在CentOS 运行你的第一个MapReduce程序

在进行本文的操作之前要先搭建一个Hadoop的环境，为了便于实验，可采用单节点部署的方式，具体方法可参见： Centos 6.5 下Hadoop 1.2.1单节点环境的创建

• 编写源码

  主要为创建一个解析气象数据的程序，可以从数据文件中选择气温最高的一年，采用Maven进行编译。下面只包含Maper,Reduce,以及Main函数的代码。完整项目代码请参见
  https://github.com/Eric-aihua/practise/tree/master/hadoop
  

• Mapper
  
  

  package com.eric.hadoop.map;
  
  import java.io.IOException;
  
  import org.apache.hadoop.io.IntWritable;
  import org.apache.hadoop.io.LongWritable;
  import org.apache.hadoop.io.Text;
  import org.apache.hadoop.mapred.MapReduceBase;
  import org.apache.hadoop.mapred.Mapper;
  import org.apache.hadoop.mapred.OutputCollector;
  import org.apache.hadoop.mapred.Reporter;
  
  public class MaxTemperatureMapper extends MapReduceBase implements
      Mapper {
  
    private static final int MISSING = 9999;
  
    public void map(LongWritable fileOffset, Text lineRecord,
        OutputCollector output, Reporter reporter) throws IOException {
      System.out.println("##Processing Record:" + lineRecord.toString());
      String line = lineRecord.toString();
      String year = line.substring(15, 19);
      int temperature;
      if (line.charAt(87) == '+') {
        temperature = Integer.parseInt(line.substring(88, 92));
      } else {
        temperature = Integer.parseInt(line.substring(87, 92));
      }
      String quality = line.substring(92, 93);
      if (temperature != MISSING && quality.matches("[01459]")) {
        output.collect(new Text(year), new IntWritable(temperature));
      }
    }
  
  }
  

• Reduce
  

  package com.eric.hadoop.reduce;
  
  import java.io.IOException;
  import java.util.Iterator;
  
  import org.apache.hadoop.io.IntWritable;
  import org.apache.hadoop.io.Text;
  import org.apache.hadoop.mapred.MapReduceBase;
  import org.apache.hadoop.mapred.OutputCollector;
  import org.apache.hadoop.mapred.Reducer;
  import org.apache.hadoop.mapred.Reporter;
  
  public class MaxTemperatureReduce extends MapReduceBase implements
      Reducer {
  
    public void reduce(Text year, Iterator temperatures,
        OutputCollector output, Reporter arg3) throws IOException {
      int maxTemperature = Integer.MIN_VALUE;
      System.out.println("##Processing temperatures:" + temperatures);
      while (temperatures.hasNext()) {
        maxTemperature = Math.max(maxTemperature, temperatures.next().get());
      }
      output.collect(year, new IntWritable(maxTemperature));
    }
  
  }
  

• Main
  

  package com.eric.hadoop.jobconfig;
  
  import java.io.IOException;
  
  import org.apache.hadoop.fs.Path;
  import org.apache.hadoop.io.IntWritable;
  import org.apache.hadoop.io.Text;
  import org.apache.hadoop.mapred.FileInputFormat;
  import org.apache.hadoop.mapred.FileOutputFormat;
  import org.apache.hadoop.mapred.JobClient;
  import org.apache.hadoop.mapred.JobConf;
  
  import com.eric.hadoop.map.MaxTemperatureMapper;
  import com.eric.hadoop.reduce.MaxTemperatureReduce;
  
  
  public class MaxTemperature {
    public static void main(String[] args) throws IOException {
      JobConf conf = new JobConf(MaxTemperature.class);
      conf.setJobName("Get Max Temperature!");
      if (args.length != 2) {
        System.err.print("Must contain 2 params:inputPath OutputPath");
        System.exit(0);
      }
  
      FileInputFormat.addInputPaths(conf, args[0]);
      FileOutputFormat.setOutputPath(conf, new Path(args[1]));
  
      conf.setMapperClass(MaxTemperatureMapper.class);
      conf.setReducerClass(MaxTemperatureReduce.class);
  
      conf.setOutputKeyClass(Text.class);
      conf.setOutputValueClass(IntWritable.class);
      JobClient.runJob(conf);
    }
  }
  

• 生成Jar文件

  进入项目目录，执行
  mvn install
  成功执行后生成名称为hadoop-0.0.1-SNAPSHOT.jar的Jar文件

• 获取测试数据

  可以使用上文中github中的数据，也可从互联网上下载，URL为：https://github.com/tomwhite/hadoop-book/tree/master/input/ncdc/all
  假设下载的数据文件名称为1902,且放到HDFS文件系统的testdata目录
  hadoop dfs -mkdir testdata
  hadoop dfs -mkdir output
  hadoop dfs -put 1902 testdata

• 执行Job

  hadoop jar hadoop-0.0.1-SNAPSHOT.jar testdata/1902 output

• 观察结果

  通过WEB控制台来监控：
  
  通过命令行输出来监控：
  [hadoop@localhost ~]$ hadoop jar hadoop-0.0.1-SNAPSHOT.jar testdata/1902 output 
  Warning: $HADOOP_HOME is deprecated. 
  
