[jira] [Created] (HIVE-26179) In tez reuse container mode, asyncInitOperations are not clear.

["zhengchenyu (Jira)" <ji...@apache.org>]

zhengchenyu created HIVE-26179:
----------------------------------

             Summary: In tez reuse container mode, asyncInitOperations are not clear.
                 Key: HIVE-26179
                 URL: https://issues.apache.org/jira/browse/HIVE-26179
             Project: Hive
          Issue Type: Bug
          Components: Hive, Tez
    Affects Versions: 1.2.1
         Environment: engine: Tez (Note: tez.am.container.reuse.enabled is true)

 
            Reporter: zhengchenyu
            Assignee: zhengchenyu
             Fix For: 4.0.0


In our cluster, we found error like this.
{code}
Vertex failed, vertexName=Map 1, vertexId=vertex_1650608671415_321290_1_11, diagnostics=[Task failed, taskId=task_1650608671415_321290_1_11_000422, diagnostics=[TaskAttempt 0 failed, info=[Error: Error while running task ( failure ) : attempt_1650608671415_321290_1_11_000422_0:java.lang.RuntimeException: java.lang.RuntimeException: Hive Runtime Error while closing operators
    at org.apache.hadoop.hive.ql.exec.tez.TezProcessor.initializeAndRunProcessor(TezProcessor.java:173)
    at org.apache.hadoop.hive.ql.exec.tez.TezProcessor.run(TezProcessor.java:135)
    at org.apache.tez.runtime.LogicalIOProcessorRuntimeTask.run(LogicalIOProcessorRuntimeTask.java:374)
    at org.apache.tez.runtime.task.TaskRunner2Callable$1.run(TaskRunner2Callable.java:73)
    at org.apache.tez.runtime.task.TaskRunner2Callable$1.run(TaskRunner2Callable.java:61)
    at java.security.AccessController.doPrivileged(Native Method)
    at javax.security.auth.Subject.doAs(Subject.java:422)
    at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1730)
    at org.apache.tez.runtime.task.TaskRunner2Callable.callInternal(TaskRunner2Callable.java:61)
    at org.apache.tez.runtime.task.TaskRunner2Callable.callInternal(TaskRunner2Callable.java:37)
    at org.apache.tez.common.CallableWithNdc.call(CallableWithNdc.java:36)
    at com.google.common.util.concurrent.TrustedListenableFutureTask$TrustedFutureInterruptibleTask.runInterruptibly(TrustedListenableFutureTask.java:125)
    at com.google.common.util.concurrent.InterruptibleTask.run(InterruptibleTask.java:57)
    at com.google.common.util.concurrent.TrustedListenableFutureTask.run(TrustedListenableFutureTask.java:78)
    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.lang.RuntimeException: Hive Runtime Error while closing operators
    at org.apache.hadoop.hive.ql.exec.tez.MapRecordProcessor.close(MapRecordProcessor.java:349)
    at org.apache.hadoop.hive.ql.exec.tez.TezProcessor.initializeAndRunProcessor(TezProcessor.java:161)
    ... 16 more
Caused by: java.lang.NullPointerException
    at org.apache.hadoop.hive.ql.exec.MapJoinOperator.closeOp(MapJoinOperator.java:488)
    at org.apache.hadoop.hive.ql.exec.Operator.close(Operator.java:684)
    at org.apache.hadoop.hive.ql.exec.Operator.close(Operator.java:698)
    at org.apache.hadoop.hive.ql.exec.tez.MapRecordProcessor.close(MapRecordProcessor.java:338)
    ... 17 more
{code}

When tez reuse container is enable, and use MapJoinOperator, if same tasks's different taskattemp execute in same container, will throw NPE.


By my debug, I found the second task attempt use first task's asyncInitOperations. asyncInitOperations are not clear when close op, then second taskattemp may use first taskattepmt's mapJoinTables which HybridHashTableContainer.HashPartition is closed, so throw NPE.

 

We must clear asyncInitOperations when op is closed.



--
This message was sent by Atlassian Jira
(v8.20.7#820007)