| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2018-08-16 | |||
| 09:51:13 | mriedem | sean-k-mooney: CERN already has that | |
| 09:51:19 | tssurya | mriedem: yea the time consumption is in the pulling of all instances info | |
| 09:51:22 | mriedem | they have tenants mapped to host aggregates which are mapped to resource provider aggregates | |
| 09:51:26 | mriedem | and that does the pre-filtering | |
| 09:51:46 | sean-k-mooney | mriedem: oh ok cool | |
| 09:52:07 | mriedem | they use this stuff https://docs.openstack.org/nova/latest/admin/configuration/schedulers.html#aggregates-in-placement | |
| 09:52:09 | sean-k-mooney | mriedem: so why do we ever pull the per host info back? | |
| 09:52:34 | openstackgerrit | jichenjc proposed openstack/nova master: Add zvm admin intro and hypervisor information https://review.openstack.org/533125 | |
| 09:52:35 | openstackgerrit | jichenjc proposed openstack/nova master: Add zvm CI information https://review.openstack.org/533512 | |
| 09:52:51 | mriedem | sean-k-mooney: for the affinity filters | |
| 09:53:18 | sean-k-mooney | mriedem: ah but that is after the placmenet query right? | |
| 09:53:25 | mriedem | yes | |
| 09:53:46 | mriedem | pre-filter -> GET /allocation_candidates -> get_host_states -> per-host-instance query -> filters/weighers | |
| 09:54:11 | sean-k-mooney | ok i miss understood. i taught you were implying we were doing the db query in the prefilter | |
| 09:54:33 | mriedem | no the only thing we query there is which host aggregates are restricted for the tenant | |
| 09:54:52 | sean-k-mooney | ya that makes sense. | |
| 09:55:03 | mriedem | https://github.com/openstack/nova/blob/master/nova/scheduler/request_filter.py#L26 | |
| 09:56:58 | openstackgerrit | Matthew Booth proposed openstack/nova master: Py3 fix in fake image service https://review.openstack.org/592056 | |
| 09:57:00 | openstackgerrit | Matthew Booth proposed openstack/nova master: Add regression test for bug 1550919 https://review.openstack.org/591733 | |
| 09:57:01 | openstackgerrit | Matthew Booth proposed openstack/nova master: Don't delete disks on shared storage during evacuate https://review.openstack.org/578846 | |
| 09:57:01 | openstack | bug 1550919 in OpenStack Compute (nova) "[Libvirt]Evacuate fail may cause disk image be deleted" [Medium,In progress] https://launchpad.net/bugs/1550919 - Assigned to Matthew Booth (mbooth-9) | |
| 09:58:15 | sean-k-mooney | mriedem: right we just get the aggreate uuids and stuff them in the destination object | |
| 09:59:41 | mdbooth | mriedem: ^^^ is my proposed for failed evacuate on shared storage. The functional tests exercise all 4 combinations of shared/nonshared instance directory and share/nonshared block. | |
| 10:00:03 | mdbooth | mriedem: Good call ;) | |
| 10:01:39 | sean-k-mooney | there is no stien spec template ye correct? im going to make one reusing the rocky template. | |
| 10:02:05 | sean-k-mooney | is it stien or stein? | |
| 10:02:29 | mdbooth | sean-k-mooney: The latter | |
| 10:03:11 | sean-k-mooney | damb german words not doing the i before e thing | |
| 10:03:35 | sean-k-mooney | that said more english word dont follow that then do also | |
| 10:03:42 | mdbooth | Our crazy european neighbours | |
| 10:04:26 | mdbooth | Don't understand the weightiness of consistent spelling | |
| 10:07:09 | sean-k-mooney | you have seen my speeling correct ( yes i know i misspelled spelling, however that was kind of a typo). sometimes constency is a good thing but that does not mean i like us spelling more then uk. | |
| 10:14:39 | openstackgerrit | Vlad Gusev proposed openstack/nova stable/pike: Fix service list for disabled compute using MC driver https://review.openstack.org/592337 | |
| 10:26:00 | openstackgerrit | Claudiu Belu proposed openstack/nova master: tests: autospecs all the mock.patch usages https://review.openstack.org/470775 | |
| 10:26:01 | openstackgerrit | Claudiu Belu proposed openstack/nova master: hyper-v: autospec classes before they are instantiated https://review.openstack.org/342211 | |
| 10:30:46 | openstackgerrit | Claudiu Belu proposed openstack/nova master: WIP: replace spec with autospec https://review.openstack.org/557299 | |
| 10:38:14 | openstackgerrit | Chen proposed openstack/nova master: Fix soft deleting vm fails after "nova resize" vm https://review.openstack.org/546920 | |
| 10:41:35 | openstackgerrit | Claudiu Belu proposed openstack/nova master: hyperv: Cleans up live migration Planned VM https://review.openstack.org/478943 | |
| 11:59:12 | openstackgerrit | Matthew Booth proposed openstack/nova master: Move static _get_power_off_values to compute_utils https://review.openstack.org/578822 | |
| 11:59:27 | openstackgerrit | Matthew Booth proposed openstack/nova master: Move static _get_power_off_values to compute_utils https://review.openstack.org/578822 | |
| 12:07:38 | mdbooth | Any py3-loving cores fancy knocking this on the head: https://review.openstack.org/#/c/592056/ Trivial: change to explicit string type required by py3. | |
