When having 2 master nodes in the Kubernetes production cluster, do they share their etcd database?
Do they have duplication of Etcd each in its own node? or the `etcd should be in a separate machine? how exactly it's working? In case one master node is dead, does the other master node has all its etcd information?