NameNode used to be a single point of failure before the release of Hadoop version 2. In Hadoop 1, each cluster consisted of a single NameNode. If this NameNode failed, then the entire cluster would be unavailable. So, until and unless the NameNode service restarted, no one could use the Hadoop cluster. In Hadoop 2, the high availability feature was introduced. It has two NameNodes, one of the NameNodes is in active state while the other NameNode is in standby state. The active NameNode serves the client requests while the standby NameNode maintains synchronization of its state to take over as the active NameNode if the current active NameNode fails.
There is a Quorum Journal Manager (QJM) runs in each NameNode. The QJM is responsible for communicating with JournalNodes using RPC; for example, sending namespace modifications, that is, edits to...
