How does Rancher handle Zookeeper failures?


Sorry if this question has been asked before but how does Rancher handle Zookeeper failures. Is there some kind of monitoring mechanism (like Exhibitor) or self-healing mechanism. Say nodes fail (1 node fails in a 3 node quorum) or for some reason the entire quorum gets botched. How would Rancher deal with this? What would be the steps to take to recover. I guess this is more Mesos based since it relies heavily on Zookeeper for HA. I am also very interested to know how this works for the other cluster managers (Kubernetes and Swarm), not too familiar with their semantics.