Is there any way to recover a cluster in an error state? I’m unable to join new nodes to this cluster because of the error state.