Runtime exception in Delta Iteration

[Janani Chakkaradhari]

Hi,

I am facing the following exception while using Delta iteration. The
dataset size is 27GB. Any suggestions to overcome this issue?

15:14:30,431 ERROR
eu.stratosphere.pact.runtime.task.RegularPactTask             - Error in
task code:  SolutionSet Delta (142/284)
java.lang.RuntimeException: Memory ran out. Compaction failed.
numPartitions: 32 minPartition: 23 maxPartition: 40 number of overflow
segments: 15 bucketSize: 242 Message: Index: 59, Size: 24
        at
eu.stratosphere.pact.runtime.hash.CompactingHashTable.insertOrReplaceRecord(CompactingHashTable.java:526)
        at
eu.stratosphere.pact.runtime.iterative.io.SolutionSetUpdateOutputCollector.collect(SolutionSetUpdateOutputCollector.java:55)
        at
eu.stratosphere.pact.runtime.task.NoOpDriver.run(NoOpDriver.java:64)
        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:853)

Thanks,
Janani

[Stephan Ewen]

The delta iterations are currently still memory sensitive (the last
component in flink that is).

We are working on providing better memory usage for them, but the only
short term solution is to increase the memory size of the JVMs that run the
code, or to use more machines.

Stephan



On Tue, Jun 24, 2014 at 4:59 PM, Janani Chakkaradhari <
[hidden email]> wrote:

> Hi,
>
> I am facing the following exception while using Delta iteration. The
> dataset size is 27GB. Any suggestions to overcome this issue?
>
> 15:14:30,431 ERROR
> eu.stratosphere.pact.runtime.task.RegularPactTask             - Error in
> task code:  SolutionSet Delta (142/284)
> java.lang.RuntimeException: Memory ran out. Compaction failed.
> numPartitions: 32 minPartition: 23 maxPartition: 40 number of overflow
> segments: 15 bucketSize: 242 Message: Index: 59, Size: 24
>         at
>
> eu.stratosphere.pact.runtime.hash.CompactingHashTable.insertOrReplaceRecord(CompactingHashTable.java:526)
>         at
>
> eu.stratosphere.pact.runtime.iterative.io.SolutionSetUpdateOutputCollector.collect(SolutionSetUpdateOutputCollector.java:55)
>         at
> eu.stratosphere.pact.runtime.task.NoOpDriver.run(NoOpDriver.java:64)
>         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:853)
>
> Thanks,
> Janani
>

[Robert Metzger]

What do you exactly mean by dataset size? Is that the size of the input
data or the size of the data that is being held in-memory by the iterations?

The exception is not caused by the input data size. I think you are working
on the graph closeness algorithms. In this case, the exception is probably
caused because your sketches in the nodes are too large (or you have too
many nodes).


On Tue, Jun 24, 2014 at 5:03 PM, Stephan Ewen <[hidden email]> wrote:

> The delta iterations are currently still memory sensitive (the last
> component in flink that is).
>
> We are working on providing better memory usage for them, but the only
> short term solution is to increase the memory size of the JVMs that run the
> code, or to use more machines.
>
> Stephan
>
>
>
> On Tue, Jun 24, 2014 at 4:59 PM, Janani Chakkaradhari <
> [hidden email]> wrote:
>
> > Hi,
> >
> > I am facing the following exception while using Delta iteration. The
> > dataset size is 27GB. Any suggestions to overcome this issue?
> >
> > 15:14:30,431 ERROR
> > eu.stratosphere.pact.runtime.task.RegularPactTask             - Error in
> > task code:  SolutionSet Delta (142/284)
> > java.lang.RuntimeException: Memory ran out. Compaction failed.
> > numPartitions: 32 minPartition: 23 maxPartition: 40 number of overflow
> > segments: 15 bucketSize: 242 Message: Index: 59, Size: 24
> >         at
> >
> >
> eu.stratosphere.pact.runtime.hash.CompactingHashTable.insertOrReplaceRecord(CompactingHashTable.java:526)
> >         at
> >
> >
> eu.stratosphere.pact.runtime.iterative.io.SolutionSetUpdateOutputCollector.collect(SolutionSetUpdateOutputCollector.java:55)
> >         at
> > eu.stratosphere.pact.runtime.task.NoOpDriver.run(NoOpDriver.java:64)
> >         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:853)
> >
> > Thanks,
> > Janani
> >
>