Integrate Spark-SQL (Spark 2.0.1 and later) with Hive

You integrate Spark-SQL with Hive when you want to run Spark-SQL queries on Hive tables. This information is for Spark 2.0.1 or later users.

About this task

For information about Spark-SQL and Hive support, see Spark Feature Support.

NOTE
If you installed Spark with the MapR Installer, the following steps are not required.

Procedure

  1. Copy the hive-site.xml file into the SPARK_HOME/conf directory so that Spark and Spark-SQL recognize the Hive Metastore configuration. Do not create a symbolic link instead of copying the file. You may need to edit the file with settings that are specific to the Spark Thrift server.
  2. Add 644 permission to the hive-site.xml using the following command: 
    sudo chmod 644 /opt/mapr/spark/spark-<sparkVersion>/conf/hive-site.xml
  3. If Hive is configured on Tez (not on MR), you must remove the Tez property from the Spark conf directory hive-site.xml. Delete this entry: 
    <property>
  <name>hive.execution.engine</name>
  <value>tez</value>
</property>
  4. If Hive is configured on PAM, set "hive.metastore.sasl.enabled = true" in the hive-site.xml located in the Spark conf directory.
  5. Add the following additional properties to the /opt/mapr/spark/spark-<version>/conf/spark-defaults.conf file:
    Property Configuration Requirements
    spark.yarn.dist.files For Spark on YARN, specify the location of the hive-site.xml file: 
    /opt/mapr/spark/spark-<spark-version>/conf/hive-site.xml
    spark.sql.hive.metastore.version Specify the Hive version that you are using.
    NOTE
    If you are using Hive Metastore 2.1, set the version to 1.2.1.
  6. Depending on whether you plan to run with impersonation, perform one of the following:
    • Configure user impersonation. See Hive User Impersonation for the steps to configure impersonation in the Spark Thrift server.
    • Set hive.server2.enable.doAs to false in the hive-site.xml file.
  7. To verify the integration, run the following command as the mapr user or as a user that mapr impersonates: 
    <spark-home>/bin/run-example --master <master> [--deploy-mode <deploy-mode>] sql.hive.SparkHiveExample

    The master URL for the cluster is either spark://<host>:7077 or yarn. The deploy-mode is either client or cluster.

What to do next

NOTE
The default port for both HiveServer 2 and the Spark Thrift server is 10000. Therefore, before you start the Spark Thrift server on a node where HiveServer 2 is running, verify that there is no port conflict.
NOTE
If you plan to access Hive tables that store data in HPE Ezmeral Data Fabric Database, you need to copy the Hive HBase handler jar into the Spark jars directory. For example:
cp /opt/mapr/hive/hive-2.1/lib/hive-hbase-handler-2.1.1-mapr-1707.jar /opt/mapr/spark/spark-2.1.0/jars/