I have Zeppelin 0.6.0 running on a linux server, this server is also an edge node to my Hadoop cluster. I have a Hadoop cluster with spark 1.5 running on it. By pointing the spark master to yarn-client, I can run spark in zeppelin without a problem. I also have spark 2.0 unpacked in my home directory and can navigate to the bin/spark-shell in my home directory and run spark 2.0 from the repl with –master yarn-client . That works fine. When I try to point my spark-home to my home directory so that zeppelin will run spark 2.0, I get the error message above. Any suggestions on what I am doing wrong? I do not have rights to simply upgrade the cluster to spark 2.0.
Thanks, Jeremy