Only users that are part of the Hadoop group (that was created in the prerequisites section) will be able to submit Hadoop Jobs. This can be relatively easily if you are using LDAP for authentication, but if you are not, you need to run the command given below on each server in the trusted storage pool and the YARN Master Server, for each user you add.
# useradd -u 1005 -g hadoop tom
The UID of 1005 is arbitrary, you can specify the UID of your choice, but it must be both unique and consistent across the trusted storage pools. Also, the Hadoop Container Executor default properties require that all UIDs be greater or equal to 1000. Thus, 999 is not acceptable as user ID, but 1000 is acceptable. If you want to lower the default minimum acceptable UID, modify the
min.user.id value in the
/etc/hadoop/conf/container-executor.cfg file on every Red Hat Gluster Storage server that is running a NodeManager.
After adding a user who is part of the hadoop group, you need to create a user directory for that user within the default Red Hat Gluster Storage Hadoop Volume. The default Red Hat Gluster Storage Hadoop Volume is the first volume that was created and enabled for Hadoop and is usually called
HadoopVol according to the examples given in installation instructions.
Run the following commands from a server within the Red Hat Gluster Storage trusted storage pool (replacing user_name with the actual user name) for each user that you are adding. You must run this command only once on the default Volume. If you add subsequent volumes, you do not need to repeat this step.
# mkdir /mnt/glusterfs/HadoopVol/user/<username>
# chown <username>:hadoop /mnt/glusterfs/HadoopVol/user/<username>
# chmod 0755 /mnt/glusterfs/HadoopVol/user/<username>
To disable a user from submitting Hadoop Jobs, remove the user from the Hadoop group.