| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2018-01-16 | |||
| 13:55:47 | Spazmotic | Poor XenAPI drivers, no love hehe | |
| 13:55:51 | zioproto | sdague: should I open a bug then ? | |
| 13:57:21 | openstackgerrit | Matthew Edmonds proposed openstack/nova master: Add taskflow to requirements https://review.openstack.org/533779 | |
| 13:57:28 | sdague | zioproto: are you sure that request_id isn't in there? | |
| 13:57:48 | sdague | because in looking at the code, it looks like it should be a key under the 'context' key, if it exists | |
| 13:58:31 | zioproto | it is not there | |
| 13:58:36 | zioproto | I even shared my all config in the thread | |
| 13:58:56 | zioproto | I see the same problem in neutron-server and nova-api | |
| 13:59:03 | gibi | gmann: I did what you suggested in https://review.openstack.org/#/c/533210/7 plus I made an override for irrelevant files to see if it works or not in this context. Based on zuul status the override does not work in this context either | |
| 13:59:05 | zioproto | I did not test other services so far | |
| 14:00:05 | sdague | zioproto: you have a json dump of what you see, I don't see that in the email thread | |
| 14:04:47 | efried | gibi You got the PowerVM OVS patch on your radar again today? https://review.openstack.org/#/c/422512/ | |
| 14:06:27 | efried | stephenfin I remember you saying you were really looking forward to that one ^ :) | |
| 14:06:46 | stephenfin | Ha! | |
| 14:06:51 | stephenfin | I'll look at it today :) | |
| 14:07:29 | efried | Thank you sir. | |
| 14:28:41 | bauzas | mdbooth: I think I addressed most of your concerns in https://review.openstack.org/#/c/528832/ and later patches | |
| 14:29:12 | bauzas | mdbooth: there are still some open issues (at least for suspend) that I'm working on, but after testing my series with a specific node, it works quite fine | |
| 14:31:13 | mdbooth | bauzas: Thanks. wrt side effects in _get_guest_xml, I wonder if I'll have to suck that up. Perhaps we should rename it. | |
| 14:31:48 | bauzas | renaming _get_guest_xml ? well, yes it's not a "get" | |
| 14:31:53 | bauzas | but anyway | |
| 14:34:30 | efried | jaypipes Given mgoddard's feedback, I'm thinking I do need to go to the trouble of reordering ProviderTree.snapshot ahead of set_traits_for_provider so we can get rid of that generation arg right away and have it be useful to him. Unless you see getting this series merged pretty quickly... | |
| 14:34:58 | jaypipes | efried: go for it. I'm workin' on the traits stuff for alloc cands | |
| 14:35:09 | jaypipes | efried: i.e. I won't get to reviews on that until this afternoon | |
| 14:35:33 | efried | ight | |
| 14:42:25 | jaypipes | sean-k-mooney: reverted that one patch in os-vif. you feel like cutting the 1.8.1 release? | |
| 14:42:50 | sean-k-mooney | jaypipes: which one kuryr? | |
| 14:42:55 | jaypipes | ya | |
| 14:43:01 | BKaminski | johnthetubaguy, You around today by chance sir? | |
| 14:43:02 | sean-k-mooney | we have another one | |
| 14:43:07 | jaypipes | oh>? | |
| 14:43:15 | sean-k-mooney | pyroute2 has a filehandel leak | |
| 14:43:32 | sean-k-mooney | im implentint an iptools driver until its patched | |
| 14:43:50 | sean-k-mooney | jaypipes: https://review.openstack.org/#/c/534307/ | |
| 14:44:00 | jaypipes | looking... | |
| 14:44:02 | sean-k-mooney | im gixing some tests currently | |
| 14:44:08 | sean-k-mooney | fixing | |
| 14:44:13 | jaypipes | hehe | |
| 14:44:26 | jaypipes | sean-k-mooney: have fun gixing the implentint | |
