| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2018-05-03 | |||
| 14:15:34 | bauzas | it's just we use the docstring elsewhere | |
| 14:15:52 | stephenfin | I think our definition of docstring differs :) | |
| 14:16:05 | bauzas | IMHO, help text is for nova.conf | |
| 14:16:10 | bauzas | but meh | |
| 14:16:18 | stephenfin | Um, that is nova.conf? | |
| 14:16:23 | bauzas | a *python* docstring | |
| 14:16:30 | bauzas | 3 brackets | |
| 14:16:38 | bauzas | that's what I call "docstring" | |
| 14:16:47 | edleafe | bauzas: that's just a string | |
| 14:16:50 | stephenfin | That's a mult-line string | |
| 14:16:52 | stephenfin | Yeah | |
| 14:17:07 | edleafe | A docstring is the comment that follows the method definition | |
| 14:17:12 | stephenfin | https://www.python.org/dev/peps/pep-0257/#id15 | |
| 14:18:27 | bauzas | well, if you should name something because of its usage, I disagree | |
| 14:18:30 | bauzas | but meh | |
| 14:18:56 | bauzas | :p | |
| 14:19:05 | edleafe | bauzas: for example: | |
| 14:19:06 | edleafe | return True | |
| 14:19:06 | edleafe | "this is a docstring" | |
| 14:19:06 | edleafe | def some_method(): | |
| 14:19:19 | edleafe | Doesn't need triple quotes | |
| 14:19:25 | edleafe | unless it's multi-line | |
| 14:19:52 | bauzas | I should write my own language | |
| 14:20:17 | stephenfin | Yeah, theoretically you could even avoid using triple quotes then by escaping the new lines but no one would be so evil | |
| 14:20:50 | stephenfin | bauzas: http://logs.openstack.org/14/564214/2/check/build-openstack-sphinx-docs/a23348b/html/configuration/config.html#devices.enabled_vgpu_types | |
| 14:21:10 | bauzas | grmblblb :p | |
| 14:21:31 | stephenfin | Eventually I'll have you all writing perfect rST | |
| 14:21:43 | stephenfin | At which point I can happily retire to Spain or something | |
| 14:21:47 | mriedem | *perfect* rST? | |
| 14:22:01 | stephenfin | (TM) | |
| 14:22:59 | bauzas | that's glorious | |
| 14:25:28 | mriedem | stephenfin: speaking of https://review.openstack.org/#/c/565526/2..3/nova/conf/compute.py@a894 | |
| 14:25:34 | mriedem | you should appreciate that | |
| 14:26:50 | stephenfin | I do :) oslo_config.sphinxext should soon start reporting those warnings too (though it won't raise an error, to avoid breaking all the stuff) | |
| 14:52:16 | openstack | Launchpad bug 1767139 in OpenStack Compute (nova) "TypeError in _get_inventory_and_update_provider_generation" [Undecided,New] | |
| 14:52:16 | mriedem | efried: https://bugs.launchpad.net/nova/+bug/1767139 | |
| 14:52:21 | mriedem | remember https://review.openstack.org/#/c/562739/ ? | |
| 14:52:37 | mriedem | apparently people are hitting a variant of that in pike starting in 16.1.1 which is when the pike version of that backport was released | |
| 14:52:44 | mriedem | looking at the code, i can't figure out what might be causing that | |
| 14:53:10 | mriedem | they are saying in the bug report that it happens if they start the compute before the control plane is up, but i'm not sure what difference that would make - but placement would have to be running | |
| 14:53:23 | efried | mriedem: Sorry, is it the backport of the original or the backport of the revert that's breaking? | |
| 14:54:18 | mriedem | backport of the original | |
| 14:54:24 | mriedem | there is no backport of the revert | |
| 14:54:29 | mriedem | so this https://review.openstack.org/#/c/524618/ | |
| 14:55:25 | efried | ^ needs to be reverted is what you're saying? | |
| 14:55:30 | mriedem | in the bug they said this happens if compute starts before controller, so i have to assume that includes placement isn't up yet, so we're hitting a flow where the RT is posting inventory for the first time and the compute node RP can't be created | |
| 14:56:03 | mriedem | i would like to avoid a revert if possible | |
| 14:56:10 | mriedem | i'm trying to inspect the change and flow to see if there is something we're missing | |
| 14:57:46 | efried | mriedem: Is it weird that https://review.openstack.org/#/c/562739/ is a revert in ocata, but pike was skipped? | |
| 14:57:59 | mriedem | we only saw the ci failures in ocata | |
| 14:58:04 | mriedem | i haven't seen those in pike | |
| 14:58:18 | efried | okay, I just thought that wasn't allowed. | |
| 14:58:42 | mriedem | i bet i know what it is | |
| 14:58:44 | mriedem | bazinga! | |
| 14:58:46 | efried | I guess it wasn't a cherry-pick of a revert; it was a revert of a cherry-pick. | |
| 14:59:29 | mriedem | set_inventory_for_provider -> _ensure_resource_provider -> _create_resource_provider -> safe_connect returns None because it can't talk to placement yet | |
