如何提取RDD的数据到Java ArrayList?

How to extract RDD's data to Java ArrayList?

显而易见的想法是添加元素。

ArrayList<String> myvalues = new ArrayList<String>();

myRdd.foreach(new VoidFunction<org.apache.spark.sql.api.java.Row>() {
    @Override
    public void call(org.apache.spark.sql.api.java.Row row) throws Exception {
        myvalues.add(row.getString(0); // Say I need only first element
    }
});

这个和其他备选方案一直在抛出 org.apache.spark.SparkException:任务不可序列化。我进一步简化了功能..显​​然我在做一些不合逻辑的事情:-

LOG.info("Let's see..");
queryRdd.foreach(new VoidFunction<org.apache.spark.sql.api.java.Row>() {
  @Override
  public void call(org.apache.spark.sql.api.java.Row row) throws Exception {
      LOG.info("Value is : "+row.getString(0));
  }
});

必须有一个简单的方法。这是供参考的堆栈跟踪:

2015-10-08 10:16:48 INFO  UpdateStatementTemplateImpl:141 - Lets see.. 
2015-10-08 10:16:48 WARN  GenericExceptionMapper:20 - Error while executing service
org.apache.spark.SparkException: Task not serializable
        at org.apache.spark.util.ClosureCleaner$.ensureSerializable(ClosureCleaner.scala:166)
        at org.apache.spark.util.ClosureCleaner$.clean(ClosureCleaner.scala:158)
        at org.apache.spark.SparkContext.clean(SparkContext.scala:1476)
        at org.apache.spark.rdd.RDD.foreach(RDD.scala:781)
        at org.apache.spark.api.java.JavaRDDLike$class.foreach(JavaRDDLike.scala:313)
        at org.apache.spark.sql.api.java.JavaSchemaRDD.foreach(JavaSchemaRDD.scala:42)
        at com.simility.cassandra.template.DeviceIDTemplateImpl.test(DeviceIDTemplateImpl.java:144)
        at com.kumbay.service.admin.BusinessEntityService.testSignal(BusinessEntityService.java:1801)
        at com.kumbay.service.admin.BusinessEntityService$$FastClassByCGLIB$7ddd50.invoke(<generated>)
        at org.springframework.cglib.proxy.MethodProxy.invoke(MethodProxy.java:204)
        at org.springframework.aop.framework.CglibAopProxy$CglibMethodInvocation.invokeJoinpoint(CglibAopProxy.java:701)
        at org.springframework.aop.framework.ReflectiveMethodInvocation.proceed(ReflectiveMethodInvocation.java:150)
        at org.springframework.transaction.interceptor.TransactionInterceptor.proceedWithInvocation(TransactionInterceptor.java:96)
        at org.springframework.transaction.interceptor.TransactionAspectSupport.invokeWithinTransaction(TransactionAspectSupport.java:260)
        at org.springframework.transaction.interceptor.TransactionInterceptor.invoke(TransactionInterceptor.java:94)
        at org.springframework.aop.framework.ReflectiveMethodInvocation.proceed(ReflectiveMethodInvocation.java:172)
        at org.springframework.security.access.intercept.aopalliance.MethodSecurityInterceptor.invoke(MethodSecurityInterceptor.java:64)
        at org.springframework.aop.framework.ReflectiveMethodInvocation.proceed(ReflectiveMethodInvocation.java:172)
        at org.springframework.aop.framework.CglibAopProxy$DynamicAdvisedInterceptor.intercept(CglibAopProxy.java:634)

我假设 LOGmyvalues 生活在包含 class 的空间中。因此整个 class(作为 call 的 "capture" 的一部分将被序列化,这是不可能的。

解决方案

首先,将 LOG 替换为简单的 System.out.println,看看是否可行。

其次,创建您在调用中使用的成员的副本;

public void call(...) {
    Log log = LOG // or
    ArrayList<String> inside = myvalues
    inside.add(...)
}

第三,永远不要在 foreach 中使用 ArrayList,因为它在不同的节点上 运行 并且每个节点都会看到自己的 ArrayList。所以,你永远不会如你所愿。

相反,请使用 rdd.collect(...) 来收集您的结果!