Understanding NodeManager
The NodeManager node is the worker node for YARN and is responsible for updating the resource availability on a node to ResourceManager. It is also responsible for monitoring the health of a node and for executing containers for an application.
The following diagram shows various subcomponents of the NodeManager daemon followed by a detailed description of these subcomponents:
Status updates
The resource capability of a cluster is calculated as the sum of the capabilities of all NodeManager nodes. To utilize cluster resources efficiently, it is important to keep track of all resources across the cluster. NodeManager nodes send regular status updates to the ResourceManager. This enables ResourceManager to schedule execution of applications efficiently and increases the cluster performance. Few of the classes defined in the NodeManager framework for sending updates are mentioned in the upcoming sections.
NodeStatusUpdater
Every slave node with the NodeManager daemon registers...