Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2020-04-08
09:30:59 bauzas stephenfin: I'd appreciate some reading of https://review.opendev.org/#/c/715489/
09:31:06 bauzas (btw.)
09:31:33 bauzas bonus stage : https://review.opendev.org/#/c/717975/8
09:32:24 lyarwood bauzas: I've got that open at the moment btw, taking a while as I've never looked at vGPU stuff before.
09:32:52 bauzas lyarwood: that's the reason why I invested a bit of time on functionally testing the feature https://review.opendev.org/#/c/717975/
09:33:31 bauzas worth reading the last bit, fresh as of this night.
09:34:13 lyarwood ack thanks
09:37:31 brinzhang gibi: seem like you missed the "--os-compute-api-version 2.85" in you CLI
09:37:45 gibi brinzhang: hm, interesting
09:37:47 gibi checking...
09:38:02 gibi nova client should default to max microversion
09:38:10 gibi and False to True worked
09:38:15 gibi but let me double check it
09:38:52 brinzhang Emm..interesting..
09:40:41 gibi brinzhang: here is a repro http://paste.openstack.org/show/791792/
09:41:05 gibi False -> True works, True -> False seems to be ignored
09:42:13 gibi let's try to attach a debugger
09:42:27 brinzhang gibi: looks like the phenomenon is not in nocalient
09:42:51 gibi you mean, this a potential bug in the nova API change?
09:43:23 brinzhang gibi: I am not sure, I will rebuild my env, I think it's not fast
09:43:43 gibi OK, I'm also looking into this in parallel with you
09:43:51 gibi I will let you know if I found something
09:44:58 brinzhang "False -> True works, True -> False seems to be ignored", that from False to True works, and I reviewed again in novalient code, it's ok for me, so I am not sure whether is it have something in nova API
09:47:26 openstackgerrit Mikhail Ushanov proposed openstack/nova stable/ocata: Support qemu >= 2.10 https://review.opendev.org/693851
09:53:14 gibi brinzhang: it seems to me that converting from the 'False' string to boolean is missing from the API code
09:53:17 gibi (Pdb) bdm.delete_on_termination = att['delete_on_termination']
09:53:19 gibi (Pdb) bdm.delete_on_termination
09:53:22 openstackgerrit Andreas Jaeger proposed openstack/nova-specs master: Cleanup py27 support https://review.opendev.org/718368
09:53:22 gibi True
09:53:24 gibi (Pdb) att['delete_on_termination']
09:53:27 gibi 'False'
09:53:29 gibi (Pdb)
09:53:42 gibi and ovo converst the non empty string ('False') to True automatically
09:55:09 gibi the schema enforces the boolean-ness of the field but we did not covert it to boolean
09:59:49 gibi yep, I confirmed that the string boolean conversion is missing
09:59:59 gibi filing bug...
10:00:57 brinzhang_ gibi: yeah, happened in https://review.opendev.org/#/c/693828/27/nova/api/openstack/compute/volumes.py@454
10:01:54 gibi yep, a bool_from_string calls is missing there, and also we don't have proper test coverage as we did not catch the bug
10:02:17 gibi I'm filling a bug, do you have time to start creating a fix?
10:02:49 brinzhang_ gibi: of course
10:03:02 gibi thanks, I wil link the bug here soon
10:03:14 brinzhang_ gibi: thanks
10:08:28 gibi brinzhang_: https://bugs.launchpad.net/nova/+bug/1871598
10:08:29 openstack Launchpad bug 1871598 in OpenStack Compute (nova) "PUT /servers/{server_id}/os-volume_attachments/{volume_id} does not update delete_on_termination from True to False" [Medium,Triaged]
10:08:58 brinzhang_ gibi: I will fix it today, before I get home
10:09:02 gibi brinzhang_: thanks
10:09:11 gibi ping me when you push the fix and I will review it
10:09:17 brinzhang_ gibi: np
10:20:23 openstackgerrit Stephen Finucane proposed openstack/nova master: Introduce 'pcpuset' field for InstanceNUMACell https://review.opendev.org/714658
10:20:23 openstackgerrit Stephen Finucane proposed openstack/nova master: hardware: Remove '_numa_fit_instance_cell_with_pinning' https://review.opendev.org/714703
10:20:24 openstackgerrit Stephen Finucane proposed openstack/nova master: hardware: Add support for 'hw:cpu_dedicated_mask' extra spec https://review.opendev.org/714706
10:20:24 openstackgerrit Stephen Finucane proposed openstack/nova master: Introduce the 'CPUAllocationPolicy.MIXED' enum https://review.opendev.org/716267
10:20:25 openstackgerrit Stephen Finucane proposed openstack/nova master: Calculate the pinned and unpinned CPU usage for mixed instance https://review.opendev.org/716269
10:20:25 openstackgerrit Stephen Finucane proposed openstack/nova master: computer: validate cpu pinning configuration for mixed instance https://review.opendev.org/716268
10:20:26 openstackgerrit Stephen Finucane proposed openstack/nova master: libvirt/hardware: refactor numa_get_constraints https://review.opendev.org/718294
10:20:26 openstackgerrit Stephen Finucane proposed openstack/nova master: mark mixed policy as ready, bump compute service version https://review.opendev.org/716271
