| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2018-01-24 | |||
| 17:43:07 | mriedem | and once the docs get published | |
| 17:43:21 | melwitt | k | |
| 17:44:47 | mriedem | bauzas: what do you need from me wrt the libvirt gpu series? | |
| 17:47:40 | dansmith | mriedem: : | |
| 17:47:46 | openstackgerrit | Dan Smith proposed openstack/nova master: Avoid suspending guest with attached vGPUs https://review.openstack.org/535693 | |
| 17:47:55 | dansmith | mriedem: I just fixed up all the things we commented on | |
| 17:47:59 | mriedem | ack | |
| 17:48:01 | mriedem | reviewing | |
| 17:48:22 | dansmith | mriedem: I can try to find someone else to be the second +2 if my hands are too dirty.. I'll let you make that call | |
| 17:54:17 | openstackgerrit | Dan Smith proposed openstack/nova master: Avoid suspending guest with attached vGPUs https://review.openstack.org/535693 | |
| 17:54:19 | dansmith | mriedem: sorry, I forgot to replace one line in the test ^ | |
| 17:59:46 | dansmith | lbragstad: hey, mriedem told me you fixed the copious warnification about policy deprecativity recently | |
| 17:59:58 | dansmith | but I still experience said pain | |
| 18:00:22 | dansmith | is there something we're doing that blocks us from experiencing intended test euphoria? | |
| 18:01:09 | lbragstad | dansmith: oh - really? | |
| 18:01:20 | dansmith | lbragstad: yah | |
| 18:01:28 | lbragstad | i had a few patches in flight for a couple different issues, let me check to see where they are at | |
| 18:02:40 | openstackgerrit | OpenStack Release Bot proposed openstack/os-traits master: Update reno for stable/queens https://review.openstack.org/537512 | |
| 18:02:55 | openstackgerrit | OpenStack Release Bot proposed openstack/os-vif master: Update reno for stable/queens https://review.openstack.org/537514 | |
| 18:03:42 | lbragstad | dansmith: what version of oslo.policy are you experiencing this with? | |
| 18:03:59 | dansmith | lbragstad: I just tox -r'd, but let me look | |
| 18:04:12 | dansmith | oslo.policy>=1.30.0 # Apache-2.0 | |
| 18:04:12 | dansmith | (r/s/b/add-suppo)% grep policy requirements.txt | |
| 18:04:45 | mriedem | dansmith: +2 on https://review.openstack.org/#/c/535693/ - i think you're ok to +W | |
| 18:04:50 | lbragstad | https://review.openstack.org/#/c/531497/ is the patch that should have fixed the issue you're seeing | |
| 18:04:56 | dansmith | mriedem: rr thanks | |
| 18:04:59 | lbragstad | and that *should* be in oslo.policy 1.33.1 | |
| 18:05:50 | lbragstad | which looks good here - https://github.com/openstack/requirements/blob/master/upper-constraints.txt#L446 | |
| 18:06:08 | dansmith | lbragstad: hmm, but I've got 1.30 in requirements.. am I missing how g-r overrides that or something? | |
| 18:07:36 | lbragstad | that's a good question | |
| 18:07:59 | mriedem | dansmith: upper-constraints should pull in 1.33.1 | |
| 18:08:01 | lbragstad | dansmith: do you know exactly which version of oslo.policy you have in your env? | |
| 18:08:10 | melwitt | mriedem: can you remind me if you know if anything is unique to the cells v1 tempest job? in looking at the libvirt errors on stable, it's both stable branches and only the cells v1 job | |
| 18:08:11 | mriedem | he's got 1.30 | |
| 18:08:14 | dansmith | lbragstad: above I quoted what is in requirements.txt | |
| 18:08:33 | mriedem | melwitt: cellsv1 job shouldn't have anything unique about the compute / virt setup | |
| 18:08:35 | lbragstad | mriedem: i thought he had >=1.30 in requirements | |
| 18:08:42 | mriedem | oh, | |
| 18:08:45 | mriedem | pip freeze your tox venv | |
| 18:08:53 | dansmith | I hacked my requirements.txt, tox -r, and still get it | |
| 18:08:56 | mriedem | 1.33.1 is in u-c | |
| 18:09:15 | melwitt | mriedem: thanks | |
| 18:09:24 | dansmith | (r/s/b/add-suppo)% .tox/py27/bin/pip freeze | grep policy | |
| 18:09:25 | dansmith | oslo.policy==1.33.1 | |
| 18:09:26 | dansmith | after my hack | |
| 18:09:57 | lbragstad | ah | |
| 18:10:04 | mriedem | bauzas: is this the final piece of the vgpu puzzle for libvirt https://review.openstack.org/#/c/535693/ ? i don't see any other libvirt patches for vgpu - so assuming we're done once that merges | |
| 18:10:15 | dansmith | trying again without the hack, but assume I'll get that same thing again | |
| 18:10:30 | mriedem | do we have any patch that has a feature support matrix update for vgpu support? | |
