Hi! Have you got any experience deploying on CDH5 cluster with the Spark available in there? Is the standalone Spark strictly necessary? Thanks