I'm new to cloudera. I'm trying to submit a mapreduce job using yarn to cloudera cluster. My application uses higher versions of thirdparty libraries. For. e.g. httpclient, httpcore. Before using cloudera, I used to run the job locally in a single node hadoop cluster. I copied the httpclient-4.3.3.jar and httpcore-4.3.2.jar to /home/user/hadoop2.5.2/share/hadoop/common/lib and replaced the existing older version of httpclient and httpcore jars.
My application jar bundle also has a lib folder with the necessary jars. How to force cloudera to use application provided jars instead of those jars on the box?