| 12:15:45 | openstackgerrit | Yikun Jiang (Kero) proposed openstack/nova master: Make monkey patch work in uWSGI mode https://review.openstack.org/592285 | |
| 12:15:58 | efried | mdbooth: I glanced at that yesterday, but I'm not a py2-to-3 expert and the commit message is... sparse. Can you help me understand what's wrong and why it needs to be fixed? | |
| 12:16:15 | mdbooth | efried: Because the argument has to be binary | |
| 12:16:31 | mdbooth | If it's not, it gives an error that the argument has to be binary | |
| 12:16:31 | efried | according to...? | |
| 12:16:58 | efried | what gives an error? data.write or _imagedata.get? | |
| 12:17:07 | efried | presumably the former | |
| 12:17:10 | mdbooth | Yeah | |
| 12:17:40 | efried | and where does the error show up? It must not be in the py35 tox job...? | |
| 12:18:02 | mdbooth | Yeah, it was happening in the functional job | |
| 12:18:18 | efried | oooohhhh | |
| 12:18:29 | mdbooth | Although it was non-deterministic until I twigged why | |
| 12:18:29 | efried | I was blinded by the fact that the fake driver lives in the /unit/ path. | |
| 12:18:34 | openstackgerrit | Yikun Jiang (Kero) proposed openstack/nova master: Make monkey patch work in uWSGI mode https://review.openstack.org/592285 | |
| 12:18:38 | efried | but it's used for the func tests too. | |
| 12:18:45 | mdbooth | They're pretty incestuous :) | |
| 12:18:54 | efried | okay, now that I have a way to try to repro locally I can take a better look at it. | |
| 12:19:17 | efried | If you wouldn't mind enhancing the commit message, future me will be grateful. | |
| 12:19:36 | mdbooth | Maybe if you open the file binary it has to have binary data written to it? | |
| 12:19:59 | efried | Well, it makes sense that you would *want* to write binary if you're opening with 'b'. | |
| 12:20:24 | mdbooth | >>> f = open('/tmp/foo', 'wb+') | |
| 12:20:24 | mdbooth | >>> f.write('foo') | |
| 12:20:24 | mdbooth | Traceback (most recent call last): | |
| 12:20:24 | mdbooth | File "<stdin>", line 1, in <module> | |
| 12:20:24 | mdbooth | TypeError: a bytes-like object is required, not 'str' | |
| 12:20:32 | mdbooth | That's the error you get, anyway | |
| 12:20:53 | mdbooth | >>> f.write(b'foo') | |
| 12:20:53 | mdbooth | 3 | |
| 12:21:04 | efried | yeah, so they must have tightened that down in py3, eh? | |
| 12:21:24 | mdbooth | py2: | |
| 12:21:30 | mdbooth | >>> f = open('/tmp/foo', 'wb+') | |
| 12:21:30 | mdbooth | >>> f.write('foo') | |
| 12:21:30 | mdbooth | >>> f.write(b'foo') | |
| 12:21:32 | mdbooth | Both work | |
| 12:21:55 | efried | which func test would fail? | |
| 12:22:19 | efried | I try not to run full suites locally unless I need to reheat my coffee. | |
| 12:22:50 | mdbooth | So... the following functional test would fail non-deterministically depending on whether or not fake image service gave it an image with a kernel and ramdisk or not | |
| 12:23:38 | efried | We have a test that sometimes uses one kind of image and sometimes another, nondeterministically?? | |
| 12:23:52 | mdbooth | efried: Yep. A few :) | |
| 12:24:16 | efried | that's... I can't... but WHY? | |
| 12:24:25 | mdbooth | I felt really good about myself for find that this morning :) | |
| 12:24:50 | mdbooth | efried: non-deterministic dict ordering | |
| 12:25:40 | efried | ...because the reward for good work is... | |
| 12:26:59 | mdbooth | efried: It's not relevant to my bug, or I would have. I was just tripping over it. | |
| 12:27:19 | efried | I was about to say: assume the topic bug isn't really related, so this should be tagged with the py3-first tag (whatever that tag is) | |
| 12:28:50 | efried | btw, I recently discovered with cdent's guidance that you can avoid gerrit re-topic'ing your whole series by using git-review -T (available only in recent versions of git-review) | |
| 12:29:07 | mdbooth | I mean the non-deterministic images, that is. My patch just needs a functioning image service in order to boot. kernel vs non-kernel is irrelevant, except that it caused a bug. | |
| 12:29:39 | mdbooth | I don't really use topics, tbh | |
| 12:30:32 | mdbooth | I've always found it more useful to use the fact that gerrit displays a series in dependency order in 'Related Changes' | |
| 12:30:51 | efried | um. For misc random patch that's fine, but when they're associated with a bug, blueprint, or initiative (like mox removal or py3 first), it's useful. | |
| 12:31:22 | efried | Well, yeah, on that note, in this case the fix really doesn't need to be in that series at all, does it? | |
| 12:31:48 | mdbooth | It does, because the functional test which follows it was tripping up on it | |
| 12:31:58 | mdbooth | non-deterministically | |
| 12:32:00 | efried | but only in py3, which isn't set up in the gate yet? | |
| 12:32:13 | mdbooth | Yeah, we run functional on py3 | |
| 12:32:20 | efried | "we" do? | |
| 12:32:23 | efried | looking again... | |
| 12:32:44 | efried | oic | |
| 12:32:49 | mdbooth | http://logs.openstack.org/33/591733/4/check/nova-tox-functional-py35/d9266d8/ | |