| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2020-07-30 | |||
| 15:28:24 | sean-k-mooney | maybe it kind of feells like soemthing that should not need to be manual | |
| 15:28:49 | stephenfin | I agree, but you could say the same about the NUMA-aware vSwitches feature in general | |
| 15:29:00 | sean-k-mooney | we did | |
| 15:29:04 | stephenfin | True :) | |
| 15:29:10 | sean-k-mooney | be we also said we did not want to wait | |
| 15:29:13 | stephenfin | but it was a means to an end | |
| 15:29:14 | stephenfin | yup | |
| 15:29:36 | stephenfin | I don't really want to pile more on top of that shaky foundation though | |
| 15:29:48 | stephenfin | like I said, we're stuck with traits once they're there | |
| 15:29:52 | sean-k-mooney | ralonsoh: stephenfin ok so do we want to put the traits on pause untill we see if we will use them in the nova half of the feature | |
| 15:30:08 | ralonsoh | sean-k-mooney, np | |
| 15:30:14 | sean-k-mooney | ralonsoh: what that means for you is you just dont need to add a trait request to the port | |
| 15:30:15 | stephenfin | if that works for ralonsoh, I'd much rather that | |
| 15:30:35 | sean-k-mooney | and if nova want to use the traits in the futrue we can do it instead of neutorn | |
| 15:30:38 | ralonsoh | sean-k-mooney, but what should I add to the port info? | |
| 15:30:39 | stephenfin | if it turns out we want a way to advertise these capabilities, then I'm all for revisiting | |
| 15:30:52 | ralonsoh | because I don't know how nova is going to read it | |
| 15:31:02 | sean-k-mooney | ralonsoh: neutron need to add the polciy via the new extention | |
| 15:31:09 | sean-k-mooney | but it does not need to add the trait | |
| 15:31:26 | sean-k-mooney | the extenion was adding a new extended atibute to the port anyway right | |
| 15:31:26 | ralonsoh | sean-k-mooney, that means another parameter in the port dict | |
| 15:31:27 | ralonsoh | ok | |
| 15:31:41 | ralonsoh | I'll need to revisit the spec | |
| 15:31:42 | sean-k-mooney | ralonsoh: yes isnt that what the spec says | |
| 15:32:36 | stephenfin | ralonsoh: again, apologies for only getting to this now. I didn't know that this work was as far along as it is. You've been busy :) | |
| 15:32:39 | sean-k-mooney | https://specs.openstack.org/openstack/neutron-specs/specs/victoria/port-numa-affinity-policy.html#rest-api-impact | |
| 15:32:48 | ralonsoh | np at all | |
| 15:32:55 | ralonsoh | I'll revisit the spec and the patches | |
| 15:32:59 | sean-k-mooney | oh its in the resouce attibute | |
| 15:33:19 | sean-k-mooney | sorry no that is just an extend port attibute | |
| 15:33:48 | sean-k-mooney | so im not sure you need to updat ehte spec other then to remove the port_resource['resource_request'] = { | |
| 15:33:51 | sean-k-mooney | 'required': [os_traits.COMPUTE_NUMA_POLICY_REQUIRED]} | |
| 15:33:53 | sean-k-mooney | bit | |
| 15:34:14 | ralonsoh | sean-k-mooney, I need to clean up this part yes | |
| 15:34:28 | sean-k-mooney | nova will just read numa_affinity_policy field form the port and use that | |
| 15:34:40 | ralonsoh | perfect for me | |
| 15:34:51 | sean-k-mooney | ralonsoh: i was worried that you were only going to pass the trait | |
| 15:35:10 | sean-k-mooney | ralonsoh: that would not have been right from a nova point of view | |
| 15:48:52 | sean-k-mooney | stephenfin: by the way you also missed the mix hypervior usecase for the trait | |
| 15:49:24 | sean-k-mooney | but we can cross that bridge if we need too | |
| 16:01:05 | openstackgerrit | Pavlo Shchelokovskyy proposed openstack/nova master: Raise 409 when removing security group from instance https://review.opendev.org/735572 | |
| 16:06:58 | openstackgerrit | sean mooney proposed openstack/nova master: Lookup nic feature by PCI address https://review.opendev.org/739131 | |
| 16:07:53 | sean-k-mooney | ralonsoh: stephenfin by the way should i drop the funcitonal tests form the coverage job or make a coverage with sudo job for os-vif https://review.opendev.org/#/c/728037/4/.zuul.yaml | |
| 16:08:10 | stephenfin | drop from coverage | |
| 16:08:14 | stephenfin | imo | |
| 16:08:26 | sean-k-mooney | cool i can do that trivially | |
| 16:08:37 | ralonsoh | sean-k-mooney, why do you need to drop them? | |
| 16:08:50 | sean-k-mooney | the openstack-cover-jobs | |
