The call to `joined_cluster?` was passing in the `to_join` node name
instead of the current node name. This resulted in the nodes trying to
rejoin whenever the `to_join` node was offline.

Resolves #347