| 18:11:56 | lbragstad | dansmith: do you have a policy file you're testing with locally, or are you just running nova tests? | |
| 18:12:11 | dansmith | lbragstad: just nova tests | |
| 18:12:18 | mriedem | dansmith: if it's always a handful of rules, it could be something in the policy fixture or a specific test | |
| 18:13:28 | lbragstad | the fix in 1.33.1 just makes it so a warning is logged iff a policy is deprecated *and* you're specifying it in a policy file or somewhere on disk | |
| 18:13:49 | dansmith | mriedem: there are 11 warnings, seemingly regardless of what tests I run | |
| 18:15:12 | mriedem | nova.tests.unit.policy_fixture.PolicyFixture maybe? | |
| 18:15:17 | mriedem | loads things up from some fake rules | |
| 18:15:33 | mriedem | from nova.tests.unit.fake_policy | |
| 18:15:44 | dansmith | mriedem: wouldn't that only affect me if I'm running those tests though? | |
| 18:15:49 | dansmith | I get this if I run a single-shot test | |
| 18:17:03 | lbragstad | so - it is writing them to disk https://github.com/openstack/nova/blob/master/nova/tests/unit/policy_fixture.py#L97 | |
| 18:17:32 | mriedem | dansmith: i think the base test case class loads up the policy fixture | |
| 18:17:33 | mriedem | for all tests | |
| 18:17:47 | lbragstad | which would trip this case - https://github.com/openstack/oslo.policy/blob/master/oslo_policy/policy.py#L610-L625 | |
| 18:17:59 | mriedem | yup self.policy = self.useFixture(policy_fixture.PolicyFixture()) | |
| 18:18:02 | mriedem | it's loaded for all tests | |
| 18:18:14 | mriedem | dansmith: so you'd have to try and remove the deprecated policy rules from the fake fixture data | |
| 18:18:43 | lbragstad | because oslo.policy thinks it needs to inform an operator about a policy override that is deprecated (i assume https://github.com/openstack/nova/blob/master/nova/tests/unit/fake_policy.py contains some deprecated policies) | |
| 18:19:15 | dansmith | mriedem: it doesn't seem to have a set of rules in the fixture though | |
| 18:20:07 | mriedem | dansmith: what's one of the ones it's complaining about? | |
| 18:20:22 | dansmith | os_compute_api:os-extended-volumes | |
| 18:20:26 | dansmith | maybe it's in fake_policy.py? | |
| 18:20:30 | mriedem | os_compute_api:os-extended-volumes | |
| 18:20:31 | mriedem | yes | |
| 18:20:33 | mriedem | that's the problem, | |
| 18:20:36 | lbragstad | https://github.com/openstack/nova/blob/master/nova/tests/unit/fake_policy.py#L50 | |
| 18:20:40 | lbragstad | ^ that one is deprecated | |
| 18:20:41 | mriedem | it's writing the fixture data to a temp file | |
| 18:20:46 | mriedem | and read in by the policy fixture | |
| 18:21:05 | dansmith | what does that mean though? there's no value set for that | |
| 18:21:11 | dansmith | or do you mean I just need to delete it? | |
| 18:21:16 | mriedem | "" means all i think | |
| 18:21:21 | mriedem | delete those entries | |
| 18:21:38 | lbragstad | i'm working on a patch now | |
| 18:21:41 | mriedem | i'm not sure why we actually even have that fake_policy file anymore | |
| 18:22:10 | mriedem | lbragstad has grown bored with keystone | |
| 18:24:15 | dansmith | lbragstad: I've got one already | |
| 18:24:18 | dansmith | about to push | |
| 18:24:41 | openstackgerrit | Dan Smith proposed openstack/nova master: Remove deprecated policy items from fake_policy https://review.openstack.org/537600 | |
| 18:24:45 | dansmith | lbragstad: mriedem ^ | |
| 18:25:25 | dansmith | hmm | |
| 18:25:26 | dansmith | I think I still get something | |
| 18:25:40 | dansmith | Its value may be silently ignored in the future | |
| 18:25:40 | dansmith | /dan/nova/.tox/py27/lib/python2.7/site-packages/oslo_policy/policy.py:623: UserWarning: Policy "os_compute_api:os-extended-volumes":"rule:admin_or_owner" was deprecated for removal in 17.0.0. Reason: Nova API extension concept has been removed in Pike. Those extensions have their own policies enforcement. As there is no extensions now, "os_compute_api:os-extended-volumes" policy which was added for extensions is not needed any more. | |
| 18:25:48 | dansmith | I think the rule:admin_or_owner bit is new now | |
| 18:26:43 | lbragstad | dansmith: oh nice | |
| 18:26:55 | dansmith | hmm, actually, no | |
| 18:27:02 | dansmith | I backed out the change and I get the same exact thing | |
| 18:28:22 | lbragstad | weird - that's still emitting https://github.com/openstack/oslo.policy/blob/master/oslo_policy/policy.py#L615-L625 | |
| 18:29:48 | lbragstad | huh... | |
| 18:29:51 | lbragstad | actually... | |