10:20:27 openstackgerrit Stephen Finucane proposed openstack/nova master: metadata: export the vCPU IDs that are pinning on the host CPUs https://review.opendev.org/688936
10:20:27 openstackgerrit Stephen Finucane proposed openstack/nova master: Setup 'mixed' instance through 'PCPU' and 'VCPU' resource https://review.opendev.org/713355
10:27:08 gibi stephenfin: one question / nit in https://review.opendev.org/#/c/717789/2/nova/api/validation/extra_specs/validators.py@50
10:32:11 stephenfin gibi: Replied. I'll have a test shortly
10:33:25 alex_xu gibi: quick question, I don't think we need obj_make_compatible method for notification object, right? https://review.opendev.org/#/c/714658/6/nova/notifications/objects/request_spec.py@187
10:36:52 openstackgerrit Stephen Finucane proposed openstack/nova master: Follow-up for flavor-extra-spec-validators series https://review.opendev.org/718357
10:37:54 gibi stephenfin: thanks. +A
10:38:04 stephenfin ta
10:38:49 gibi alex_xu: at the momen there is no way to ask nova to emit older than the latest version of the notifications. So obj_make_compatible is unused for notification ovos
10:39:00 gibi alex_xu: so no, we don't need it
10:40:39 alex_xu gibi: thanks
10:54:06 openstackgerrit Brin Zhang proposed openstack/nova master: Convert delete_on_termination from string to boolean https://review.opendev.org/718387
10:54:21 brinzhang gibi: fixed done, pls review https://review.opendev.org/718387
10:56:09 openstackgerrit Stephen Finucane proposed openstack/nova master: Follow-up for flavor-extra-spec-validators series https://review.opendev.org/718357
10:56:16 stephenfin bauzas: had to respin that ^
10:56:27 bauzas ack
10:57:25 gibi brinzhang: ack, will check, thanks
11:02:38 openstackgerrit Merged openstack/nova master: Add new default roles in security group policies https://review.opendev.org/716793
11:11:53 openstackgerrit Merged openstack/nova master: Allow versioned discovery unauthenticated https://review.opendev.org/685181
11:12:01 openstackgerrit Merged openstack/nova master: Fix new context comparison workaround in base tests class https://review.opendev.org/717825
11:20:39 openstackgerrit Brin Zhang proposed openstack/python-novaclient master: Microversion 2.85: Change volume-update CLI https://review.opendev.org/712651
11:22:06 lyarwood https://review.opendev.org/#/q/topic:workaround_native_luksv1+status:open+project:openstack/nova - looking for core reviews on this btw, appreciate everyone is busy ahead of M3 but wanted to ask anyway.
11:22:48 brinzhang_ gibi, lyarwood: novalient for microversion 2.85 updated too.
11:23:11 gibi brinzhang_: ack
11:34:26 gibi stephenfin: feedback in https://review.opendev.org/#/c/708436/ would you like to fix it in a FUP?
11:34:57 lyarwood brinzhang_: ack thanks, I think the call to bool_from_string should use strict=True
11:37:25 brinzhang_ lyarwood: I looked you comment, agree add strict=True
11:45:16 brinzhang_ lyarwood: [True, 'True', 'TRUE', 'true', '1', 'ON', 'On', 'on', 'YES', 'Yes', 'yes'] for true, [False, 'False', 'FALSE', 'false', '0', 'OFF', 'Off', 'off', 'NO', 'No', 'no'] for False, does them all tests?
11:45:33 brinzhang_ if yes, I will do a for cycle to test
11:45:55 sean-k-mooney what is this for?
11:46:59 sean-k-mooney if your trying to do string to bool conversion use https://github.com/openstack/oslo.utils/blob/master/oslo_utils/strutils.py#L123-L158
11:47:15 sean-k-mooney if you use bool_from_string you dont need to test all the combinations
11:47:20 sean-k-mooney we know the oslo function works
11:47:23 lyarwood brinzhang_: I don't think we need to test them all but just more than False
11:48:07 lyarwood brinzhang_: and a bad value obviously
11:49:16 openstackgerrit Merged openstack/nova master: Add test coverage of existing simple tenant usage policies https://review.opendev.org/717546
11:49:28 openstackgerrit Merged openstack/nova master: Add test coverage of existing server migrations policies https://review.opendev.org/717531
11:49:45 openstackgerrit Merged openstack/nova master: Fix server tags policy to be admin_or_owner https://review.opendev.org/717947
11:49:54 openstackgerrit Merged openstack/nova master: Add test coverage of existing server tags policies https://review.opendev.org/717425
11:50:04 brinzhang ok, I will add a true value e.g. '1' to tests
11:50:05 openstackgerrit Merged openstack/nova master: Introduce scope_types in server tags policy https://review.opendev.org/717948
11:50:14 openstackgerrit Merged openstack/nova master: Add new default roles in server tags policies https://review.opendev.org/717954
11:50:22 openstackgerrit Merged openstack/nova master: Introduce scope_types in server migration https://review.opendev.org/717588
11:50:32 openstackgerrit Merged openstack/nova master: Add new default roles in server migration policies https://review.opendev.org/717590

Earlier   Later