Description of problem: Currently whenever the rhs-hadoop-install script doesn't handle cleanup. So suppose a user runs the script and encounters an error (eg. ntp failure), the user will have to manually cleanup the setup (like brick mount, /etc/fstab entries ) before re-running the script. Version-Release number of selected component (if applicable): 1_34-1.el6rhs How reproducible: Always Actual results: Script doesn't handle cleanup or rollback of operations from the previous run. Expected results: Script should be capable to handling the cleanup. It is a tedious task to cleanup all the host machines. Additional info:
The user does not have to undo the steps done by any of the scripts when an error is detected. The scripts are designed to be idempotent. However, there is a script named bin./cleanup.sh that will do most of the necessary clean up if that is desired.