I had my exam three days ago. Let me answer my own question. I do not know which HDP version it was. The default version `running `spark-shell` in the terminal was Spark 1.6. I did not try to change it. Yes, I was solving the tasks with Scala in the Spark Shell. However you have to save all you commands in the provided text files. It was not necessary to build a JAR manually to submit it. But there could be a task to submit a provided JAR to YARN. I do not know. You can use `hadoop fs` commands in the terminal to browse the HDFS. I do not think so. You do not have to. Since the VM is running in your browser it automatically uses your local one. Further information: I think there were some links on the desktop to the documentation. But I did no use it. You do not have to write CSV files with a header. Read carefully, the delimiter do not have to be the same in all tasks. The general question pattern is: read this from here, do something with it, write the results to here. Because only the output counts you have to read the tasks carefully (ordering of columns, sorting of the data, delimiter in CSV files, ...) It is up to you how to solve the tasks. You can use RDD or SparkSQL API. The exam is not really difficult if you work through the exam objectives.
... View more
