apache-sparkdockerneo4jhdfsneo4j-mazerunner

neo4j-mazerunner, How to Increase memory size in docker-compose.yml

Using kbastani/spark-neo4j with docker-compose on MacBook pro (16gb mem), I'm trying to analyze strongly_connected_components of my graph.

I have a graph with about 60,000 nodes (n1:Node {id:1})-[r:NEXT {count:100}]->(n2:Node {id:2}).

Using the neo4j browser I've managed to get pagerank processed back to my nodes.

However, when I try to run a more complex algorithm like strongly_connected_components, I get the following error:

mazerunner_1  | 16/11/29 14:58:01 ERROR Utils: Uncaught exception in thread SparkListenerBus
mazerunner_1  | java.lang.OutOfMemoryError: Java heap space
mazerunner_1  |     at org.apache.spark.ui.jobs.JobProgressListener$$anonfun$onJobStart$5$$anonfun$apply$9.apply(JobProgressListener.scala:200)
mazerunner_1  |     at org.apache.spark.ui.jobs.JobProgressListener$$anonfun$onJobStart$5$$anonfun$apply$9.apply(JobProgressListener.scala:200)
mazerunner_1  |     at scala.collection.mutable.MapLike$class.getOrElseUpdate(MapLike.scala:189)
mazerunner_1  |     at scala.collection.mutable.AbstractMap.getOrElseUpdate(Map.scala:91)
mazerunner_1  |     at org.apache.spark.ui.jobs.JobProgressListener$$anonfun$onJobStart$5.apply(JobProgressListener.scala:200)
mazerunner_1  |     at org.apache.spark.ui.jobs.JobProgressListener$$anonfun$onJobStart$5.apply(JobProgressListener.scala:198)
mazerunner_1  |     at scala.collection.IndexedSeqOptimized$class.foreach(IndexedSeqOptimized.scala:33)
mazerunner_1  |     at scala.collection.mutable.WrappedArray.foreach(WrappedArray.scala:34)
mazerunner_1  |     at org.apache.spark.ui.jobs.JobProgressListener.onJobStart(JobProgressListener.scala:198)
mazerunner_1  |     at org.apache.spark.scheduler.SparkListenerBus$class.onPostEvent(SparkListenerBus.scala:34)
mazerunner_1  |     at org.apache.spark.scheduler.LiveListenerBus.onPostEvent(LiveListenerBus.scala:31)
mazerunner_1  |     at org.apache.spark.scheduler.LiveListenerBus.onPostEvent(LiveListenerBus.scala:31)
mazerunner_1  |     at org.apache.spark.util.ListenerBus$class.postToAll(ListenerBus.scala:53)
mazerunner_1  |     at org.apache.spark.util.AsynchronousListenerBus.postToAll(AsynchronousListenerBus.scala:36)
mazerunner_1  |     at org.apache.spark.util.AsynchronousListenerBus$$anon$1$$anonfun$run$1.apply$mcV$sp(AsynchronousListenerBus.scala:76)
mazerunner_1  |     at org.apache.spark.util.AsynchronousListenerBus$$anon$1$$anonfun$run$1.apply(AsynchronousListenerBus.scala:61)
mazerunner_1  |     at org.apache.spark.util.AsynchronousListenerBus$$anon$1$$anonfun$run$1.apply(AsynchronousListenerBus.scala:61)
mazerunner_1  |     at org.apache.spark.util.Utils$.logUncaughtExceptions(Utils.scala:1618)
mazerunner_1  |     at org.apache.spark.util.AsynchronousListenerBus$$anon$1.run(AsynchronousListenerBus.scala:60)
mazerunner_1  | Exception in thread "SparkListenerBus" java.lang.OutOfMemoryError: Java heap space
mazerunner_1  |     at org.apache.spark.ui.jobs.JobProgressListener$$anonfun$onJobStart$5$$anonfun$apply$9.apply(JobProgressListener.scala:200)
mazerunner_1  |     at org.apache.spark.ui.jobs.JobProgressListener$$anonfun$onJobStart$5$$anonfun$apply$9.apply(JobProgressListener.scala:200)
mazerunner_1  |     at scala.collection.mutable.MapLike$class.getOrElseUpdate(MapLike.scala:189)
mazerunner_1  |     at scala.collection.mutable.AbstractMap.getOrElseUpdate(Map.scala:91)
mazerunner_1  |     at org.apache.spark.ui.jobs.JobProgressListener$$anonfun$onJobStart$5.apply(JobProgressListener.scala:200)
mazerunner_1  |     at org.apache.spark.ui.jobs.JobProgressListener$$anonfun$onJobStart$5.apply(JobProgressListener.scala:198)
mazerunner_1  |     at scala.collection.IndexedSeqOptimized$class.foreach(IndexedSeqOptimized.scala:33)
mazerunner_1  |     at scala.collection.mutable.WrappedArray.foreach(WrappedArray.scala:34)
mazerunner_1  |     at org.apache.spark.ui.jobs.JobProgressListener.onJobStart(JobProgressListener.scala:198)
mazerunner_1  |     at org.apache.spark.scheduler.SparkListenerBus$class.onPostEvent(SparkListenerBus.scala:34)
mazerunner_1  |     at org.apache.spark.scheduler.LiveListenerBus.onPostEvent(LiveListenerBus.scala:31)
mazerunner_1  |     at org.apache.spark.scheduler.LiveListenerBus.onPostEvent(LiveListenerBus.scala:31)
mazerunner_1  |     at org.apache.spark.util.ListenerBus$class.postToAll(ListenerBus.scala:53)
mazerunner_1  |     at org.apache.spark.util.AsynchronousListenerBus.postToAll(AsynchronousListenerBus.scala:36)
mazerunner_1  |     at org.apache.spark.util.AsynchronousListenerBus$$anon$1$$anonfun$run$1.apply$mcV$sp(AsynchronousListenerBus.scala:76)
mazerunner_1  |     at org.apache.spark.util.AsynchronousListenerBus$$anon$1$$anonfun$run$1.apply(AsynchronousListenerBus.scala:61)
mazerunner_1  |     at org.apache.spark.util.AsynchronousListenerBus$$anon$1$$anonfun$run$1.apply(AsynchronousListenerBus.scala:61)
mazerunner_1  |     at org.apache.spark.util.Utils$.logUncaughtExceptions(Utils.scala:1618)
mazerunner_1  |     at org.apache.spark.util.AsynchronousListenerBus$$anon$1.run(AsynchronousListenerBus.scala:60)

I have tried to modify my docker-compose.yml file like so:

hdfs:
  environment:
    - "JAVA_OPTS=-Xmx5g"
  image: sequenceiq/hadoop-docker:2.4.1
  command: /etc/bootstrap.sh -d -bash
mazerunner:
  environment:
    - "JAVA_OPTS=-Xmx5g"
  image: kbastani/neo4j-graph-analytics:latest
  links:
   - hdfs
graphdb:
  environment:
    - "JAVA_OPTS=-Xmx2g"
  image: kbastani/docker-neo4j:latest
  ports:
   - "7474:7474"
   - "1337:1337"
  volumes:
   - /opt/data
  links:
   - mazerunner
   - hdfs

with no success. How do I configure spark & hdfs to use the maximum available memory?

Solution

  • My solution was increasing the memory size of the virtual machine. On my Virtual Box UI, I have adgjusted the "Base memory" size slider.

    enter image description here