Search code examples
hadoopoozie

Is configuring queue optional for Oozie actions

I am browsing on a number of Oozie examples and the three actions I'm most interested in are Sqoop, Hive and Pig actions. On some examples, the mapreduce.job.queuename property (mapred.queue.name for older version) is configured, normally to default, like this:

<action>
  <sqoop>
  ...
    <configuration>
      <property>
        <name>mapreduce.job.queuename</name>
        <value>default</value>
      </property>
    </configuration>
  </sqoop>
<action>

On the other hand, other examples didn't specify any queue (no global configuration is specified either). Does configuring the queue necessary? What would happen if I didn't specify any?

Solution

  • Does configuring the queue necessary?

    No It is not necessary.

    What would happen if I didn't specify any?

    If you do not specify the property then the default value which is default queue will be used. But it is a good (most of the time must) practice to specify the queue name in production system, as resources are allocated to the queues based on different business parameters.

    The default values for Hadoop MapReduce 2.7 can be found here.