Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2017-01-23
16:24:56 bauzas cdent: I'm not saying it's due to the UC, just that the problem is popping up now because all our computes have 'fake-mini'
16:25:01 BobBall jroll: OK - my reading is that I think that setting the default per-compute will work as the scheduler reads from the host_state rather than using a default value from the config file.
16:25:22 bauzas cdent: I fixed that in https://review.openstack.org/#/c/424110/ but that blows up, probably because it makes a shit ton of assumptions
16:25:31 jroll BobBall: ah, then yeah, conf.set_default in the driver's initialization would probably work out
16:25:37 jroll I guess it's easily tested
16:25:43 BobBall I'm already hacking it up to test :)
16:25:56 cdent bauzas: your use of the phrase "very important" made it sound like it would be hard to fix and might impact actually using the system, but it sounds more like it is something we need to address (and may be hard to address) but once addressed, not a problem?
16:26:18 bauzas cdent: it's a show-stopper for merging the scheduler upcall
16:26:25 bauzas tbc
16:26:27 edleafe bauzas: thinking about real-world deployments: are we requiring operators to give every compute node a unique name?
16:27:07 bauzas edleafe: well, I think we made this hypothesis a while ago because you can get_by_name a RP
16:27:20 bauzas now, it's part of the world
16:27:33 bauzas and we can't easily go against that way
16:28:03 edleafe bauzas: ok, so how likely is it that a deployment with thousands of compute nodes has named them all uniquely?
16:28:42 bauzas edleafe: well, that's certainly a good point, but you're advocating of removing that UC, then ?
16:28:48 bauzas sounds a big deal to me
16:29:01 edleafe bauzas: not advocating - more like thinking aloud
16:29:23 edleafe bauzas: as someone who hasn't ever managed a large deployment
16:29:47 bauzas edleafe: AFAIR, libvirt is taking the hostname as the hypervisor_hostname
16:29:57 bauzas edleafe: and that goes the same for other drivers
16:30:05 edleafe bauzas: causing tests to fail is bad. Causing deployments to fail is... worse
16:30:24 bauzas edleafe: given it's based on the service hostname, that looks not difficult to me to have different hostnames :)
16:31:31 sfinucan DinaBelova: Quick question - if I run something like 'tox -e py27 nova.tests.unit.scheduler.filters.test_numa_topology_filters', is it expected that the os_profiler test be run?
16:32:00 openstackgerrit Eric Fried proposed openstack/nova: Introduce base PowerVM driver (disk drv) https://review.openstack.org/409401
16:32:11 edleafe bauzas: ok, that's reassuring
16:33:07 cdent bauzas: so sounds like the first thing to do is fix the fixtures and the functional tests? Do you need/want help with that?
16:33:40 bauzas cdent: http://logs.openstack.org/10/424110/1/check/gate-nova-tox-db-functional-ubuntu-xenial/e2972f1/testr_results.html.gz
16:33:48 mriedem so rp.name is optional
16:33:57 bauzas cdent: giving you a sense of what I call "a shit ton of assumptions"
16:33:57 mriedem is it the resource tracker that's setting it to hypervisor_hostname?
16:34:04 bauzas mriedem: it is indeed
16:34:20 mriedem so how about we name it hypervisor_hostname + nodename?
16:34:25 mriedem surely those would be unique in the real world
16:34:47 mriedem for our fixtures and tests, we probably have those copied all over the place, but we could just fix that
16:34:50 mriedem with uuids or something
16:34:50 bauzas mriedem: hypervisor_hostname == nodename but I get your idea
16:35:06 bauzas mriedem: you mean service hostname + hypervisor_hostname
16:35:24 bauzas lemme try this
16:35:29 openstackgerrit Dan Smith proposed openstack/nova: Multicell support for instance listing https://review.openstack.org/396775
16:35:29 openstackgerrit Dan Smith proposed openstack/nova: Make api_samples tests use simple cell environment https://review.openstack.org/410473
16:35:30 openstackgerrit Dan Smith proposed openstack/nova: WIP Move instance creation to conductor https://review.openstack.org/319379
16:35:33 dansmith sfinucan: thanks for catching that, fixed here ^
16:36:11 BobBall jroll: What do you know - it looks like it works! Many thanks
16:36:13 openstackgerrit Claudiu Belu proposed openstack/nova: hyper-v: Adds support for VMs with nested virtualization https://review.openstack.org/398509
16:36:14 openstackgerrit Claudiu Belu proposed openstack/nova: Adds CPUFeaturesField to ImageMetaProps https://review.openstack.org/420960
16:36:30 cdent mriedem, bauzas: wait, are we saying that hypervisor hostname is not unique? is it not an FQDN?
16:36:51 bauzas cdent: not with the FakeDriver at least
16:37:00 cdent bauzas: the fake driver is _fake_
16:37:04 cdent and can be changed
16:37:09 cdent so isn't really a true issue
16:37:13 bauzas cdent: sure, what I did
16:37:21 cdent I'm trying to figure out if there is a real issue
16:37:25 bauzas cdent: I changed that to being defaulted to the service hostname
16:37:32 bauzas cdent: but kaboom
16:37:46 cdent hypervisor hostname is unique, job done
16:37:53 bauzas because in lots of places, we litterally say hypervisor_hostname == 'fake-mini'
16:37:55 cdent that's because in a test context service hostname is not unique, right?
16:38:05 jroll BobBall: \o/
16:38:13 cdent so we just fix that, right?
16:38:56 bauzas cdent: that's because we stupidely get the value from the first CONF.host value which is set to 'fake-mini' (because of the ConfFixture)
16:39:29 bauzas so we wrote around 500 tests saying "cool, let's just assume it's fake-mini whatever the compute node name is"
16:39:56 bauzas honestly, mriedem's proposal seems the less boring approach
16:40:41 cdent that means changing the names of existing resource providers out there in a deployed world?
16:40:42 mriedem bauzas: i also agree with cdent,
16:40:54 mriedem in that if the hostname is just fqdn, and we have a bunch of dumb tests, we just fix the tests that need fixing,
16:41:02 mriedem i doubt it's 500 that we need to fix 'right now'
16:41:28 bauzas mriedem: sure, we can do that, I leave others fix that then
16:41:42 bauzas https://review.openstack.org/#/c/424110/1 is just red tomato
16:42:24 cdent bauzas: If it helps, I can do that.
16:42:37 cdent As it has to be done, right?
16:43:02 bauzas cdent: if we agree on that being a necessary prereq for merging the scheduler change, I'm fine
16:43:07 openstackgerrit Bob Ball proposed openstack/nova: XenAPI: fix memory usage miscalculation https://review.openstack.org/394857
16:43:25 bauzas cdent: yup, lemme show you why it's important
16:43:28 cdent bauzas, fine with me doing it, or fine you'll take care of it?
16:43:34 cdent I know why it is important
16:44:09 bauzas cdent: http://paste.openstack.org/show/595821/ L62
16:45:42 mriedem bauzas: so let's just focus on fixing the tests that need fixing right now to move you along
16:45:56 bauzas okay
16:47:06 cdent bauzas: If there's some task or chunk of a task that you can split off to me, let me know.
16:47:11 bauzas mriedem: cdent: so we all agree with the fact that the FakeDriver should stop reporting the same hypervisor hostname?
16:47:26 mriedem sure
16:47:35 mriedem i mean yeah probably
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/)

Earlier   Later