Uploaded image for project: 'CDAP'
  1. CDAP
  2. CDAP-16471

Remote Hadoop Provisioner can leave behind zombie processes

    Details

    • Type: Bug
    • Status: Closed
    • Priority: Major
    • Resolution: Fixed
    • Affects Version/s: 6.1.1
    • Fix Version/s: 6.2.0, 6.1.2
    • Component/s: Cloud Provisioner
    • Labels:
      None
    • Release Notes:
      Fixed a bug that would cause zombie processes when using the Remote Hadoop Provisioner
    • Rank:
      1|i00uzr:

      Description

      There is a problem with the remote workflow driver where it would occasionally not shut down after the program completes. This will eventually cause java processes to pile up on the remote hadoop master node, consuming all the memory.

      This manifests itself as future program runs failing a few seconds into the run, with errors about being unable to allocate heap memory.

      The root cause is the leveldb compaction thread is hanging around, due to certain tables not getting properly closed in race conditions.

        Attachments

          Activity

            People

            • Assignee:
              ashau Albert Shau
              Reporter:
              ashau Albert Shau
            • Votes:
              0 Vote for this issue
              Watchers:
              3 Start watching this issue

              Dates

              • Created:
                Updated:
                Resolved: