java - MRUnit reducer 测试 : Mismatch in value class

Question

我的 reducer 单元测试在迁移到 MapReduce 2 后抛出一个“值类不匹配”异常:

值类不匹配:预期:类 org.apache.hadoop.io.IntWritable 实际:类 com.company.MyWritable

错误信息本身对我来说很清楚，但我不明白为什么 MRUnit 得到的是临时可写类而不是 IntWritable。

reducer 实现:

public static class TestCountReduce extends
        Reducer<Text, MyWritable, Text, IntWritable> {

    public void reduce(Text key, Iterator<MyWritable> values,
            Context context) throws IOException, InterruptedException {

        ...
        context.write(key, new IntWritable(s.size()));
    }
}

测试设置:

public void setUp() throws IOException {
    Mapper<Object, Text, Text, MyWritable> mapper = new MyMapper();
    Reducer<Text, MyWritable, Text, IntWritable> reducer = new MyReducer();

    mapDriver = new MapDriver<Object, Text, Text, MyWritable>();
    mapDriver.setMapper(mapper);

    reduceDriver = new ReduceDriver<Text, MyWritable, Text, IntWritable>();
    reduceDriver.setReducer(reducer);
}

最后是测试用例:

@Test
public void testReducer() throws IOException {
    List<MyWritable> values = new ArrayList<MyWritable>();
    values.add(new MyWritable("1"));
    values.add(new MyWritable("1"));
    reduceDriver.withInput(new Text("testkey"), values);
    reduceDriver.withOutput(new Text("testkey"), new IntWritable(1));
    reduceDriver.runTest();
}

Answer 1

请检查你的 reducer 实现中的 reduce 方法签名

应该是

public void reduce(Text key, Iterable<MyWritable> values, Context context) throws IOException, InterruptedException {

代替

public void reduce(Text key, Iterator<MyWritable> values, Context context) throws IOException, InterruptedException {