spark-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Matt Cheah <>
Subject Re: Visitor function to RDD elements
Date Tue, 22 Oct 2013 20:02:02 GMT
In this context, it would be able to create a visitor mapping for each partition. However,
I'm looking for the ability to use a single visitor object that will walk over all partitions.

I suppose I could do this if I used coalesce() to combine everything to one partition but
that's too much memory in one partition. Am I misinterpreting how to use it?

From: Mark Hamstra <<>>
Reply-To: "<>"
Date: Tuesday, October 22, 2013 12:51 PM
To: user <<>>
Subject: Re: Visitor function to RDD elements


With care, you can use these and maintain the iteration order within partitions.  Beware,
though, that any reduce functions need to be associative and commutative.

On Tue, Oct 22, 2013 at 12:28 PM, Matt Cheah <<>>
Hi everyone,

I have a driver holding a reference to an RDD. The driver would like to "visit" each item
in the RDD in order, say with a visitor object that invokes visit(item) to modify that visitor's
internal state. The visiting is not commutative (e.g. Visiting item A then B makes a different
internal state from visiting item B then item A). Items in the RDD also are not necessarily

I've looked into accumulators which don't work because they require the operation to be commutative.
Collect() will not work because the RDD is too large; in general, bringing the whole RDD into
one partition won't work since the RDD is too large.

Is it possible to iterate over the items in an RDD in order without bringing the entire dataset
into a single JVM at a time, and/or obtain chunks of the RDD in order on the driver? We've
tried using the internal iterator() method. In some cases, we get a stack trace (running locally
with 3 threads). I've included the stack trace below.


-Matt Cheah

org.apache.spark.SparkException: Error communicating with MapOutputTracker
at org.apache.spark.MapOutputTracker.askTracker(MapOutputTracker.scala:84)
at org.apache.spark.MapOutputTracker.getServerStatuses(MapOutputTracker.scala:170)
at org.apache.spark.BlockStoreShuffleFetcher.fetch(BlockStoreShuffleFetcher.scala:39)
at org.apache.spark.rdd.ShuffledRDD.compute(ShuffledRDD.scala:59)
at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:237)
at org.apache.spark.rdd.RDD.iterator(RDD.scala:226)
at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:36)
at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:237)
at org.apache.spark.rdd.RDD.iterator(RDD.scala:226)
at org.apache.spark.rdd.MappedRDD.compute(MappedRDD.scala:29)
at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:237)
at org.apache.spark.rdd.RDD.iterator(RDD.scala:226)
at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
at sun.reflect.NativeMethodAccessorImpl.invoke(
at sun.reflect.DelegatingMethodAccessorImpl.invoke(
at java.lang.reflect.Method.invoke(
at org.junit.runners.model.FrameworkMethod$1.runReflectiveCall(
at org.junit.runners.model.FrameworkMethod.invokeExplosively(
at org.junit.internal.runners.statements.InvokeMethod.evaluate(
at org.junit.rules.ExternalResource$1.evaluate(
at org.junit.rules.RunRules.evaluate(
at org.junit.runners.ParentRunner.runLeaf(
at org.junit.runners.BlockJUnit4ClassRunner.runChild(
at org.junit.runners.BlockJUnit4ClassRunner.runChild(
at org.junit.runners.ParentRunner$
at org.junit.runners.ParentRunner$1.schedule(
at org.junit.runners.ParentRunner.runChildren(
at org.junit.runners.ParentRunner.access$000(
at org.junit.runners.ParentRunner$2.evaluate(
at org.junit.internal.runners.statements.RunBefores.evaluate(
at org.junit.rules.ExternalResource$1.evaluate(
at org.junit.rules.RunRules.evaluate(
at org.eclipse.jdt.internal.junit.runner.RemoteTestRunner.runTests(
at org.eclipse.jdt.internal.junit.runner.RemoteTestRunner.runTests(
at org.eclipse.jdt.internal.junit.runner.RemoteTestRunner.main(
Caused by: java.util.concurrent.TimeoutException: Futures timed out after [10000] milliseconds
at org.apache.spark.internal.akka.dispatch.DefaultPromise.ready(Future.scala:870)
at org.apache.spark.internal.akka.dispatch.DefaultPromise.result(Future.scala:874)
at org.apache.spark.internal.akka.dispatch.Await$.result(Future.scala:74)
at org.apache.spark.MapOutputTracker.askTracker(MapOutputTracker.scala:81)
... 46 more

View raw message