| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-01-23 | |||
| 16:47:44 | mriedem | it depends on context | |
| 16:47:46 | cdent | At least in the context of the scheduler related tests | |
| 16:47:52 | mriedem | jinx | |
| 16:47:56 | cdent | buy me a coke | |
| 16:48:23 | bauzas | mriedem: cdent: we could compare with other drivers | |
| 16:48:26 | cdent | Perhaps a UniqueHostnameFakeDriver subclass could help? | |
| 16:48:41 | mriedem | yeah that's not a bad idea | |
| 16:48:42 | bauzas | cdent: interesting idea | |
| 16:48:53 | bauzas | that would prevent us to fix the world | |
| 16:48:55 | mriedem | then just override the config with that in setUp for the tests that care | |
| 16:49:11 | bauzas | mriedem: cdent: I like that idea, I can spin a patch easily then | |
| 16:49:12 | mriedem | well, you could TODO it in ocata, | |
| 16:49:23 | mriedem | TODO(bauzas): this is temporary until we can fix all of the tests | |
| 16:49:39 | bauzas | mriedem: agreed | |
| 16:49:44 | cdent | \o/ | |
| 16:49:52 | bauzas | mriedem: I also need to verify the other drives | |
| 16:49:53 | cdent | we have just experienced successful collaboration | |
| 16:50:00 | gabor_antal | I'd like to notice, that we also have this hostname problem when we want to test live migration notification (e.g. https://review.openstack.org/#/c/402124/) | |
| 16:50:03 | cdent | achievement unlocked! | |
| 16:50:06 | bauzas | I'm pretty sure for libvirt that the nodename is equal to the service name | |
| 16:50:26 | bauzas | for ironic, it's easy : it's the ironic node uuid | |
| 16:50:39 | bauzas | but I leave the question open for other drivers | |
| 16:53:48 | openstackgerrit | Bob Ball proposed openstack/nova: XenAPI: fix memory usage miscalculation https://review.openstack.org/394857 | |
| 16:58:39 | andreykurilin | mriedem: it looks like "token" auth is compitely broken in novaclient. It doesn't work for CLI where keystoneauth creates session object itself(I do not know how many releases it was broken) and in python API (my patch broke it). I'll try to fix is asap and we need to cut new release | |
| 17:03:45 | mriedem | andreykurilin: https://review.openstack.org/#/c/419441/ right? | |
| 17:03:50 | openstackgerrit | STEW TY proposed openstack/nova: Transform instance.unrescue notifications https://review.openstack.org/388275 | |
| 17:04:23 | andreykurilin | mriedem: 1) it contains a bug. It fixes only python api | |
| 17:04:45 | andreykurilin | I'll re-submit it to pass gates | |
| 17:06:59 | openstackgerrit | Andrey Kurilin proposed openstack/python-novaclient: Pass relevant parameters to Token based authentication https://review.openstack.org/419441 | |
| 17:07:20 | openstackgerrit | Dan Smith proposed openstack/nova: Only warn about hostmappings during ocata upgrade https://review.openstack.org/424243 | |
| 17:08:28 | dims | mriedem : notice the gate-tempest-dsvm-py35-ubuntu-xenial-nv in check queue? (example https://review.openstack.org/#/c/323250/) | |
| 17:09:51 | mdbooth | bauzas: When doing a migration with an explicit destination, would it make sense to do make a scheduler call anyway restricted to just that destination to see if it succeeds? | |
| 17:10:05 | mdbooth | For all I know we already do this somewhere, just thinking... | |
| 17:10:16 | bauzas | mdbooth: for a live migration, we do that already | |
| 17:10:32 | bauzas | mdbooth: for cold ones, there is a BP from takashi | |
| 17:10:32 | mdbooth | I'm looking here: https://review.openstack.org/#/c/244489/58/nova/conductor/tasks/live_migrate.py | |
| 17:10:54 | mdbooth | explicit destination doesn't call _find_destination() there | |
| 17:11:13 | mdbooth | But perhaps it's called elsewhere. | |
| 17:11:38 | mdbooth | Well, not _find_destination, but some other scheduler call. | |
| 17:11:56 | mriedem | dims: notice what? | |
| 17:12:06 | mriedem | dims: if it's non-voting i'm not noticing it right now | |
| 17:14:42 | bauzas | mriedem: https://github.com/openstack/nova/blob/c9eb9530314d047f5013941ebcfd5ef0192a9dc3/nova/compute/api.py#L3773 | |
| 17:16:08 | dims | mriedem : ack. it's green that's all :) | |
| 17:16:35 | mriedem | dims: oh you were looking for some back patting, i see | |
| 17:16:51 | mriedem | #success dims got gate-tempest-dsvm-py35-ubuntu-xenial-nv green | |
| 17:16:53 | openstackstatus | mriedem: Added success to Success page | |
| 17:17:04 | dims | LOL, just wanted to be sure i won't get yelled at if you notice it later in the week (when it goes RED) :) | |
| 17:18:09 | mriedem | i won't notice | |
| 17:18:35 | mriedem | dims: i'm on vacation so only water temperature and some other key things are high priority for me right now | |
| 17:18:39 | dims | fair enough | |
| 17:19:11 | mriedem | i don't have to officially care about py35 until march | |
| 17:19:12 | mriedem | :) | |
| 17:19:22 | dims | ack mriedem : have fun! | |
| 17:26:34 | openstackgerrit | Dan Smith proposed openstack/nova: WIP Move instance creation to conductor https://review.openstack.org/319379 | |
