[HIVE-26459] ReduceRecordProcessor: move to using a timeout version of waitForAllInputsReady(TEZ-3302)

• Type: Improvement
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 4.0.0-alpha-2
• Component/s: HiveServer2
• Labels: pull-request-available
• Assignee: Butao Zhang
• Reporter: Butao Zhang
• Created: 08/Aug/22 09:14
• Updated: 29/Jan/23 00:23

Description

Cluster environment: Haoop 3.1.0, Hive 3.1.0, Tez 0.9.2

I often see some tez jobs stuck for a long time.  I find partial tez reduce tasks seems to wait all inputs ready, and if inputs are not ready, the reduce tasks will wait forever. 

I think we can invoke the timeout version of waitForAllInputsReady, and this optimization was also mentioned in HIVE-14042. So, let's move forward with this optimization.

 

"TezChild" #32 daemon prio=5 os_prio=0 tid=0x0000fffc0a7cb800 nid=0xd012 waiting on condition [0x0000fffbb9b5c000]
   java.lang.Thread.State: WAITING (parking)
    at sun.misc.Unsafe.park(Native Method)
    - parking to wait for  <0x000000068c08cd60> (a java.util.concurrent.locks.AbstractQueuedSynchronizer$ConditionObject)
    at java.util.concurrent.locks.LockSupport.park(LockSupport.java:175)
    at java.util.concurrent.locks.AbstractQueuedSynchronizer$ConditionObject.await(AbstractQueuedSynchronizer.java:2039)
    at org.apache.tez.runtime.InputReadyTracker$InputReadyMonitor.awaitCondition(InputReadyTracker.java:147)
    at org.apache.tez.runtime.InputReadyTracker.waitForAllInputsReady(InputReadyTracker.java:107)
    at org.apache.tez.runtime.api.impl.TezProcessorContextImpl.waitForAllInputsReady(TezProcessorContextImpl.java:141)
    at org.apache.tez.runtime.api.impl.TezProcessorContextImpl.waitForAllInputsReady(TezProcessorContextImpl.java:136)
    at org.apache.hadoop.hive.ql.exec.tez.ReduceRecordProcessor.init(ReduceRecordProcessor.java:122)
    at org.apache.hadoop.hive.ql.exec.tez.TezProcessor.initializeAndRunProcessor(TezProcessor.java:266)
    at org.apache.hadoop.hive.ql.exec.tez.TezProcessor.run(TezProcessor.java:250)
    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$1.run(TaskRunner2Callable.java:61)
    at java.security.AccessController.doPrivileged(Native Method)
    at javax.security.auth.Subject.doAs(Subject.java:422)
    at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1699)
    at org.apache.tez.runtime.task.TaskRunner2Callable.callInternal(TaskRunner2Callable.java:61)
    at org.apache.tez.runtime.task.TaskRunner2Callable.callInternal(TaskRunner2Callable.java:37)
    at org.apache.tez.common.CallableWithNdc.call(CallableWithNdc.java:36)
    at com.google.common.util.concurrent.TrustedListenableFutureTask$TrustedFutureInterruptibleTask.runInterruptibly(TrustedListenableFutureTask.java:108)
    at com.google.common.util.concurrent.InterruptibleTask.run(InterruptibleTask.java:41)
    at com.google.common.util.concurrent.TrustedListenableFutureTask.run(TrustedListenableFutureTask.java:77)
    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) 

Issue Links

• relates to: HIVE-14042 Potential improvements to abort task handling - tez/llap
• relates to: TEZ-3302 Add a version of processorContext.waitForAllInputsReady and waitForAnyInputReady with a timeout
• links to: GitHub Pull Request #3523