hadoop reduce task running even after telling on command line as -D mapred.reduce.tasks=0

问题

I have aMapReduce program as

public static class MapClass extends MapReduceBase implements Mapper<Text, Text, IntWritable, IntWritable> {

        private final static IntWritable uno = new IntWritable(1);
        private IntWritable citationCount = new IntWritable();

        public void map(Text key, Text value, OutputCollector<IntWritable, IntWritable> output, Reporter reporter) throws IOException {
            citationCount.set(Integer.parseInt(value.toString()));
            output.collect(citationCount, uno);
        }
    }

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

        public void reduce(IntWritable key, Iterator<IntWritable> values, OutputCollector<IntWritable, IntWritable> output, Reporter reporter) throws IOException {
            int count = 0;
            while (values.hasNext()) {
                count += values.next().get();
            }

            output.collect(key, new IntWritable(count));
        }
    }

I want to run only map task, where output should be of form <number, 1> When I run this from command-line I say

$ hadoop jar Hadoop-programs.jar com/hadoop/patent/CitationHistogram input output -Dmapred.reduce.tasks=0

But this is what I see on command-line output

12/07/30 06:13:14 INFO mapred.JobClient:  map 50% reduce 0%
12/07/30 06:13:23 INFO mapred.JobClient:  map 58% reduce 0%
12/07/30 06:13:26 INFO mapred.JobClient:  map 60% reduce 8%
12/07/30 06:13:29 INFO mapred.JobClient:  map 68% reduce 8%
12/07/30 06:13:32 INFO mapred.JobClient:  map 76% reduce 8%
12/07/30 06:13:35 INFO mapred.JobClient:  map 85% reduce 16%
12/07/30 06:13:38 INFO mapred.JobClient:  map 93% reduce 16%
12/07/30 06:13:41 INFO mapred.JobClient:  map 98% reduce 16%
12/07/30 06:13:44 INFO mapred.JobClient:  map 100% reduce 16%
12/07/30 06:13:55 INFO mapred.JobClient:  map 100% reduce 69%
12/07/30 06:13:58 INFO mapred.JobClient:  map 100% reduce 78%
12/07/30 06:14:01 INFO mapred.JobClient:  map 100% reduce 94%
12/07/30 06:14:08 INFO mapred.JobClient:  map 100% reduce 100%

When I see the output of job, I see entries like

1       2
13      2
24      1
29      1
31      2
42      3
6796    7
6799    1
6806    1
6815    1
6824    2

Which means data is getting aggregated

How can I not run reducer at all?

回答1:

That will work only if you implment ToolRunner.run method and pass it arguments in your main method.

ToolRunner.run(new Configuration(), new YourClasImplmentingToolRunner(), args);

If you don't want to do try setting

job.setNumReduceTasks(0);

or the other option is to set the value in conf and use that configuration in job.

Configuration conf = new Configuration();
conf.set("mapred.reduce.tasks", "0");
Job job = new Job(conf, "My job Name");

回答2:

Add a space after the -D option and it should work ;)

hadoop jar Hadoop-programs.jar com/hadoop/patent/CitationHistogram input output -D mapred.reduce.tasks=0

来源：https://stackoverflow.com/questions/11722424/hadoop-reduce-task-running-even-after-telling-on-command-line-as-d-mapred-reduc