This project has retired. For details please refer to its Attic page.
Falcon - Hive Disaster Recovery

Hive Disaster Recovery

Overview

Falcon provides feature to replicate Hive metadata and data events from source cluster to destination cluster. This is supported for secure and unsecure cluster through Falcon Recipes.

Prerequisites

Following is the prerequisites to use Hive DR

  • Hive 1.2.0+
  • Oozie 4.2.0+

Note: Set following properties in hive-site.xml for replicating the Hive events on source and destination Hive cluster:

    <property>
        <name>hive.metastore.event.listeners</name>
        <value>org.apache.hive.hcatalog.listener.DbNotificationListener</value>
        <description>event listeners that are notified of any metastore changes</description>
    </property>

    <property>
        <name>hive.metastore.dml.events</name>
        <value>true</value>
    </property>

Usage

Bootstrap

Perform initial bootstrap of Table and Database from source cluster to destination cluster

  • Database Bootstrap
For bootstrapping DB replication, first destination DB should be created. This step is expected, since DB replication definitions can be set up by users only on pre-existing DB’s. Second, Export all tables in the source db and Import it in the destination db, as described in Table bootstrap.

  • Table Bootstrap
For bootstrapping table replication, essentially after having turned on the DbNotificationListener on the source db, perform an Export of the table, distcp the Export over to the destination warehouse and do an Import over there. Check the following Hive Export-Import for syntax details and examples. This will set up the destination table so that the events on the source cluster that modify the table will then be replicated.

Setup cluster definition

    $FALCON_HOME/bin/falcon entity -submit -type cluster -file /cluster/definition.xml

Update recipes properties

Update recipe properties file in addons/recipes/hive-disaster-recovery with required attributes for replicating Hive data and metadata from source cluster to destination cluster.

Submit Hive DR recipe

    $FALCON_HOME/bin/falcon recipe -name hive-disaster-recovery -operation HIVE_DISASTER_RECOVERY

Recipe templates for Hive DR is available in addons/recipes/hive-disaster-recovery and copy it to recipe path (falcon.recipe.path=<recipe directory path>) by specifying in client.properties.

Note: If kerberos security is enabled on cluster, use the secure templates for Hive DR from addons/recipes/hive-disaster-recovery