Apache Hadoop by itself does not require root privileges to run
(assuming a non-secure setup). You can run it out of a tarball from a
home directory you have on the server machines.
However, many prefer using packages, such as those from Apache
Bigtop/etc., to install Hadoop and use it. This facilitates easier
upgrades and is closer to the regular management/administration of an
OS. Installing packages in an OS requires root-level permissions due
to the kind of paths it touches. Root isn't a user that will run the
services in the end, however.
The only real reason you need root to actually run the cluster is when
you set it up for security and you'd need to start the DN over
reserved ports. In this case, we need the DN to start as root, but
post-port-bind we switch back down to the actual user.
On Mon, May 13, 2013 at 8:12 PM, Raj Hadoop <[EMAIL PROTECTED]> wrote:
> I am planning to install Hadoop on Linux in a Pseudo Distributed Mode ( One
> Machine ). Do I require 'root' privileges for install ? Please advise.