Analyzing Spark jobs

You can use the Spark History Server web UI to analyze Spark jobs on CDP One. You can view application data for both completed and running Spark jobs.

The Spark History Server application UI includes the following information:

  • An event timeline that displays the relative ordering and interleaving of application events. The timeline view is available on three levels: across all jobs, within one job, and within one stage. The timeline also shows executor allocation and deallocation.
  • A list of stages and tasks.
  • The execution directed acyclic graph (DAG) for each job.
  • A summary of RDD sizes and memory usage.
  • Environment - runtime information, property settings, library paths.
  • Information about Spark SQL jobs.