When buying the hardware, I was told that it is desirable to have 3 management nodes so that if one fails the cluster still works. After reading Cloudera documentation, I am rather confused with all this HA. There seems to be several services that can be configured in HA mode. But what about various databases? I have not seen anywhere in the documentation that database should be replicated to several nodes. So how to configure Hadoop so that a failure of one of the management nodes would still allow users to login and submit jobs?