| 16:08:55 | sean-k-mooney | does not allow sudo to be used | |
| 16:09:00 | ralonsoh | right | |
| 16:09:06 | sean-k-mooney | so privsep fails | |
| 16:09:33 | sean-k-mooney | our functional tests actully need elevated right to run | |
| 16:09:37 | sean-k-mooney | becaue they acutlly use ovs | |
| 16:09:50 | sean-k-mooney | which actully is alos not in stalled in that job | |
| 16:10:25 | sean-k-mooney | so ya if i want to run them i need a custom job that has sudo and installs ovs like our custom functional job | |
| 16:10:44 | ralonsoh | sean-k-mooney, you can take the neutron functional base one | |
| 16:10:48 | ralonsoh | if you want | |
| 16:11:16 | sean-k-mooney | no we have a base job for fucntional test | |
| 16:11:29 | sean-k-mooney | openstack-tox-functional-ovs-with-sudo | |
| 16:11:39 | ralonsoh | or this one yes | |
| 16:12:05 | sean-k-mooney | but i either need to inherit form that and chagne the env form functional to cover | |
| 16:12:16 | sean-k-mooney | or i drop the fucntional test | |
| 16:12:38 | sean-k-mooney | i was leaning to removing them so that you did not need sudo and ovs to run tox -e cover | |
| 16:13:02 | sean-k-mooney | we dont actully have covage report for os-vif currently so this is adding them | |
| 16:13:48 | sean-k-mooney | jangutter: any opipion ^ | |
| 16:14:54 | sean-k-mooney | jangutter: question is shoudl os-vifs coverage tox enve run just unit test or unit and functional | |
| 16:17:14 | jangutter | sean-k-mooney: :reading: | |
| 16:18:27 | jangutter | sean-k-mooney: can we achieve full coverage with just unit tests? I'm almost leaning yeah. | |
| 16:18:49 | sean-k-mooney | we could | |
| 16:18:56 | sean-k-mooney | although that is not the goal | |
| 16:19:19 | sean-k-mooney | but i do want to more or less know what is and is not tested | |
| 16:19:49 | sean-k-mooney | 100% coverage is not need but if we are low in one module it would be good to look at why | |
| 16:22:15 | jangutter | I agree - full coverage might might just be for bragging rights (and can be done badly). My feeling is that unit testing should be slightly more than functional, but if we don't measure it we won't know it. | |
| 16:23:06 | sean-k-mooney | so are you ok with the coverage job just doing unit. i could add a second coverage-all that ran both | |
| 16:23:33 | sean-k-mooney | i would like to enable coverage by default when you run just tox | |
| 16:23:49 | sean-k-mooney | but if its runing the functional test i would prefer not to | |
| 16:24:08 | sean-k-mooney | basically i want you to be able to jsut git clone and then run "tox" | |
| 16:24:13 | sean-k-mooney | without args and have it work | |
| 16:24:27 | jangutter | Same for me: I tend to run just unit when I dev, but rely on the CI to run functional. | |
| 16:24:42 | sean-k-mooney | i run the fucnational test but not on my laptop | |
| 16:24:46 | sean-k-mooney | i do it in my devstack vms | |
| 16:25:00 | sean-k-mooney | so for small change i dont want to have to instlal ovs | |
| 16:25:58 | openstackgerrit | Sylvain Bauza proposed openstack/nova-specs master: Offline Reshape tool spec https://review.opendev.org/742908 | |
| 16:26:08 | jangutter | I'm OK with adding more non-sudo stuff to the default tox run. | |
| 16:27:22 | sean-k-mooney | in os-vif the test are really fast so i was going to enable py3,pep8,docs,releasenotes,cover | |
| 16:27:38 | sean-k-mooney | by default | |
| 16:28:43 | sean-k-mooney | basically that is what i would like peple to run before pushing | |
| 16:29:01 | jangutter | I'm in agreement. | |
| 16:30:03 | sean-k-mooney | ok ill remvoe the functional test from cover so for now at least | |
| 16:30:41 | sean-k-mooney | stephenfin: jangutter ralonsoh as a head up i also want to change the default ovsdb backend in os-vif this cycle | |
| 16:30:54 | sean-k-mooney | i ment to do it af ew releases ago | |
| 16:31:02 | sean-k-mooney | but we are still defaulting to ovs-vsctl | |
| 16:31:18 | ralonsoh | sean-k-mooney, in os-vif? | |
| 16:31:32 | sean-k-mooney | ralonsoh: ya | |
| 16:31:36 | ralonsoh | we removed ovs-vsctl, we only have native now | |
| 16:31:51 | sean-k-mooney | no | |
| 16:31:56 | ralonsoh | in neutron | |
| 16:32:03 | sean-k-mooney | in neutron yes | |
| 16:32:08 | ralonsoh | just to let you know | |
| 16:32:13 | sean-k-mooney | we never got around to doing it in os-vif | |