[ 
https://issues.apache.org/jira/browse/FLINK-945?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14063449#comment-14063449
 ] 

Robert Metzger commented on FLINK-945:
--------------------------------------

I came across this issue while testing the 0.5.2 bugfix release with a large 
KMeans job (27 gb of data). I'm currently trying out if I can reproduce the 
error.

{code}
12:21:19,844 ERROR eu.stratosphere.nephele.taskmanager.Task                     
 - java.lang.NullPointerException: in 
eu.stratosphere.example.java.clustering.KMeans
$.Centroid null of eu.stratosphere.example.java.clustering.KMeans$.Centroid
        at 
org.apache.avro.reflect.ReflectDatumWriter.write(ReflectDatumWriter.java:145)
        at 
org.apache.avro.generic.GenericDatumWriter.write(GenericDatumWriter.java:58)
        at 
eu.stratosphere.api.java.typeutils.runtime.AvroSerializer.serialize(AvroSerializer.java:104)
        at 
eu.stratosphere.pact.runtime.iterative.io.WorksetUpdateOutputCollector.collect(WorksetUpdateOutputCollector.java:52)
        at 
eu.stratosphere.pact.runtime.task.chaining.ChainedMapDriver.collect(ChainedMapDriver.java:71)
        at 
eu.stratosphere.pact.runtime.task.ReduceDriver.run(ReduceDriver.java:121)
        at 
eu.stratosphere.pact.runtime.task.RegularPactTask.run(RegularPactTask.java:505)
        at 
eu.stratosphere.pact.runtime.iterative.task.AbstractIterativePactTask.run(AbstractIterativePactTask.java:132)
        at 
eu.stratosphere.pact.runtime.iterative.task.IterationTailPactTask.run(IterationTailPactTask.java:100)
        at 
eu.stratosphere.pact.runtime.task.RegularPactTask.invoke(RegularPactTask.java:370)
        at 
eu.stratosphere.nephele.execution.RuntimeEnvironment.run(RuntimeEnvironment.java:284)
        at java.lang.Thread.run(Thread.java:745)
Caused by: java.lang.NullPointerException
        at 
eu.stratosphere.core.memory.MemorySegment.putInt(MemorySegment.java:432)
        at 
eu.stratosphere.core.memory.MemorySegment.putIntBigEndian(MemorySegment.java:480)
        at 
eu.stratosphere.nephele.services.memorymanager.AbstractPagedOutputView.writeInt(AbstractPagedOutputView.java:257)
        at 
eu.stratosphere.api.java.typeutils.runtime.DataOutputEncoder.writeInt(DataOutputEncoder.java:55)
        at 
org.apache.avro.reflect.FieldAccessUnsafe$UnsafeIntField.write(FieldAccessUnsafe.java:124)
        at 
org.apache.avro.reflect.ReflectDatumWriter.writeField(ReflectDatumWriter.java:161)
        at 
org.apache.avro.generic.GenericDatumWriter.writeRecord(GenericDatumWriter.java:104)
        at 
org.apache.avro.generic.GenericDatumWriter.write(GenericDatumWriter.java:66)
        at 
org.apache.avro.reflect.ReflectDatumWriter.write(ReflectDatumWriter.java:143)
        ... 11 more
{code}

> Iterations sometimes release memory prematurely
> -----------------------------------------------
>
>                 Key: FLINK-945
>                 URL: https://issues.apache.org/jira/browse/FLINK-945
>             Project: Flink
>          Issue Type: Bug
>          Components: Iterations, Local Runtime
>    Affects Versions: 0.6-incubating, pre-apache-0.5.2
>            Reporter: Stephan Ewen
>
> It seems that the iteration tail is sometimes releasing the memory 
> prematurely, causing still working operations to fail.
> {code}
> java.lang.NullPointerException
>       at eu.stratosphere.core.memory.MemorySegment.put(MemorySegment.java:186)
>       at 
> eu.stratosphere.nephele.services.memorymanager.AbstractPagedOutputView.writeByte(AbstractPagedOutputView.java:214)
>       at 
> eu.stratosphere.nephele.services.memorymanager.AbstractPagedOutputView.write(AbstractPagedOutputView.java:167)
>       at eu.stratosphere.types.Record.serialize(Record.java:1208)
>       at 
> eu.stratosphere.api.java.typeutils.runtime.record.RecordSerializer.serialize(RecordSerializer.java:79)
>       at 
> eu.stratosphere.api.java.typeutils.runtime.record.RecordSerializer.serialize(RecordSerializer.java:27)
>       at 
> eu.stratosphere.pact.runtime.iterative.io.WorksetUpdateOutputCollector.collect(WorksetUpdateOutputCollector.java:52)
>       at 
> eu.stratosphere.test.recordJobs.kmeans.KMeansBroadcast$RecomputeClusterCenter.reduce(KMeansBroadcast.java:255)
>       at 
> eu.stratosphere.pact.runtime.task.GroupReduceDriver.run(GroupReduceDriver.java:103)
>       at 
> eu.stratosphere.pact.runtime.task.RegularPactTask.run(RegularPactTask.java:505)
>       at 
> eu.stratosphere.pact.runtime.iterative.task.AbstractIterativePactTask.run(AbstractIterativePactTask.java:132)
>       at 
> eu.stratosphere.pact.runtime.iterative.task.IterationTailPactTask.run(IterationTailPactTask.java:100)
>       at 
> eu.stratosphere.pact.runtime.task.RegularPactTask.invoke(RegularPactTask.java:370)
>       at 
> eu.stratosphere.nephele.execution.RuntimeEnvironment.run(RuntimeEnvironment.java:284)
>       at java.lang.Thread.run(Thread.java:744)
> {code}



--
This message was sent by Atlassian JIRA
(v6.2#6252)

Reply via email to