Really apreciated the article, thanks! I’m currently trying to install SparkR on a CDH 5.5 cluster with Spark deployed via Cloudera Manager.
Based on your last paragraph, if I want to run SparkR on YARN, I should delete Spark services from Cloudera Manager and apply the procedure to all worker nodes. Is that correct ?
In this case, do you think I still need both HDFS and YARN gateway services ?