How does the cluster administrator ensure that the paths are identical on each node so the ZFS pool can fail over without any problems?