Monitoring HBase with CloudWatch

Cloud watch provides and extensive way of monitoring the Hbase/Hadoop nodes in various ways as listed here:

• Cluster status

• Node status

• I/O

• Hbase

Let's discuss Cluster status—it provides the following areas to look at:

Is Idle, Container allocated—container reserved, container pending, apps completed.

Apps failed, apps killed, apps pending, apps running, apps submitted. We will go over one area for explanation.

By double-clicking, a child window is popped up (overlay), which provides details as follows. You can explicitly get the detail by customizing the graph based on hours, days, or mins ago; the graph pulls the data accordingly.

Node status provides details of the following:

• Core Nodes running /pending: As we have two nodes the graph is marked at 2

• Live data nodes: The data node is fully live hence we see the 100 percentage

• Map Reduce total nodes: The total MR node us 2 hence its showing 2

• MR Active nodes: The total active MR nodes are 2 hence its showing 2...