I am trying to deploy a cluster and run some sample Spark/scala code on it and while everything works fine when I use zeppelin in standalone mode with default parameters, I cannot get it working in cluster mode. I tried to manually create spark cluster using spark-class and start-master standard shell-files in spark and then passed spark://.. URL to Zeppelin, however after running the code I keep getting different exotic errors (like java sys.process._ library is missing) and after a while spark workers' status changes into DEAD in Spark MasterUI
I also tried to put yarn-client instead of a spark-URL in zeppelin spark.home variable but it did not work producing similar errors.
Could you please help me deploy zeppelin on Ambari cluster and/or hint what am I doing wrong?
