正如问题中所述,问题是它期望IntWritable,但我在reducer的collector.collect中传递了一个Text对象.

我的作业配置有以下映射器输出类：

conf.setMapOutputKeyClass(IntWritable.class);
conf.setMapOutputValueClass(IntWritable.class);

以下减速机输出类：

conf.setOutputKeyClass(Text.class);
conf.setOutputValueClass(IntWritable.class);

我的映射类具有以下定义：

public static class Reduce extends MapReduceBase implements Reducer<IntWritable,IntWritable,Text,IntWritable>

具有所需功能：

public void reduce(IntWritable key,Iterator<IntWritable> values,OutputCollector<Text,IntWritable> output,Reporter reporter)

然后当我打电话时失败：

output.collect(new Text(),new IntWritable());

我是相当新的map reduce,但所有类型似乎都匹配,它编译但是然后在该行上失败,说它期望IntWritable作为reduce类的键.如果重要的话我使用的是0.21版本的Hadoop

这是我的地图类：

public static class Map extends MapReduceBase implements Mapper<LongWritable,IntWritable> {
    private IntWritable node = new IntWritable();
    private IntWritable edge = new IntWritable();
 
    public void map(LongWritable key,Text value,OutputCollector<IntWritable,Reporter reporter) throws IOException {
        String line = value.toString();
        StringTokenizer tokenizer = new StringTokenizer(line);
 
        while (tokenizer.hasMoreTokens()) {
            node.set(Integer.parseInt(tokenizer.nextToken()));
            edge.set(Integer.parseInt(tokenizer.nextToken()));
            if(node.get() < edge.get())
                output.collect(node,edge);
        }
    }
}

和我的减少类：

public static class Reduce extends MapReduceBase implements Reducer<IntWritable,IntWritable> {
 
    IntWritable $= new IntWritable(Integer.MAX_VALUE);
    Text keyText = new Text();
 
    public void reduce(IntWritable key,Reporter reporter) throws IOException {
        ArrayList<IntWritable> valueList = new ArrayList<IntWritable>();
 
        //outputs original edge pair as key and $for value
        while (values.hasNext()) {
            IntWritable value = values.next();
            valueList.add(value);
            keyText.set(key.get() + "," + value.get());
            output.collect(keyText,$);
        }
 
        //outputs all the 2 length pairs 
        for(int i = 0; i < valueList.size(); i++)
            for(int j = i+1; i < valueList.size(); j++)
                output.collect(new Text(valueList.get(i).get() + "," + valueList.get(j).get()),key);
    }
}

和我的工作配置：

JobConf conf = new JobConf(Triangles.class);
conf.setJobName("mapred1");
 
conf.setMapOutputKeyClass(IntWritable.class);
conf.setMapOutputValueClass(IntWritable.class);
 
conf.setOutputKeyClass(Text.class);
conf.setOutputValueClass(IntWritable.class);
 
conf.setMapperClass(Map.class);
conf.setCombinerClass(Reduce.class);
conf.setReducerClass(Reduce.class);
 
conf.setInputFormat(TextInputFormat.class);
conf.setOutputFormat(TextOutputFormat.class);
 
FileInputFormat.setInputPaths(conf,new Path(args[0]));
FileOutputFormat.setOutputPath(conf,new Path("mapred1"));
 
JobClient.runJob(conf);