Uploaded image for project: 'Hive'
  1. Hive
  2. HIVE-24322

In case of direct insert, the attempt ID has to be checked when reading the manifest files

    XMLWordPrintableJSON

    Details

    • Type: Bug
    • Status: Resolved
    • Priority: Major
    • Resolution: Fixed
    • Affects Version/s: 4.0.0
    • Fix Version/s: 4.0.0
    • Component/s: None

      Description

      In IMPALA-10247 there was an exception from Hive when tyring to load the data:

      2020-10-13T16:50:53,424 ERROR [HiveServer2-Background-Pool: Thread-23832] exec.Task: Job Commit failed with exception 'org.apache.hadoop.hive.ql.metadata.HiveException(java.io.EOFException)'
      org.apache.hadoop.hive.ql.metadata.HiveException: java.io.EOFException
       at org.apache.hadoop.hive.ql.exec.FileSinkOperator.jobCloseOp(FileSinkOperator.java:1468)
       at org.apache.hadoop.hive.ql.exec.Operator.jobClose(Operator.java:798)
       at org.apache.hadoop.hive.ql.exec.Operator.jobClose(Operator.java:803)
       at org.apache.hadoop.hive.ql.exec.Operator.jobClose(Operator.java:803)
       at org.apache.hadoop.hive.ql.exec.tez.TezTask.close(TezTask.java:627)
       at org.apache.hadoop.hive.ql.exec.tez.TezTask.execute(TezTask.java:342)
       at org.apache.hadoop.hive.ql.exec.Task.executeTask(Task.java:213)
       at org.apache.hadoop.hive.ql.exec.TaskRunner.runSequential(TaskRunner.java:105)
       at org.apache.hadoop.hive.ql.Executor.launchTask(Executor.java:357)
       at org.apache.hadoop.hive.ql.Executor.launchTasks(Executor.java:330)
       at org.apache.hadoop.hive.ql.Executor.runTasks(Executor.java:246)
       at org.apache.hadoop.hive.ql.Executor.execute(Executor.java:109)
       at org.apache.hadoop.hive.ql.Driver.runInternal(Driver.java:721)
       at org.apache.hadoop.hive.ql.Driver.run(Driver.java:488)
       at org.apache.hadoop.hive.ql.Driver.run(Driver.java:482)
       at org.apache.hadoop.hive.ql.reexec.ReExecDriver.run(ReExecDriver.java:166)
       at org.apache.hive.service.cli.operation.SQLOperation.runQuery(SQLOperation.java:225)
       at org.apache.hive.service.cli.operation.SQLOperation.access$700(SQLOperation.java:87)
       at org.apache.hive.service.cli.operation.SQLOperation$BackgroundWork$1.run(SQLOperation.java:322)
       at java.security.AccessController.doPrivileged(Native Method)
       at javax.security.auth.Subject.doAs(Subject.java:422)
       at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1876)
       at org.apache.hive.service.cli.operation.SQLOperation$BackgroundWork.run(SQLOperation.java:340)
       at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
       at java.util.concurrent.FutureTask.run(FutureTask.java:266)
       at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
       at java.util.concurrent.FutureTask.run(FutureTask.java:266)
       at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
       at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
       at java.lang.Thread.run(Thread.java:748)
      Caused by: java.io.EOFException
       at java.io.DataInputStream.readInt(DataInputStream.java:392)
       at org.apache.hadoop.hive.ql.exec.Utilities.handleDirectInsertTableFinalPath(Utilities.java:4587)
       at org.apache.hadoop.hive.ql.exec.FileSinkOperator.jobCloseOp(FileSinkOperator.java:1462)
       ... 29 more
      

      The reason of the exception was that Hive was trying to read an empty manifest file. Manifest files are used in case of direct insert to determine which files needs to be kept and which one needs to be cleaned up. They are created by the tasks and they use the task attempt Id as postfix. In this particular test what happened is that one of the container ran out of memory so Tez decided to kill it right after the manifest file got created but before the paths got written into the manifest file. This was the manifest file for the task attempt 0. Then Tez assigned a new container to the task, so a new attempt was made with attemptId=1. This one was successful, and wrote the manifest file correctly. But Hive didn't know about this, since this out of memory issue got handled by Tez under the hood, so there was no exception in Hive, therefore no clean-up in the manifest folder. And when Hive is reading the manifest files, it just reads every file from the defined folder, so it tried to read the manifest files for attempt 0 and 1 as well.
      If there are multiple manifest files with the same name but different attemptId, Hive should only read the one with the biggest attempt Id.

        Attachments

          Issue Links

            Activity

              People

              • Assignee:
                kuczoram Marta Kuczora
                Reporter:
                kuczoram Marta Kuczora
              • Votes:
                0 Vote for this issue
                Watchers:
                2 Start watching this issue

                Dates

                • Created:
                  Updated:
                  Resolved:

                  Time Tracking

                  Estimated:
                  Original Estimate - Not Specified
                  Not Specified
                  Remaining:
                  Remaining Estimate - 0h
                  0h
                  Logged:
                  Time Spent - 20m
                  20m