| 14:44:51 | sean-k-mooney | i have a revert open also https://review.openstack.org/#/c/512396/ so if i dont have https://review.openstack.org/#/c/534307 ready be this evening we can just straight revert | |
| 14:45:12 | jaypipes | yup, saw that | |
| 14:45:24 | sean-k-mooney | jaypipes: hehe im fixing pep8 first then ill fix spelling and add unit tests | |
| 14:49:18 | openstackgerrit | Lajos Katona proposed openstack/nova master: Deduplicate service status notification samples https://review.openstack.org/534335 | |
| 14:51:02 | damini | Hi while installing newton in devstack enviournment i am getting error of | |
| 14:51:05 | damini | Command "/usr/bin/python -u -c "import setuptools, tokenize;__file__='/tmp/pip-build-EW0AN9/PyECLib/setup.py';f=getattr(tokenize, 'open', open)(__file__);code=f.read().replace('\r\n', '\n');f.close();exec(compile(code, __file__, 'exec'))" install --record /tmp/pip-iZslP6-record/install-record.txt --single-version-externally-managed --compile" failed with error code 255 in /tmp/pip-build-EW0AN9/PyECLib/ | |
| 14:51:37 | damini | Can anyone tell me what the problem is ..i am new to this | |
| 14:52:11 | openstackgerrit | Merged openstack/os-vif master: Revert "Move 'ips' field from Subnet object to VIF object" https://review.openstack.org/534206 | |
| 14:52:31 | zioproto | sdague: taking a random line of the file, it looks like this https://pastebin.com/CcQ0bwi8 | |
| 14:52:37 | zioproto | sdague: this is from nova-api | |
| 14:59:26 | openstackgerrit | Balazs Gibizer proposed openstack/nova master: Deduplicate aggregate notification samples https://review.openstack.org/531162 | |
| 14:59:26 | openstackgerrit | Balazs Gibizer proposed openstack/nova master: Make sure that functional test triggered on sample changes https://review.openstack.org/533210 | |
| 15:00:45 | sdague | zioproto: what version? | |
| 15:01:35 | sdague | because I don't see how that could come from current code, there should be a context key | |
| 15:01:42 | sdague | even if it's an empty array | |
| 15:02:09 | sdague | https://github.com/openstack/oslo.log/blob/e29dc73e9eaf009e7fccd17dcb96abdaf787efe7/oslo_log/formatters.py#L246-L251 | |
| 15:11:55 | openstackgerrit | Jay Pipes proposed openstack/nova master: func tests for nested providers in alloc candidate https://review.openstack.org/531443 | |
| 15:11:56 | openstackgerrit | Jay Pipes proposed openstack/nova master: placement: _get_trees_matching_all() https://review.openstack.org/531512 | |
| 15:11:56 | openstackgerrit | Jay Pipes proposed openstack/nova master: add _has_provider_trees() utility function https://review.openstack.org/531474 | |
| 15:11:57 | openstackgerrit | Jay Pipes proposed openstack/nova master: add test for scenario with sum of child resources https://review.openstack.org/534339 | |
| 15:11:57 | openstackgerrit | Jay Pipes proposed openstack/nova master: add tests for _get_trees_matching_all() with trait https://review.openstack.org/531899 | |
| 15:12:56 | jaypipes | gibi, efried, edleafe, cdent: ready for your signoff on ^^. gibi, the last patch contains the additional test case you'd asked for (which exposed buggy behaviour, btw) | |
| 15:13:16 | efried | ack | |
| 15:13:34 | edleafe | jaypipes: in meeting hell this morning. Will check it out later | |
| 15:13:41 | jaypipes | edleafe: np | |
| 15:26:39 | lyarwood | mdbooth: https://review.openstack.org/#/c/523958/ - do you have time to go over the LUKS change again, I've had to introduce the _format logic again for block device volumes for now | |
| 15:27:06 | mdbooth | lyarwood: Yep. Looking now. | |
