Thanks Harsh. That is a very good explanation.
I am trying to understand how in a production cluster, hadoop user and hadoop clients would be.
What users should exist in NN, JT, DN ?
From: Harsh J <[EMAIL PROTECTED]>
To: "<[EMAIL PROTECTED]>" <[EMAIL PROTECTED]>
Sent: Thursday, August 29, 2013 1:25 AM
Subject: Re: Hadoop client user
The user1 will mainly require a home directory on the HDFS, created by
the HDFS administrator user ('hadoop' in your case): sudo -u hadoop
hadoop fs -mkdir /user/user1; sudo -u hadoop hadoop fs -chown
user1:user1 /user/user1. After this, the user should be able to run
jobs and manipulate files in their own directory.
On Thu, Aug 29, 2013 at 10:21 AM, Hadoop Raj <[EMAIL PROTECTED]> wrote:
> I have a hadoop learning environment on a pseudo distributed mode. It is owned by the user 'hadoop'.
> I am trying to get an understanding on how can another user on this box can act as a Hadoop client and able to create HDFS files and run Map Reduce jobs. Say I have a Linux user 'user1'.
> What permissions , privileges and configuration settings are required for 'user1' to act as a Hadoop client?