Github user Leemoonsoo commented on the issue: https://github.com/apache/zeppelin/pull/3240 > Awesome! that is a great news. Just out of curiosity, does the Zeppelin server in KB must be the one which has acess to the Spark_Home or the Spark interpreter which is now a pod? (Just to manage which one should be a gateway in Cloudera cluster). Spark Interpreter is running inside of its own Pod. And Spark driver is running in Spark Interpreter. So Spark Interpreter Pod is the one, who need SPARK_HOME and access from/to Spark master/executors in your cluster.
---