Spark:当迭代太大时,PageRank示例抛出stackoverflowError

Question

我测试了spark默认的PageRank示例并将迭代设置为1024,然后它抛出stackoverflower.我也在我的其他程序中遇到了这个问题.我怎么能解决它.

object SparkPageRank {
  def main(args: Array[String]) {
    if (args.length < 3) {
      System.err.println("Usage: PageRank <master> <file> <number_of_iterations>")
      System.exit(1)
    }
    var iters = args(2).toInt
    val ctx = new SparkContext(args(0), "PageRank",System.getenv("SPARK_HOME"), SparkContext.jarOfClass(this.getClass))
    val lines = ctx.textFile(args(1), 1)
    val links = lines.map{ s => val parts = s.split("\\s+")
    (parts(0), parts(1))
    }.distinct().groupByKey().cache()
    var ranks = links.mapValues(v => 1.0)

    for (i <- 1 to iters) {
        val contribs = links.join(ranks).values.flatMap{ case (urls, rank) =>
        val size = urls.size
        urls.map(url => (url, rank / size))
      }
    ranks = contribs.reduceByKey(_ + _).mapValues(0.15 + 0.85 * _)
    }

    val output = ranks.collect()
    output.foreach(tup => println(tup._1 + " has rank: " + tup._2 + "."))

    System.exit(0)
  }
}

我在这里发布错误.

    [spark-akka.actor.default-dispatcher-15] ERROR LocalActorRefProvider(akka://spark) - guardian failed, shutting down system
java.lang.StackOverflowError
    at scala.collection.mutable.FlatHashTable$class.containsEntry(FlatHashTable.scala:119)
    at scala.collection.mutable.HashSet.containsEntry(HashSet.scala:41)
    at scala.collection.mutable.HashSet.contains(HashSet.scala:58)
    at scala.collection.GenSetLike$class.apply(GenSetLike.scala:43)
    at scala.collection.mutable.AbstractSet.apply(Set.scala:45)
    at org.apache.spark.scheduler.DAGScheduler.org$apache$spark$scheduler$DAGScheduler$$visit$1(DAGScheduler.scala:312)
    at org.apache.spark.scheduler.DAGScheduler$$anonfun$org$apache$spark$scheduler$DAGScheduler$$visit$1$1.apply(DAGScheduler.scala:321)
    at org.apache.spark.scheduler.DAGScheduler$$anonfun$org$apache$spark$scheduler$DAGScheduler$$visit$1$1.apply(DAGScheduler.scala:316)
    at scala.collection.immutable.List.foreach(List.scala:318)
    at org.apache.spark.scheduler.DAGScheduler.org$apache$spark$scheduler$DAGScheduler$$visit$1(DAGScheduler.scala:316)
    at org.apache.spark.scheduler.DAGScheduler$$anonfun$org$apache$spark$scheduler$DAGScheduler$$visit$1$1.apply(DAGScheduler.scala:321)
    at org.apache.spark.scheduler.DAGScheduler$$anonfun$org$apache$spark$scheduler$DAGScheduler$$visit$1$1.apply(DAGScheduler.scala:316)
    at scala.collection.immutable.List.foreach(List.scala:318)
    at org.apache.spark.scheduler.DAGScheduler.org$apache$spark$scheduler$DAGScheduler$$visit$1(DAGScheduler.scala:316)
    at org.apache.spark.scheduler.DAGScheduler.getParentStages(DAGScheduler.scala:326)

Answer 1

我的猜测是，发生错误是因为中间 RDD 直到 才被评估collect()。并在收集时对它们进行递归评估。

尝试cache()在每次迭代中添加评估 RDD，它可能会有所帮助：

ranks = contribs.reduceByKey(_ + _).mapValues(0.15 + 0.85 * _).cache