If you delete a server that a volume is attached to, and then quickly try to delete the volume before the server has finished deleting (at least, that’s where I think the problem stems from; it could just be that this happens even if you waited for it to fully delete), the volume can get stuck saying that it’s still attached. You cannot run
openstack server remove volume <server id> < volume id> because it will complain that the server doesn’t exist. The only way I could figure out how to get around this is to use the
cinder tool to remove the attachement ID
openstack volume show <volume id> # Grab the attachment ID for use in next step cinder --os-volume-api-version 3.50 attachment-delete <attachment_id> # For some reason you have to specify at least api version 3.50. Why it can't figure this out on it's own is beyond me
Hopefully this will help someone out.