Help Center/ MapReduce Service/ FAQs/ Application Development/ How Do I Manage and Use Third-Party JAR Packages for MRS Cluster Components?
Updated on 2025-03-17 GMT+08:00

How Do I Manage and Use Third-Party JAR Packages for MRS Cluster Components?

Some components within an MRS cluster support the use of custom third-party JAR packages to meet specific customization needs. When a user application invokes a third-party JAR package while using a component, it can be stored in a designated directory on each node. Upon restarting the component, the JAR package in that directory will be automatically loaded by the corresponding component program.

The directory is as follows: <Cluster installation directory>/third_lib/<Component name> of the component installation node.

Here are two examples of saving JAR files to their respective directories:

  • Save the JAR package file used by the HBase service to the ${BIGDATA_HOME}/third_lib/HBase directory of the HBase node.
  • Save the JAR package file used by the Hive service to the ${BIGDATA_HOME}/third_lib/Hive directory of each HiveServer instance node.
  • Currently, custom JAR files are supported on HDFS, YARN, MapReduce, Spark, HBase, and Hive.
  • Hive supports only Hive custom JAR files developed by users for UDF implementation. Do not store other open-source JAR files in the directory, including org.apache.*, com.google.*, and javax.*.
  • This directory is supported only by MRS 3.x clusters.