Wrong libvirt UUID attaching volume to instance

asked 2020-02-21 07:32:01 -0500

LinkDiegoHylia gravatar image

Hello everyone, I've searched the site and used google, but all the problems I've found are similar to mine, but not quite the same.

I can create, launch and use instances on my openstack system (created manually using the official openstack guide with juju), even if only with the cli and not trough horizon.

The problem is that I cannot attach a secondary volume to any instance. I can create the volume no problem, but then if I try to attach it, it fail silently, even with the debug option on.

Looking trhough the log on the compute node hosting the instance, I've found that the task fail with these errors:

2020-02-21 12:49:18.374 4426 ERROR nova.virt.libvirt.driver [instance: 4b56ba03-77a8-4fae-8e38-eb1b60a71def] libvirtError: Secret not found: no secret with matching uuid '443f5303-e4fb-4b4b-9f2c-a84e37d98621'

...

2020-02-21 12:49:18.379 4426 ERROR nova.virt.block_device [instance: 4b56ba03-77a8-4fae-8e38-eb1b60a71def] libvirtError: Secret not found: no secret with matching uuid '443f5303-e4fb-4b4b-9f2c-a84e37d98621'

...

2020-02-21 12:49:18.786 4426 ERROR nova.compute.manager [instance: 4b56ba03-77a8-4fae-8e38-eb1b60a71def] libvirtError: Secret not found: no secret with matching uuid '443f5303-e4fb-4b4b-9f2c-a84e37d98621'

and so on. The problem is that that uuid is not the correct one, the correct one is correctly saved in libvirt/secret.

If I attach the drive on a stopped instance, then I cannot boot the vm until I detach the volume.

I'm using ceph and cinder (as for the guide).

Thanks for the help!

edit retag flag offensive close merge delete

Comments

Have you compared rbd_secret_uuid in nova.conf with virsh secret-list on your compute node(s)? Those should match.

eblock gravatar imageeblock ( 2020-02-25 03:07:18 -0500 )edit