| 14:59:37 | mriedem | https://review.openstack.org/#/c/524618/2/nova/scheduler/client/report.py@516 | |
| 14:59:40 | mriedem | so we put None in the cache | |
| 14:59:43 | mriedem | kafuckingboom | |
| 14:59:43 | efried | That sounds totally right. | |
| 14:59:57 | melwitt | mriedem: do you have any idea if this bug is still a thing? it's ancient https://bugs.launchpad.net/nova/+bug/1366168 | |
| 15:00:00 | openstack | Launchpad bug 1366168 in OpenStack Compute (nova) "delete private flavor will cause access problem with existed instance" [Medium,Confirmed] | |
| 15:00:42 | mriedem | melwitt: would have to look later, we need a fix for this placement thing in stable | |
| 15:00:51 | melwitt | k, mah bad | |
| 15:00:54 | mriedem | mp | |
| 15:00:56 | mriedem | *np | |
| 15:02:25 | mriedem | I'll wip up a thing | |
| 15:02:48 | mriedem | why we hit this in ocata ci and not pike, i'm not sure | |
| 15:02:59 | mriedem | maybe devstack changed the order of install stuff in pike for the fleetify | |
| 15:11:41 | openstackgerrit | Merged openstack/nova stable/pike: libvirt: Make `cpu_model_extra_flags` case-insensitive for real https://review.openstack.org/565670 | |
| 15:16:32 | melwitt | frickler: hi, could you please lift the -2 on this devstack change https://review.openstack.org/484973? the change that depends on it has been approved https://review.openstack.org/333990 | |
| 15:16:52 | kashyap | mriedem: Thanks for the review here, I wrote that warning in a hurry (hence Workflow -1): https://review.openstack.org/#/c/565242/1/nova/virt/libvirt/driver.py | |
| 15:17:29 | melwitt | frickler: and then the follow up patch (that depends-on the nova change) can be approved to re-enable the novnc tests https://review.openstack.org/527812 | |
| 15:17:32 | kashyap | I want get the wording right. I will update that, and address the unit tests. | |
| 15:26:04 | efried | gibi: o notification deity supreme... | |
| 15:26:12 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Handle @safe_connect returns None side effect in _ensure_resource_provider https://review.openstack.org/566096 | |
| 15:26:40 | efried | gibi: Do you know of any examples you can point to where an op/admin wrote a script/daemon that listens for notifications and triggers some action? | |
| 15:27:24 | gibi | efried: there should be a couple of services listening to notifications, let me dig up links | |
| 15:29:14 | gibi | efried: searchlight: https://github.com/openstack/searchlight/blob/master/searchlight/elasticsearch/plugins/nova/notification_handler.py | |
| 15:29:41 | mriedem | designate sink | |
| 15:29:43 | mriedem | ceilometer | |
| 15:29:45 | mriedem | mistral | |
| 15:29:56 | openstackgerrit | Sylvain Bauza proposed openstack/nova master: libvirt: Configure multiple vGPU types https://review.openstack.org/564214 | |
| 15:29:57 | openstackgerrit | Sylvain Bauza proposed openstack/nova master: libvirt: mdevs returning parent and vendor PCI info https://review.openstack.org/562304 | |
| 15:29:58 | openstackgerrit | Sylvain Bauza proposed openstack/nova master: libvirt: Get inventories per vGPU type https://review.openstack.org/564215 | |
| 15:30:05 | efried | gibi: Thanks. | |
| 15:31:53 | gibi | efried: https://github.com/openstack/watcher/blob/master/watcher/decision_engine/model/notification/nova.py | |
| 15:32:24 | gibi | efried: I mriedem listed the others in the meanwhile | |
| 15:32:39 | efried | gibi, mriedem: Thanks y'all. | |
| 15:38:38 | openstack | Launchpad bug 1766661 in OpenStack Compute (nova) "'host' property is missing for Cinder volume connector when boot from volume" [Medium,Confirmed] | |
| 15:38:38 | mriedem | TheJulia: left some comments in https://bugs.launchpad.net/nova/+bug/1766661 | |
| 15:38:54 | mriedem | tl;dr nova knows how to get the host connector from the db if needed | |
| 15:39:11 | openstackgerrit | Hongbin Lu proposed openstack/nova-specs master: Choose default network on ambiguity https://review.openstack.org/520247 | |
| 15:39:33 | openstackgerrit | Balazs Gibizer proposed openstack/nova-specs master: Network bandwidth resource provider https://review.openstack.org/502306 | |
| 15:42:58 | exarlos | Hi all. I had a user attempt a resize of a VM (horizon) and it has failed and reduced the VM to an error state | |
| 15:43:13 | exarlos | ocata / ubuntu / lvm storage | |
| 15:43:39 | exarlos | It mentions not being able to stat the /var/lib/nova/instances folder | |
| 15:44:09 | mriedem | https://github.com/openstack/nova/blob/stable/ocata/nova/virt/libvirt/driver.py#L7412 | |
| 15:44:34 | mriedem | is it volume-backed or not? | |
| 15:45:17 | exarlos | mriedem: local lvm disks as the backend | |