如何在Reducer的close方法中发射 [英] How to emit in the close method of Reducer

问题描述

从 MapReduceBase 继承的 close 方法没有参数：

  public void close（）
  

没有 OutputCollector 作为 reduce 方法：

<$ p $ K2> $ reduce $（K2 key，Iterator V2> value，OutputCollector< K3，

在新的hadoop API中，我发现Reducer类有一个 setup 方法：

  void setup（Context context）
  

I认为上下文参数可以发出键和值。
问题是我必须使用旧的API，我不知道如何在关闭方法中发出。
我想要的是在上次调用reduce方法并发出数据时运行一些代码。不幸的是你正在使用旧的API。新的API有一个 cleanup 方法（补充 setup ）。

我不知道Hadoop是否会允许您在Reducer类的私有数据成员中作弊并存储 OutputCollector ，然后请参阅close方法。我猜你不能，但可能值得一试。

The close method inherited from MapReduceBase has no argument:

public void close()

So I don't have OutputCollector as reduce method does:

void reduce(K2 key, Iterator<V2> values, OutputCollector<K3, V3> output, Reporter reporter)

In the new hadoop API, I find the Reducer class has a setup method:

void setup(Context context)

I think the context argument can emit key and value. The problem is I have to use the old API and I don't know how to emit in the close method. What I want is to run some code when the last time reduce method is called and emit data.

解决方案

It is unfortunate you are using the old API. The new API has a cleanup method (complements setup).

I wonder if Hadoop will let you cheat and store the OutputCollector in a private data member in your Reducer class, then refer to that in the close method. I'm guessing you can't, but it might be worth a try.

这篇关于如何在Reducer的close方法中发射的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！