Well... I have found a little free tool for integrating linux operating systems with Windows Active Directory fairly seamlessly. It is called LikeWise Open (http://www.likewise.com/community/index.php/download/). Just using a couple of simple command-line statements, I was able to
a)Add the machine to the active directory
b)Allow users to log in, changing the default shell to /bin/bash
The only thing I needed to do was manually modify the /etc/sudoers file to add our ADs group called "hadoop users" be able to run all commands
DOMAIN\\hadoop^users ALL=(ALL) NOPASSWD: ALL
Done! Now Hive and Sqoop can be executed using 'sudo hive -e "SELECT * FROM table1;"', or "sudo sqoop-import ......."
This is by far the easiest way to get your windows users to get access to the the cluster without having to manage every account individually.
Ok, just a little update to get rid of the requirement for sudo in the hive statement.
ReplyDeleteExecute the following for each active directory user, it is a bit tedious, but will get the job done!
chown -hR hdfs:hadoop /var/lib/hadoop-0.20/cache/