On Mon, Apr 9, 2012 at 1:22 PM, Rahul Das <[EMAIL PROTECTED]> wrote:
> I am trying out the possibilities of Hadoop with kerberos. Can somebody clear my doubts in following questions
> 1) Does the Hadoop with Kerberos prevent the HDFS user data from sharing?
Kerberos adds authentication. What you need on top of that is a proper
> 2) If some user wants to share the HDFS data with other user/group then what are the steps to follow.
Same as how you would do it on any other filesystem. Make sure they
exist in same group and that the data is group readable. Use "fs
-chown" and "fs -chmod" to your benefit here.
> 3) Can I create a super user who can access all the jobs and data?
Your namenode-running user ("hdfs" if you use packages), is a
superuser automatically. To create more superusers, set
"dfs.permissions.supergroup" to a specific group name and all users
that also belong to that specified group name will then be elevated as
HDFS superuser. Be careful about doing this, however. Also see
for some more details along the same lines.