System memory 259522560 must be at least 4.718592E8. Please use a larger heap size.

报错内容：

System memory 259522560 must be at least 4.718592E8. Please use a larger heap size.

在Eclipse里开发spark项目，尝试直接在spark里运行程序的时候，遇到下面这个报错：

ERROR SparkContext: Error initializing SparkContext.
java.lang.IllegalArgumentException: System memory 468189184 must be at least 4.718592E8. Please use a larger heap size.

很明显，这是JVM申请的memory不够导致无法启动SparkContext。但是该怎么设呢？

通过查看spark源码，发现源码是这么写的：

/**
   * Return the total amount of memory shared between execution and storage, in bytes.
   */
  private  def getMaxMemory(conf: SparkConf): Long = {
    val systemMemory = conf.getLong( "spark.testing.memory" , Runtime.getRuntime.maxMemory)
    val reservedMemory = conf.getLong( "spark.testing.reservedMemory" ,
      if  (conf.contains( "spark.testing" )) 0  else  RESERVED_SYSTEM_MEMORY_BYTES)
    val minSystemMemory = reservedMemory * 1.5
    if  (systemMemory < minSystemMemory) {
      throw  new  IllegalArgumentException(s "System memory $systemMemory must "  +
        s "be at least $minSystemMemory. Please use a larger heap size." )
    }
    val usableMemory = systemMemory - reservedMemory
    val memoryFraction = conf.getDouble( "spark.memory.fraction" , 0.75)
    (usableMemory * memoryFraction).toLong
  }

　　

所以，这里主要是val systemMemory = conf.getLong("spark.testing.memory", Runtime.getRuntime.maxMemory)。

conf.getLong()的定义和解释是

getLong(key: String, defaultValue: Long): Long
Get a parameter as a long, falling back to a default if not set

所以，我们应该在conf里设置一下spark.testing.memory.

通过尝试，发现可以有2个地方可以设置

1. 自己的源代码处，可以在conf之后加上：

    val conf = new SparkConf().setAppName("word count")
    conf.set("spark.testing.memory", "2147480000")//后面的值大于512m即可

2. 可以在Eclipse的Run Configuration处，有一栏是Arguments，下面有VMarguments，在下面添加下面一行(值也是只要大于512m即可)

-Dspark.testing.memory=1073741824

其他的参数，也可以动态地在这里设置，比如-Dspark.master=spark://hostname:7077

再运行就不会报这个错误了。

解决：

Window——Preference——Java——Installed JREs——选中一个Jre 后 
Edit在Default VM arguments 里加入：-Xmx512M

OK！！！

 

标签:  Spark