708317/
07:05:44 <ricolin> #topic CI followup
07:05:49 <kevinz> OK
07:06:50 <kevinz> that is a bit tricky one, and I reduce the concurrency to 2 ,but still has error occurs
07:06:57 <ricolin> I saw it passed now with certain failure
07:07:26 <kevinz> well sometimes it has  error with reboot_hard
07:07:58 <kevinz> but tempest team said that we should at least pass either reboot_soft or reboot_hard
07:08:06 <kevinz> should pass one case
07:08:23 <ricolin> what's that mean?
07:08:53 <ricolin> to pass either one
07:08:53 <kevinz> means that we should pass test_server_reboot_hard, or pass test_server_reboot_soft
07:09:15 <kevinz> we can not skip both of them
07:09:55 <ricolin> okay, you just remind me that our plan was to skip tests that keep failing
07:10:03 <kevinz> apart from this error, after reduce the tempest_concurrency can make thing better
07:10:04 <kevinz> yes
07:11:06 <ricolin> how long will it take if we running with no concurrency at all?
07:11:46 <kevinz> I don not tried, but concurrency=2 usually take 2.5h
07:13:02 <kevinz> Enlarge the Concurrency to higher we can see some OVS error, or Rabbitmq error, I believe this is due to low performance.
07:14:31 <kevinz> Now I skip those cases: https://review.opendev.org/#/c/724239/
07:16:10 <ricolin> kevinz, looks like we need some update on devstack patch and this one
07:16:16 <ricolin> #link  https://review.opendev.org/#/c/724239/
07:16:21 <kevinz> yes
07:16:28 <kevinz> I will update today
07:16:33 <ricolin> Thx
07:16:57 <kevinz> np
07:18:08 <kevinz> I'm working on some schduler limitation, to make sure that the Nodepool provision vm is rational...I saw somethings they are tending to launch a lot of VM to the same 1 node
07:18:32 <kevinz> And at that time the CI jobs could fail more cases...
07:19:14 <ricolin> that's a bug for sure
07:20:35 <ricolin> kevinz, is there a link to the issue?
07:21:06 <kevinz> not yet, I just "observed" that cases..
07:22:45 <ricolin> got it, just curious, because it's not happen when I using Nodepool. Would love to learn why that's happening
07:24:11 <kevinz> OK, I think we should adjust  the scheduler policy in Linaro Cluster
07:25:41 <ricolin> yeah, the test performance might getting better if we can make sure the schedule not result all in one node
07:26:37 <kevinz> Yes definitely
07:27:34 <ricolin> for another CI job
07:27:43 <ricolin> UT with py38
07:27:44 <ricolin> https://review.opendev.org/#/c/742090/
07:28:11 <ricolin> I guess it's reasonable to land it now
07:28:20 <ricolin> #link https://review.opendev.org/#/c/742090/
07:28:47 <ricolin> ianw, please help to review it when you got time
07:29:39 <ricolin> I guess the next task for that will be make sure we packaging with arm64 for services
07:30:46 <kevinz> cool ricolin
07:31:19 <ricolin> after that, the job should be reasonable to ask projects to run it as non-voting check job
07:31:45 <ricolin> Okay that's all I have for update
07:32:02 <ricolin> Thanks kevinz for the update on devstack job
07:32:19 <kevinz> np, thanks ricolin
07:32:44 <ricolin> let's end the meeting here, thanks for join:)
07:33:20 <ricolin> #endmeeting