| 15:28:05 | gibi | jaypipes: the whole series looks good to me. Thanks for the extra test case | |
| 15:30:20 | mdbooth | lyarwood: How does that work for rbd volumes? | |
| 15:31:29 | lyarwood | mdbooth: with just this patch it doesn't, still waiting on a c-vol patch that formats and empty rbd volume with LUKS based on my image to volume example - https://review.openstack.org/#/c/527979/ | |
| 15:31:40 | lyarwood | an* | |
| 15:32:18 | mdbooth | lyarwood: Why are we only expecting c-vol to format rbd volumes, not all of them? | |
| 15:33:19 | lyarwood | mdbooth: it's a different flow, you can format anything providing a local block device easily | |
| 15:33:29 | lyarwood | mdbooth: with rbd at the moment you need to recreate the device | |
| 15:33:47 | lyarwood | mdbooth: image even | |
| 15:33:58 | mdbooth | lyarwood: Right, from Nova's side. But why would we expect different contracts from different volume types provided by c-vol? | |
| 15:34:35 | mdbooth | lyarwood: I mean, if we're waiting for c-vol to do rbd, why not just wait for all volume types? | |
| 15:35:09 | lyarwood | mdbooth: if this was earlier in the cycle sure but with limited time before M3 I assume we would only have time to get the rbd flow in | |
| 15:35:29 | lyarwood | mdbooth: and then follow up with others in R, removing this call to format volumes then | |
| 15:35:45 | lyarwood | mdbooth: as it wouldn't run anyway if the volumes already contained a LUKS header | |
| 15:46:28 | mdbooth | lyarwood: What's the thinking behind the volume_use_native_luks tunable? Why would you turn that off? | |
| 15:48:27 | lyarwood | mdbooth: TBH that's only there to allow easier upgrade testing without needing to jump between releases downstream | |
| 15:49:28 | lyarwood | mdbooth: moving it to False once encrypted volumes are attached would cause failures, maybe it's best to drop it? | |
| 15:49:36 | mdbooth | lyarwood: Gotcha. That's really useful info. Where can we stash that such that future generations will find it? | |
| 15:50:45 | mdbooth | I mean, ideally we'd drop an unnecessary tunable. Is there any other way to do this for upgrades testing? | |
| 15:51:59 | jaypipes | gibi: thanks for asking for that test case! it highlighted a breakdown of the method in question :) | |
| 15:52:16 | mdbooth | lyarwood: Thinking about upgrades... do we support upgrading nova-compute without rebooting the host? | |
| 15:52:26 | jaypipes | gibi: you're so good at highlighting cracks in my code :) and alex_xu too. he's very good at that. | |
| 15:52:27 | lyarwood | mdbooth: during an actual upgrade in grenade upstream or tripleo-ci downstream | |
| 15:52:30 | lyarwood | mdbooth: yeah | |
| 15:53:06 | mdbooth | ... I think that's ok. We might leave orphaned encryptors, though. | |
| 15:53:26 | mdbooth | But that might prevent disconnect_volume? | |
| 15:53:35 | lyarwood | mdbooth: how so? | |
| 15:53:50 | lyarwood | mdbooth: the current logic will still try to detach an encryptor if a secret isn't found for the volume | |
| 15:54:10 | lyarwood | mdbooth: and a secret is only created when connecting the volume | |
| 15:54:16 | mdbooth | ... and host secret is only added during connect | |
| 15:54:23 | mdbooth | Nice | |
| 15:56:00 | mdbooth | lyarwood: Just thinking about grenade, when would we set that to false even during testing? | |
| 15:56:30 | mdbooth | Surely the only interesting upgrade is across the no-native to native boundary | |
| 15:56:53 | mdbooth | And libvirt/qemu version before is irrelevant, as we always used cryptsetup | |
| 15:56:57 | lyarwood | mdbooth: yeah you wouldn't set it to false | |