  14/11/26 13:33:39 WARN mapred.JobClient: Use GenericOptionsParser for parsing the arguments. Applications should implement Tool for the same. 
  14/11/26 13:33:39 INFO util.NativeCodeLoader: Loaded the native-hadoop library 
  14/11/26 13:33:39 WARN snappy.LoadSnappy: Snappy native library not loaded 
  14/11/26 13:33:39 INFO mapred.FileInputFormat: Total input paths to process : 1 
  14/11/26 13:33:40 INFO mapred.JobClient: Running job: job_201411261331_0002 #job的标识
  14/11/26 13:33:41 INFO mapred.JobClient: map 0% reduce 0% 
  14/11/26 13:33:47 INFO mapred.JobClient: map 100% reduce 0% #Mapper的进度
  14/11/26 13:33:54 INFO mapred.JobClient: map 100% reduce 33% 
  14/11/26 13:33:56 INFO mapred.JobClient: map 100% reduce 100%#Reduce的进度
  14/11/26 13:33:57 INFO mapred.JobClient: Job complete: job_201411261331_0002 
  14/11/26 13:33:57 INFO mapred.JobClient: Counters: 30 
  14/11/26 13:33:57 INFO mapred.JobClient: Job Counters 
  14/11/26 13:33:57 INFO mapred.JobClient: Launched reduce tasks=1 
  14/11/26 13:33:57 INFO mapred.JobClient: SLOTS_MILLIS_MAPS=7744 
  14/11/26 13:33:57 INFO mapred.JobClient: Total time spent by all reduces waiting after reserving slots (ms)=0 
  14/11/26 13:33:57 INFO mapred.JobClient: Total time spent by all maps waiting after reserving slots (ms)=0 
  14/11/26 13:33:57 INFO mapred.JobClient: Launched map tasks=2 
  14/11/26 13:33:57 INFO mapred.JobClient: Data-local map tasks=2 
  14/11/26 13:33:57 INFO mapred.JobClient: SLOTS_MILLIS_REDUCES=9008 
  14/11/26 13:33:57 INFO mapred.JobClient: File Input Format Counters 
  14/11/26 13:33:57 INFO mapred.JobClient: Bytes Read=890953 
  14/11/26 13:33:57 INFO mapred.JobClient: File Output Format Counters 
  14/11/26 13:33:57 INFO mapred.JobClient: Bytes Written=9 
  14/11/26 13:33:57 INFO mapred.JobClient: FileSystemCounters 
  14/11/26 13:33:57 INFO mapred.JobClient: FILE_BYTES_READ=72221 
  14/11/26 13:33:57 INFO mapred.JobClient: HDFS_BYTES_READ=891143 
  14/11/26 13:33:57 INFO mapred.JobClient: FILE_BYTES_WRITTEN=309368 
  14/11/26 13:33:57 INFO mapred.JobClient: HDFS_BYTES_WRITTEN=9 
  14/11/26 13:33:57 INFO mapred.JobClient: Map-Reduce Framework 
  14/11/26 13:33:57 INFO mapred.JobClient: Map output materialized bytes=72227 
  14/11/26 13:33:57 INFO mapred.JobClient: Map input records=6565 #Mapper的输入记录数
  14/11/26 13:33:57 INFO mapred.JobClient: Reduce shuffle bytes=72227 
  14/11/26 13:33:57 INFO mapred.JobClient: Spilled Records=13130 
  14/11/26 13:33:57 INFO mapred.JobClient: Map output bytes=59085 
  14/11/26 13:33:57 INFO mapred.JobClient: Total committed heap usage (bytes)=478543872 
  14/11/26 13:33:57 INFO mapred.JobClient: CPU time spent (ms)=4400 #CPU耗时
  14/11/26 13:33:57 INFO mapred.JobClient: Map input bytes=888978 
  14/11/26 13:33:57 INFO mapred.JobClient: SPLIT_RAW_BYTES=190 
  14/11/26 13:33:57 INFO mapred.JobClient: Combine input records=0 
  14/11/26 13:33:57 INFO mapred.JobClient: Reduce input records=6565 #Reduce的输出记录数
  14/11/26 13:33:57 INFO mapred.JobClient: Reduce input groups=1 
  14/11/26 13:33:57 INFO mapred.JobClient: Combine output records=0 
  14/11/26 13:33:57 INFO mapred.JobClient: Physical memory (bytes) snapshot=501690368 
  14/11/26 13:33:57 INFO mapred.JobClient: Reduce output records=1 #Reduce的输出记录数
  14/11/26 13:33:57 INFO mapred.JobClient: Virtual memory (bytes) snapshot=2167922688 
  14/11/26 13:33:57 INFO mapred.JobClient: Map output records=6565#Mapper的输出记录数
  

• 检查运行结果

  
  

• 故障以及解析

  问题描述：hadoop 的map阶段正常，但是reduce却卡在00%那里，等了好久进度仍然不变
  
  日志报错：2011-10-03 09:46:13,349 INFO org.apache.hadoop.mapred.JobInProgress: Failed fetch notification #1 for task attempt_201110022127_0003_m_000000_0

  1. 将/etc/hosts中的主机名与/etc/sysconfig/network中的HOSTNAME一致，修改对应的文件后重启系统