Enabling the hdfs user to run the YARN jobs on the HDP cluster

You must make configuration changes to enable the hdfs user to run YARN jobs on the HDP cluster.

In the HDP cluster, perform the following steps on the Ambari host:

  1. Open the following file:
    /var/lib/ambari-server/resources/common-services/YARN/2.1.0.2.0/package/templates/container-executor.cfg.j2
  2. Remove the hdfs entry from banned-users list and save the file.
    Sample file contents:
    yarn.nodemanager.local-dirs={{nm_local_dirs}}
    yarn.nodemanager.log-dirs={{nm_log_dirs}}
    yarn.nodemanager.linux-container-executor.group={{yarn_executor_container_group}}
    banned.users=yarn,hdfs,mapred,bin
    min.user.id={{min_user_id}}
    
  3. On the YARN configuration page, verify whether the container-executor configuration template contains hdfs in the banned.users list.
  4. If hdfs is listed in the banner.users list, remove it from the template and save the template.
  5. Restart the following services:
    • Stale services, if any.
    • Ambari server
    • Ambari agent on each host of the cluster.
  6. In the yarn.admin.acl file, add hdfs.
  7. In the etc/hadoop/capacity-scheduler.xml fileSearch file, append hdfs to the yarn.scheduler.capacity.root.acl_submit_applications property.
  8. Restart the YARN service.
  9. Run the kinit command with the hdfs user’s keytab file to authenticate the hdfs user to the Key Distribution Center (KDC).