About Monitoring & Metrics

The Supervisor monitors the standard Hadoop program daemon processes, such as ResourceManager, NodeManager, and HDFS, that are typically part of big data clusters. The monitoring includes collecting a rich array of metrics from monitored programs’ JMX services. If you need data beyond what Pepperdata automatically makes available, you can extend its program monitoring to any UNIX process, such as other common Hadoop stack daemons (for example, Impala), and extend its metrics collection to JMX (Java Management Extensions) beans exposed by Java-based daemons (such as Kafka, Zookeeper, and Presto) through HTTP or RMI interfaces.

Programs that you add for custom monitoring appear in the Pepperdata dashboard visualizations (charts, reports, and tables) in the same way that the standard set of Pepperdata monitored jobs does. Likewise, you can view, analyze, and draw conclusions from extended JMX metrics in the same manner as the built-in collected metrics. In short, the Pepperdata dashboard provides an integrated, consolidated view of the health, status, and resource usage of standard big data jobs and any program of interest, along with details of any JMX metrics the programs make available.

Next: Preconfigured Custom Program Monitoring