Hadoop Common
  1. Hadoop Common
  2. HADOOP-3771

JobClient.runJob() should not kill the job on IOExceptions

    Details

    • Type: Bug Bug
    • Status: Closed
    • Priority: Major Major
    • Resolution: Duplicate
    • Affects Version/s: None
    • Fix Version/s: None
    • Component/s: None
    • Labels:
      None

      Description

      Currently the JobClient.runJob() submits a job to the job tracker and then periodically asks the JT for the job's progress. On successive IOExceptions the JobClient kills the job. This is not a desired behaviour since the JobClient is issuing a kill-job command to the JT which is not reachable. This is a problem for HADOOP-3245 since its highly possible that the JT can come up anytime and then it makes no sense to kill the job.

        Issue Links

          Activity

          Amar Kamat created issue -
          Amar Kamat made changes -
          Field Original Value New Value
          Link This issue is related to HADOOP-3245 [ HADOOP-3245 ]
          Hide
          dhruba borthakur added a comment -

          This feature will be really useful in clusters that have long-running JobTrackers.

          There are times when we have to restart the JT. HADOOP-3245 gives us the flexibility of restarting the JT anytime without losing any currently submitted jobs. But the entire story is complete only if new job-submissions do not error out when the JT is restarting. This JIRA will be immensely helpful if it can make the JobClient aware of JT restarts and not fail a new job submission.

          Show
          dhruba borthakur added a comment - This feature will be really useful in clusters that have long-running JobTrackers. There are times when we have to restart the JT. HADOOP-3245 gives us the flexibility of restarting the JT anytime without losing any currently submitted jobs. But the entire story is complete only if new job-submissions do not error out when the JT is restarting. This JIRA will be immensely helpful if it can make the JobClient aware of JT restarts and not fail a new job submission.
          Hide
          Amar Kamat added a comment -

          This JIRA will be immensely helpful if it can make the JobClient aware of JT restarts and not fail a new job submission.

          I think HADOOP-3618 is related to this. JobClient should keep on retrying until the JobTracker is ready. Its probably ok not to reveal that the JobTracker is re-starting / initializing.

          Show
          Amar Kamat added a comment - This JIRA will be immensely helpful if it can make the JobClient aware of JT restarts and not fail a new job submission. I think HADOOP-3618 is related to this. JobClient should keep on retrying until the JobTracker is ready. Its probably ok not to reveal that the JobTracker is re-starting / initializing .
          Amar Kamat made changes -
          Assignee Amar Kamat [ amar_kamat ]
          Hide
          Hudson added a comment -
          Show
          Hudson added a comment - Integrated in Hadoop-trunk #581 (See http://hudson.zones.apache.org/hudson/job/Hadoop-trunk/581/ )
          Hide
          Amar Kamat added a comment -

          HADOOP-5577 fixed this.

          Show
          Amar Kamat added a comment - HADOOP-5577 fixed this.
          Amar Kamat made changes -
          Status Open [ 1 ] Resolved [ 5 ]
          Resolution Duplicate [ 3 ]
          Nigel Daley made changes -
          Status Resolved [ 5 ] Closed [ 6 ]
          Owen O'Malley made changes -
          Component/s mapred [ 12310690 ]

            People

            • Assignee:
              Amar Kamat
              Reporter:
              Amar Kamat
            • Votes:
              0 Vote for this issue
              Watchers:
              1 Start watching this issue

              Dates

              • Created:
                Updated:
                Resolved:

                Development