Uploaded image for project: 'Hadoop Common'
  1. Hadoop Common
  2. HADOOP-6762

exception while doing RPC I/O closes channel

    Details

    • Type: Bug
    • Status: Closed
    • Priority: Critical
    • Resolution: Fixed
    • Affects Version/s: 0.20.2
    • Fix Version/s: 2.0.3-alpha
    • Component/s: None
    • Labels:
      None

      Description

      If a single process creates two unique fileSystems to the same NN using FileSystem.newInstance(), and one of them issues a close(), the leasechecker thread is interrupted. This interrupt races with the rpc namenode.renew() and can cause a ClosedByInterruptException. This closes the underlying channel and the other filesystem, sharing the connection will get errors.

        Attachments

        1. HADOOP-6762.patch
          13 kB
          Harsh J
        2. hadoop-6762.txt
          13 kB
          Todd Lipcon
        3. hadoop-6762.txt
          13 kB
          Todd Lipcon
        4. hadoop-6762.txt
          13 kB
          Todd Lipcon
        5. hadoop-6762-1.txt
          7 kB
          sam rash
        6. hadoop-6762-10.txt
          10 kB
          sam rash
        7. hadoop-6762-2.txt
          6 kB
          sam rash
        8. hadoop-6762-3.txt
          7 kB
          sam rash
        9. hadoop-6762-4.txt
          8 kB
          sam rash
        10. hadoop-6762-6.txt
          9 kB
          sam rash
        11. hadoop-6762-7.txt
          10 kB
          sam rash
        12. hadoop-6762-8.txt
          10 kB
          sam rash
        13. hadoop-6762-9.txt
          10 kB
          sam rash

          Issue Links

            Activity

              People

              • Assignee:
                rash37 sam rash
                Reporter:
                rash37 sam rash
              • Votes:
                0 Vote for this issue
                Watchers:
                16 Start watching this issue

                Dates

                • Created:
                  Updated:
                  Resolved: