- When the EBS AMI is hanging around but the components don't exist any more
- A device is in 'deleting' state then goes straight to non-existent (no such ID) therefore throwing an exception during wait_for_volume_status instead of returning deleted state.
Somewhat confusingly there are 2 no longer exists situations. One is when you have recently deleted a device, it should return a 'deleted' state. The other is after the device has been gone for an extended period, and you will get a 'no such ID' type error.
- Device stuck in a non-optimal state. Should just catch time-out, evaluate to see whether the device is in a state that should become acceptable (e.g. if a block device is stuck in deleting state after time-out we will try to continue, ideally we want deleted state, but we can handle it).