You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@drill.apache.org by "Sorabh Hamirwasia (JIRA)" <ji...@apache.org> on 2019/01/24 01:09:00 UTC

[jira] [Created] (DRILL-7000) Queries failing with "Failed to aggregate or route the RFW" do not complete

Sorabh Hamirwasia created DRILL-7000:
----------------------------------------

             Summary: Queries failing with "Failed to aggregate or route the RFW" do not complete
                 Key: DRILL-7000
                 URL: https://issues.apache.org/jira/browse/DRILL-7000
             Project: Apache Drill
          Issue Type: Bug
          Components: Execution - Flow
    Affects Versions: 1.16.0
            Reporter: Sorabh Hamirwasia
            Assignee: Sorabh Hamirwasia
             Fix For: 1.16.0


The query does not return even though Fragment 0:0 reports a state change from RUNNING -> FINISHED

 
{code:java}
Following is the jstack output of the Frag0:0.
"23b85137-b102-39a9-70d9-72381c5fb93b:frag:0:0" #16037 daemon prio=10 os_prio=0 tid=0x00007f5f48d415d0 nid=0x1a61 waiting on condition [0x00007f61b32b2000]
 java.lang.Thread.State: TIMED_WAITING (sleeping)
 at java.lang.Thread.sleep(Native Method)
 at org.apache.drill.exec.work.filter.RuntimeFilterSink.close(RuntimeFilterSink.java:116)
 at org.apache.drill.exec.work.filter.RuntimeFilterRouter.waitForComplete(RuntimeFilterRouter.java:113)
 at org.apache.drill.exec.work.foreman.Foreman$ForemanResult.close(Foreman.java:738)
 at org.apache.drill.exec.work.foreman.QueryStateProcessor.wrapUpCompletion(QueryStateProcessor.java:315)
 at org.apache.drill.exec.work.foreman.QueryStateProcessor.running(QueryStateProcessor.java:276)
 at org.apache.drill.exec.work.foreman.QueryStateProcessor.moveToState(QueryStateProcessor.java:92)

locked <0x000000055f9a7468> (a org.apache.drill.exec.work.foreman.QueryStateProcessor)
 at org.apache.drill.exec.work.foreman.QueryStateProcessor$StateSwitch.processEvent(QueryStateProcessor.java:349)
 at org.apache.drill.exec.work.foreman.QueryStateProcessor$StateSwitch.processEvent(QueryStateProcessor.java:342)
 at org.apache.drill.common.EventProcessor.processEvents(EventProcessor.java:107)
 at org.apache.drill.common.EventProcessor.sendEvent(EventProcessor.java:65)
 at org.apache.drill.exec.work.foreman.QueryStateProcessor$StateSwitch.addEvent(QueryStateProcessor.java:344)
 at org.apache.drill.exec.work.foreman.QueryStateProcessor.addToEventQueue(QueryStateProcessor.java:155)
 at org.apache.drill.exec.work.foreman.Foreman.addToEventQueue(Foreman.java:213)
 at org.apache.drill.exec.work.foreman.QueryManager.nodeComplete(QueryManager.java:519)
 at org.apache.drill.exec.work.foreman.QueryManager.access$100(QueryManager.java:65)
 at org.apache.drill.exec.work.foreman.QueryManager$NodeTracker.fragmentComplete(QueryManager.java:483)
 at org.apache.drill.exec.work.foreman.QueryManager.fragmentDone(QueryManager.java:155)
 at org.apache.drill.exec.work.foreman.QueryManager.access$400(QueryManager.java:65)
 at org.apache.drill.exec.work.foreman.QueryManager$1.statusUpdate(QueryManager.java:546)
 at org.apache.drill.exec.rpc.control.WorkEventBus.statusUpdate(WorkEventBus.java:63)
 at org.apache.drill.exec.work.batch.ControlMessageHandler.requestFragmentStatus(ControlMessageHandler.java:253)
 at org.apache.drill.exec.rpc.control.LocalControlConnectionManager.runCommand(LocalControlConnectionManager.java:130)
 at org.apache.drill.exec.rpc.control.ControlTunnel.sendFragmentStatus(ControlTunnel.java:89)
 at org.apache.drill.exec.work.fragment.FragmentStatusReporter.sendStatus(FragmentStatusReporter.java:122)
 at org.apache.drill.exec.work.fragment.FragmentStatusReporter.stateChanged(FragmentStatusReporter.java:91)
 at org.apache.drill.exec.work.fragment.FragmentExecutor.sendFinalState(FragmentExecutor.java:367)
 at org.apache.drill.exec.work.fragment.FragmentExecutor.cleanup(FragmentExecutor.java:219)
 at org.apache.drill.exec.work.fragment.FragmentExecutor.run(FragmentExecutor.java:330)
 at org.apache.drill.common.SelfCleaningRunnable.run(SelfCleaningRunnable.java:38)
 at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
 at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
 at java.lang.Thread.run(Thread.java:748) {code}
 - From the code, it seems that RuntimeFilterSink.close is stuck at

 
{code:java}
while (!asyncAggregateWorker.over.get()) {
 try
{ Thread.sleep(100); }
catch (InterruptedException e)
{ logger.error("interrupted while sleeping to wait for the aggregating worker thread to exit", e); }
}
{code}

 This is because AsyncAggregateWorker exits due to the following exception, before it could set asyncAggregateWorker.over is set to false.
{code:java}
2019-01-22 16:01:18,773 [drill-executor-1301] ERROR o.a.d.e.w.filter.RuntimeFilterSink - Failed to aggregate or route the RFW
 java.lang.ArrayIndexOutOfBoundsException: 1
 at org.apache.drill.exec.work.filter.RuntimeFilterWritable.unwrap(RuntimeFilterWritable.java:67) ~[drill-java-exec-1.16.0-SNAPSHOT.jar:1.16.0-SNAPSHOT]
 at org.apache.drill.exec.work.filter.RuntimeFilterWritable.aggregate(RuntimeFilterWritable.java:78) ~[drill-java-exec-1.16.0-SNAPSHOT.jar:1.16.0-SNAPSHOT]
 at org.apache.drill.exec.work.filter.RuntimeFilterSink.aggregate(RuntimeFilterSink.java:140) ~[drill-java-exec-1.16.0-SNAPSHOT.jar:1.16.0-SNAPSHOT]
 at org.apache.drill.exec.work.filter.RuntimeFilterSink.access$600(RuntimeFilterSink.java:52) ~[drill-java-exec-1.16.0-SNAPSHOT.jar:1.16.0-SNAPSHOT]
 at org.apache.drill.exec.work.filter.RuntimeFilterSink$AsyncAggregateWorker.run(RuntimeFilterSink.java:246) ~[drill-java-exec-1.16.0-SNAPSHOT.jar:1.16.0-SNAPSHOT]
 at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511) [na:1.8.0_151]
 at java.util.concurrent.FutureTask.run(FutureTask.java:266) [na:1.8.0_151]
 at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149) [na:1.8.0_151]
 at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624) [na:1.8.0_151]
 at java.lang.Thread.run(Thread.java:748) [na:1.8.0_151]{code}

 A simple fix would be to add over.set(true) to the finally block in AsyncAggregateWorker.run.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)