Strange Cinder rpc timeouts after Xena upgrade

Gorka Eguileor geguileo at redhat.com
Fri Jun 9 08:25:21 UTC 2023


On 06/06, Colby Walsworth wrote:
> Hey Everyone,
>
> We just upgraded from Wallaby to Xena. We use Ceph as our volume/image
> backend. All instances from before the upgrade seem to be unable to be
> shelved/unshelved or live migrated (have not tried offline migration yet).
> Any new instances that are created work fine and all those tasks work fine
> on those. The problem seems to be with cinder:
>
> cinder.api.v3.attachments [req-2af34fab-a4b5-43f3-bd9a-94f360201533
> e28435e0a66740968c523e6376c57f68 179f701c810c425ab004548cc1f76bc9 - default
> default] Unable to create attachment for volume.:
> oslo_messaging.exceptions.MessagingTimeout: Timed out waiting for a reply to
> message ID 8eea6a00568a477c89ced8db17f1179d
>
> the migrations and shelve/unshelve processes start on the hypervisor but
> eventually error out with the above message. Any new instances I create do
> not have this issue at all. I have done all the db upgrades so I don't
> believe this is a db record update issue.
>
> Any ideas of what might be causing this?
>
> Thanks,
>
> Colby
>
>

Hi,

Do you have DEBUG log levels for the request? It would help if we could
have the logs from cinder-api and cinder-volume for the request.

I assume that the cinder-volume service doesn't appear as down in
`cinder service-list`.

Cheers,
Gorka.




More information about the openstack-discuss mailing list