Connecting to the Cluster

Application are often run on nodes that are not part of the MapR Cluster. There are many methods to connect to a MapR cluster; this section will briefly describe each option.

MapR Client
The MapR client includes the libraries and utilities required on an edge node to perform the following: connect to the cluster, submit MapReduce jobs, submit YARN applications, run hadoop fs commands, and run hadoop mfs commands. However, to run applications that access data from MapR-DB or MapR Streams, you must configure additional dependencies. For more information about the MapR client, see MapR Client and Application Connections to the Cluster.
Note: Although it is not recommended, you can include the MapR-FS JAR file in the application instead of installing the MapR client. However, there are caveats and specific requirements to make this work. For information, see Using the MapR-FS JAR to Connect to the Cluster.
MapR Persistent Application Client Container (PACC)
The MapR Persistent Application Client Container (PACC) is a Docker-based container image that includes a container-optimized MapR client. The PACC provides seamless access to MapR Converged Data Platform services, including MapR-FS, MapR-DB, and MapR Streams. The PACC includes the POSIX client, the MapR client, and the libraries required to build MapR-DB and MapR Streams applications. For more information, see About the MapR Persistent Application Client Container (PACC).
MapR POSIX Clients
MapR POSIX clients enable app servers, web servers, and other client nodes and applications to read and write directly and securely to the MapR-FS. For more information about the POSIX clients, see POSIX Clients and MapR POSIX Clients.
MapR NFS Clients
You can mount the cluster itself via NFS so that your applications can read and write data directly. For more information, see Setting Up MapR NFS.