| 17:27:50 | jaosorior | gibi: you around? | |
| 17:28:07 | openstackgerrit | Stephen Finucane proposed openstack/nova: scheduler: Don't modify RequestSpec.numa_topology https://review.openstack.org/419479 | |
| 17:30:53 | openstackgerrit | Andrey Kurilin proposed openstack/python-novaclient: [CLI] Fix token auth type https://review.openstack.org/424256 | |
| 17:31:51 | andreykurilin | mriedem: have a nice vacation! I'll try to ping some other cores to merge fixes when it will be ready and propose a release | |
| 17:33:13 | mriedem | ack | |
| 17:33:21 | mriedem | andreykurilin: thanks for being on top of it | |
| 17:33:41 | sfinucan | dansmith: Added tests to https://review.openstack.org/419479. Looking at the cells limit patch again now | |
| 17:45:07 | openstackgerrit | Stephen Finucane proposed openstack/nova: objects: add new field cpuset_reserved in NUMACell https://review.openstack.org/400874 | |
| 17:45:15 | mdbooth | Can anybody tell me what the 'limits' in the host object returned by scheduler_client.select_destinations() means? | |
| 17:45:19 | mdbooth | bauzas: ^^^ ? | |
| 17:45:52 | bauzas | mdbooth: it's what the claim uses for verifying the limit | |
| 17:46:06 | mdbooth | What's its scope? | |
| 17:46:23 | bauzas | mdbooth: for example, see the CoreFilter, it's telling what the max amount of the resouce is given the allocation ratio | |
| 17:46:31 | mdbooth | i.e. can it race with the clam? | |
| 17:47:05 | mdbooth | Ok, so it's essentially a constant? It doesn't vary as resources are used. | |
| 17:48:28 | mdbooth | I'm looking at a patch where conductor fetches this limits value and passes it to compute, which then uses it to create a claim | |
| 17:49:03 | mdbooth | IIUC, what you describe means there's no race there, as the value can't change in the presence of other placements. | |
| 17:50:09 | mdbooth | I assumed that was the case because in claim: free = limit - used | |
| 17:50:23 | mdbooth | where used is dynamic | |
| 17:50:32 | dansmith | sfinucan: re-reviewed, thanks | |
| 17:52:34 | andreykurilin | dims: I do not what had happened, but rally-nova-py3 jobs is ok now:) | |
| 17:53:02 | mlavalle | mriedem: is it still the plan in Ocata to have the scheduler use the data in resource providers and inventories for instances placement in host? | |
| 17:53:03 | andreykurilin | *I do not know | |
| 17:54:16 | edleafe | mlavalle: yes. The scheduler will get a pared-down list of hosts from placement that match the basic resources in the request | |
| 17:54:33 | edleafe | mlavalle: it will then use its other filters to select, as it currently does | |
| 17:55:47 | mriedem | mlavalle: yes | |
| 17:56:16 | openstackgerrit | John Garbutt proposed openstack/nova: WIP Support cinder v3 API https://review.openstack.org/420201 | |
| 17:58:14 | mlavalle | edleafe, mriedem: so, given that the patches that I wrote for resource providers and routed networks are merging, i can remove the first bullet here, right: http://docs.openstack.org/newton/networking-guide/config-routed-networks.html#limitations? | |
| 17:58:56 | edleafe | mlavalle: no | |
| 17:59:18 | edleafe | mlavalle: I guess I should have specified what I meant by "basic" :) | |
| 17:59:32 | edleafe | mlavalle: just disk, RAM, and VCPU for Ocata | |
| 18:00:19 | mriedem | mlavalle: what ed said. this is also going to be a pre-req for scheduling with routed networks https://review.openstack.org/#/q/status:open+project:openstack/nova+branch:master+topic:bp/prep-for-network-aware-scheduling-ocata | |
| 18:00:34 | mriedem | mlavalle: and it's looking like at this point that series won't be happening in ocata | |
| 18:01:32 | mlavalle | edleafe, mriedem: so IPV4_ADDRESS inventories are postponed to Pike? | |
| 18:02:23 | mlavalle | as far as being used by the scheduler | |
| 18:02:24 | edleafe | mlavalle: They can be reported and tracked in Ocata. The scheduler won't be able to do anything with them, though | |
| 18:02:52 | mlavalle | edleafe, mriedem: thanks for the update! | |
| 18:03:11 | edleafe | mlavalle: IOW, the placement service will be able to handle it. The scheduler won't be using that information, yet' | |
| 18:03:14 | mriedem | mlavalle: yeah you definitely want them reporting in as soon as possible | |
| 18:03:44 | mriedem | so that when the scheduler can handle it, the data is already there | |
| 18:04:18 | dims | andreykurilin : we must have fixed some stuff in nova while we got the dsvm+tempest working :) | |
| 18:04:29 | dims | andreykurilin : thanks for the good news :) | |
| 18:04:35 | mlavalle | mriedem, edleafe: ok, on our side the functionality is complete. the last patchset remaining has already 2 +2's, so it should merge today. I even update the documentation | |
| 18:05:01 | andreykurilin | dims: there is only one failed scenario. will look at it more close a bit later | |
| 18:05:17 | dims | ack andreykurilin : thanks | |
| 18:05:18 | mlavalle | mriedem, edleafe: I will focus next on creating a tempest test to exercise it in the gate, so we all get feedback | |