HBase High Availability

Most aspects of HBase are highly available in a standard configuration. A cluster typically consists of one Master and three or more RegionServers, with data stored in HDFS. To ensure that every component is highly available, configure one or more backup Masters. The backup Masters run on other hosts than the active Master.

Enabling HBase High Availability Using Cloudera Manager

  1. Go to the HBase service.
  2. Follow the process for adding a role instance and add a backup Master to a different host than the one on which the active Master is running.

Enabling HBase High Availability Using the Command Line

To configure backup Masters, create a new file in the conf/ directory which will be distributed across your cluster, called backup-masters. For each backup Master you wish to start, add a new line with the hostname where the Master should be started. Each host that will run a Master needs to have all of the configuration files available. In general, it is a good practice to distribute the entire conf/ directory across all cluster nodes.

After saving and distributing the file, restart your cluster for the changes to take effect. When the master starts the backup Masters, messages are logged. In addition, you can verify that an HMaster process is listed in the output of the jps command on the nodes where the backup Master should be running. 
$ jps
 15930 HRegionServer
 16194 Jps
 15838 HQuorumPeer
 16010 HMaster
To stop a backup Master without stopping the entire cluster, first find its process ID using the jps command, then issue the kill command against its process ID. 
$ kill 16010