Create Your Own Metastore Event Listeners in Hive With Scala

Hive metastore event listeners are used to detect every single event that takes place whenever an event is executed in Hive. If you want a certain action to take place for an event, you can override MetaStorePreEventListener and provide your own implementation.

In this article, we will learn how to create our own metastore event listeners in Hive using Scala and SBT.

So let's get started!

First, add the following dependencies in your build.sbt file:

libraryDependencies += "org.apache.hive" % "hive-exec" % "1.2.1" excludeAll
  ExclusionRule(organization = "org.pentaho")

libraryDependencies += "org.apache.hadoop" % "hadoop-common" % "2.7.3"

libraryDependencies += "org.apache.httpcomponents" % "httpclient" % "4.3.4"

libraryDependencies += "org.apache.hadoop" % "hadoop-client" % "2.6.0"

libraryDependencies += "org.apache.hive" % "hive-service" % "1.2.1"

unmanagedJars in Compile += file("/usr/lib/hive/lib/hive-exec-1.2.1.jar")

assemblyMergeStrategy in assembly := {
  case PathList("META-INF", xs @ _*) => MergeStrategy.discard
  case x => MergeStrategy.first
}

Now, create your first class. You can name it anything, but I named it OrcMetastoreListener. This class must extend the MetaStorePreEventListener class of Hive and take the Hadoop conf as the constructor argument:

import org.apache.hadoop.conf.Configuration
import org.apache.hadoop.hive.metastore.MetaStorePreEventListener
import org.apache.hadoop.hive.metastore.events.PreEventContext.PreEventType._
import org.apache.hadoop.hive.metastore.events._

class OrcMetastoreListener(conf: Configuration) extends MetaStorePreEventListener(conf) {

  override def onEvent(preEventContext: PreEventContext): Unit = {
    preEventContext.getEventType match {
      case CREATE_TABLE =>
        val tableName = preEventContext.asInstanceOf[PreCreateTableEvent].getTable
        tableName.getSd.setInputFormat("org.apache.hadoop.hive.ql.io.orc.OrcInputFormat")
        tableName.getSd.setOutputFormat("org.apache.hadoop.hive.ql.io.orc.OrcOutputFormat")
      case ALTER_TABLE =>
        val newTableName = preEventContext.asInstanceOf[PreAlterTableEvent].getNewTable
        newTableName.getSd.setInputFormat("org.apache.hadoop.hive.ql.io.orc.OrcInputFormat")
        newTableName.getSd.setOutputFormat("org.apache.hadoop.hive.ql.io.orc.OrcOutputFormat")
      case _ => //do nothing

    }

  }
}

The pre-event context contains all the Hive metastore events. In my case, I want all tables generated in Hive to use the Hive input format and output format — and the same thing for the altering command.

The best use case for this listener is when somebody wants to query a data source such as spark or any other data source using its own custom input format and even don't want to alter the schema of hive table to use his custom input format

Now, let's build a JAR from the core and use it in Hive.

First, add an SBT assembly plugin in your plugins.sbt file:

addSbtPlugin("com.eed3si9n" % "sbt-assembly" % "0.14.5")

Now, go to your root project and build the JAR with command SBT assembly. It will build your JAR, collect your JAR, and put it in your $HIVE_HOME/lib path. Inside the $HIVE_HOME/conf  folder, add the following contents in hive-site.xml:

<configuration>
	<property>
		<name>javax.jdo.option.ConnectionURL</name>
		<value>jdbc:derby:metastore_db;create=true</value>
		<description>JDBC connect string for a JDBC metastore</description>
	</property>
	<property>
		<name>hive.metastore.schema.verification</name>
		<value>false></value>
	</property>
	<property>
		<name>hive.metastore.pre.event.listeners</name>
		<value>metastorelisteners.OrcMetastoreListener</value>
	</property>
</configuration>

Now, create a table in Hive and describe it:

hive> CREATE TABLE HIVETABLE(ID INT);
OK
Time taken: 2.742 seconds
hive> DESC EXTENDED HIVETABLE
    > ;
OK
id                  int                                     

Detailed Table Information Table(tableName:hivetable, dbName:default, owner:hduser,e, inputFormat:org.apache.hadoop.hive.ql.io.orc.OrcInputFormat, outputFormat:org.apache.hadoop.hive.ql.io.orc.OrcOutputFormat, compressed:false, 
Time taken: 0.611 seconds, Fetched: 3 row(s)

And that's it!