JSON Import Job Fails with Tez Execution Framework

Problem

Executing an import job of a JSON object fails when using the Tez execution framework. The same job however is successfully run when force to run in the Hadoop execution framework.

Cause

This is a configuration issue for the particular Tez Job.

Solution

To resolve this issue, increase the available resources for the following parameters:

tez.am.resource.memory.mb

As an example, you may want to double these initial values. The initial values may be seen in the Job Trace logs in the job-conf-cluster.xml file. 

 

To work-around this issue, run the job using the MapReduce execution framework by adding the following parameter to the job: das.execution-framework=Hadoop (for Datameer <5.3) or das.execution-framework=MapReduce (since Datameer 5.3).