WMI job failed with status 10. Error details: Failed to add device 'Ethernet Connection' [closed]

edit

Yes, the instance has a vnic configured, although it shows status degraded ? see: https://nextcloud.dataenligne.com/index.php/s/N848waxS4kpatby These VMs are running on a non tagged VLAN, IPv4. Let me know if you need further information. Thanks!

Interesting, so it's already connected to a vswitch. What's the neutron port status? You can do: neutron port-show c17da5c8-61ea-4da8-ae81-a02989ab3226 I'd grep for log messages including the port id to see why the port may be handled as unbound, in which case we retry binding it.

there you go: https://nextcloud.dataenligne.com/index.php/s/NX93bedjD7wXr6p status is: BUILD which doesn't seem right to me ?

Thanks, that confirms it, something else must've failed. We'll try to make the retry logic more robust, ensuring that the port bind operations are idempotent.

It would be really helpful if we could see what failed the first time the port was processed. You should check the logs around the time the VM was spawned.

At a first look, this looks like an invalid cache issue. You can either attempt to restart the neutron agent, or disable the wmi cache, adding the following in your neutron config: [os_win] cache_temporary_wmi_objects = false

Thanks for getting back. We've been trying to add the following to neutron_hyperv_agent.conf where the instance is currently running but so far, we're still seeing the same error messages: [os_win] cache_temporary_wmi_objects = false Are we missing something ? Thanks!

That's odd. Just to make sure, you've restarted the neutron agent service after making the config file changes, right? Can you please confirm that the neutron agent log contains this message: 'WMI caching is disabled.' ?

Yep, neutron agent was restarted, however, I'm not seeing any WMI caching disabled messages in the neutron's logfile. We're currently using Ocata; would that make a difference ?

Ocata should include os-win 2.2.0, which has this option. Could you please double check the os-win version? You should be able to just do a "pip freeze | findstr os-win".

The Python scripts folder may not be added to %PATH%, so you may have to use the full path (probably "C:\Progra~1\cloudbase solutions\openstack\nova\python27\scripts\pip.exe").

Looks like we might have our culprit! https://nextcloud.dataenligne.com/index.php/s/RPEKKq3A5DHr8qL It shows: os-win==1.4.2.dev11 We currently have the CloudBase HyperV Compute Ocata version 15.0.0.0 installed.

Hi @lpetrut ; just wondering if you had any thoughts on the previous comment ? thanks !

That's a bit strange, could you please try to update os-win to 2.2.0? I'll double check the ocata msi as well.

I have tried to update os-win to 2.2.0 using the following command: ".\pip.exe install os-win==2.2.0" --- which appears to have worked, however, when doing so, ceilometer & nova compute service (neutron doesn't seem to be affected) won't start.

By running the command through CLI, it says "pbr" needs to be <2.0.0,>=1.8. If I force an install of pbr 1.10.0 (which seems to be the version being used by the msi package), nova won't start because it doesn't meet other dependencies requirements (Requirement.parse('pbr!=2.1.0,>=2.0.0').

Any help would be appreciated.

Looks like the Ocata MSI from cloudbase.it is still using os-win 1.4.2, which doesn't have a bunch of fixes (including that wmi caching option). We do have an unpublished MSI that uses 2.2.0 instead, also taking care of the dependencies.

You may either wait until we release a new Ocata MSI, or fetch os-win from https://github.com/openstack/os-win, checkout the 2.2.0 tag and modify setup.py and requirements.txt to limit the pbr version so that it matches the one used by Nova and then install it with pip, see if this fixes your issue.

I would prefer using the MSI - I'll be sure I don't mix up things. Or, I could use your "unpublished/unofficial" MSI and give feedbacks if you'd like.