Uploaded image for project: 'Spark'
  1. Spark
  2. SPARK-3734

DriverRunner should not read SPARK_HOME from submitter's environment

    XMLWordPrintableJSON

    Details

    • Type: Bug
    • Status: Closed
    • Priority: Major
    • Resolution: Fixed
    • Affects Version/s: 1.1.0, 1.2.0
    • Fix Version/s: 1.1.1, 1.2.0
    • Component/s: Deploy
    • Labels:
      None

      Description

      If you use spark-submit in cluster mode to submit a job to a Spark Standalone cluster, and the JAVA_HOME environment variable is set on the submitting machine, then DriverRunner will attempt to use the submitter's JAVA_HOME to launch the driver process (instead of the worker's JAVA_HOME), which can cause the job to fail unless the submitter and worker have Java installed in the same location.

      This has a pretty simple fix: read JAVA_HOME from sys.env instead of command.environment; PR pending shortly.

        Attachments

          Activity

            People

            • Assignee:
              joshrosen Josh Rosen
              Reporter:
              joshrosen Josh Rosen
            • Votes:
              0 Vote for this issue
              Watchers:
              2 Start watching this issue

              Dates

              • Created:
                Updated:
                Resolved: