As far as I know and the experience I have with CB 4.0 the writes only fail if there is a failed node which isn’t failover.
So once the node got hard failover because of the policy or a manual operation, the cluster will accept writes again.
Hope this helps you