In all versions of Apache Spark, its standalone resource manager accepts code to execute on a 'master' host, that then runs that code on 'worker' hosts. The master itself does not, by design, execute user code. A specially-crafted request to the mast