I am wanting to enable Kerberos on a Hadoop 2.5 cluster that has data in it we want to keep. Going into the Kerberos wizard in Ambari it warns:
YARN log and local dir will be deleted and ResourceManager will be formatted...
Will this harm the data and application that we have deployed? We have Hive data on dedicated data nodes but it isn’t clear to me whether we need to backup and restore all the data. Do we need to back things up, and if so, how?
You should be okay.
The local directory that is mentioned in your message:
YARN log and local dir will be deleted and ResourceManager will be formatted...
is referring to is YARN's local directory where the YARN stores jar files and files of running jobs. This directory sits on the local filesystem of the machines where the Node Managers run. Hive data is stored in HDFS and not affected.
In addition, when you enable Kerberos in YARN, you have to stop YARN so any formatting that is done is done before jobs are running again.
