如何使用eclipse调试Hadoop作业

使用eclipse来调试hadoop作业是非常简洁方便的，散仙以前也有用eclipse开发过hadoop程序，但是一直没有深入了解eclipse调试的一些模式，有些时候也会出一些莫名奇妙的异常，最常见的就是下面这个

java.lang.RuntimeException: java.lang.ClassNotFoundException: com.qin.sort.TestSort$SMapper
	at org.apache.hadoop.conf.Configuration.getClass(Configuration.java:857)
	at org.apache.hadoop.mapreduce.JobContext.getMapperClass(JobContext.java:199)
	at org.apache.hadoop.mapred.MapTask.runNewMapper(MapTask.java:718)
	at org.apache.hadoop.mapred.MapTask.run(MapTask.java:364)
	at org.apache.hadoop.mapred.Child$4.run(Child.java:255)
	at java.security.AccessController.doPrivileged(Native Method)
	at javax.security.auth.Subject.doAs(Subject.java:415)
	at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1190)

这个异常是最莫名其妙的一个，明明自己的MR类里面有这个Mapper的内部类，但是一运行程序，就报这个异常，说找不到这个类，然后就百般查找问题，找来找去，也没找出个所以然。

其实这并不是程序的问题，而是对eclipse的调试模式不够了解的问题，eclipse上运行hadoop总的来说有2种模式，第一种就是Local模式，也叫本地模式，第二种就是我们正式的线上集群模式，当运行本地模式的时候，程序并不会被提交到Hadoop集群上，而是基于单机的模式跑的，但是单机的模式，运行的结果仍在是存储在HDFS上的，只不过没有利用hadoop集群的资源，单机的模式不要提交jar包到hadoop集群上，因此一般我们使用local来测试我们的MR程序是否能够正常运行，
下面我们来看下，基于Local模式跑的一个排序作业：

排序数据：
a 784
b 12
c -11
dd 99999

程序源码：

package com.qin.sort;

import java.io.IOException;

import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.IntWritable;
import org.apache.hadoop.io.LongWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.io.WritableComparator;
import org.apache.hadoop.mapred.JobConf;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.Mapper;
import org.apache.hadoop.mapreduce.Reducer;
import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;
import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;


 
/**
 * 测试排序的
 * MR作业类
 * 
 * QQ技术交流群：324714439
 * @author qindongliang
 * 
 * 
 * **/
public class TestSort {
	
	
	/**
	 * Map类
	 * 
	 * **/
	private static class SMapper extends Mapper<LongWritable, Text, IntWritable, Text>{
		
		private Text text=new Text();//输出
		 private static final IntWritable one=new IntWritable();
		
		@Override
		protected void map(LongWritable key, Text value,Context context)
				throws IOException, InterruptedException {
			String s=value.toString();
		 	//System.out.println("abc: "+s);
		//	if((s.trim().indexOf(" ")!=-1)){
			 String ss[]=s.split(" ");
 			 one.set(Integer.parseInt(ss[1].trim()));//
 			 text.set(ss[0].trim());  
 			 context.write(one, text);
		}
	}
	
	/**
	 * Reduce类
	 *
	 * */
	 private static class SReduce extends Reducer<IntWritable, Text, Text, IntWritable>{
		 private Text text=new Text();
		 @Override
		protected void reduce(IntWritable arg0, Iterable<Text> arg1,Context context)
				throws IOException, InterruptedException {
			 
			 
			 for(Text t:arg1){
				 text.set(t.toString());
				
				 context.write(text, arg0);
			 } 
		}
	 }
	 
	 /**
	  * 排序的类
	  * 
	  * **/
	 private static class SSort extends WritableComparator{
		 
		 public SSort() {
			 super(IntWritable.class,true);//注册排序组件
		}
		 @Override
		public int compare(byte[] arg0, int arg1, int arg2, byte[] arg3,
				int arg4, int arg5) {
			/**
			 * 控制升降序的关键控制-号是降序
			 * */
			return -super.compare(arg0, arg1, arg2, arg3, arg4, arg5);//注意使用负号来完成降序
		}
		 
		 @Override
		public int compare(Object a, Object b) {
	 
			return    -super.compare(a, b);//注意使用负号来完成降序
		}
		 
		 
	 }
	
	 /**
	  * main方法
	  * */
	 public static void main(String[] args) throws Exception{
		 String inputPath="hdfs://192.168.75.130:9000/root/output";	    
		  String outputPath="hdfs://192.168.75.130:9000/root/outputsort";
		  JobConf conf=new JobConf();
		//Configuration conf=new Configuration();
		   //在你的文件地址前自动添加：hdfs://master:9000/
		 // conf.set("fs.default.name", "hdfs://192.168.75.130:9000");
		  //指定jobtracker的ip和端口号，master在/etc/hosts中可以配置
		//  conf.set("mapred.job.tracker","192.168.75.130:9001");
		 // conf.get("mapred.job.tracker");
		 System.out.println("模式：  "+conf.get("mapred.job.tracker"));
		//  conf.setJar("tt.jar");
		  FileSystem  fs=FileSystem.get(conf);
		  Path pout=new Path(outputPath);
		  if(fs.exists(pout)){
			  fs.delete(pout, true);
			  System.out.println("存在此路径, 已经删除......");
		  } 		  
		  Job job=new Job(conf, "sort123"); 
          job.setJarByClass(TestSort.class);
          job.setOutputKeyClass(IntWritable.class);//告诉map，reduce输出K,V的类型
          FileInputFormat.setInputPaths(job, new Path(inputPath));  //输入路径
          FileOutputFormat.setOutputPath(job, new Path(outputPath));//输出路径  
          job.setMapperClass(SMapper.class);//map类
          job.setReducerClass(SReduce.class);//reduce类
          job.setSortComparatorClass(SSort.class);//排序类
//          job.setInputFormatClass(org.apache.hadoop.mapreduce.lib.input.TextInputFormat.class);
//		  job.setOutputFormatClass(TextOutputFormat.class);
          System.exit(job.waitForCompletion(true) ? 0 : 1);  
		 
		 
	}
	

}

打印结果如下：

模式：  local
存在此路径, 已经删除......
WARN - NativeCodeLoader.<clinit>(52) | Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
WARN - JobClient.copyAndConfigureFiles(746) | Use GenericOptionsParser for parsing the arguments. Applications should implement Tool for the same.
WARN - JobClient.copyAndConfigureFiles(870) | No job jar file set.  User classes may not be found. See JobConf(Class) or JobConf#setJar(String).
INFO - FileInputFormat.listStatus(237) | Total input paths to process : 1
WARN - LoadSnappy.<clinit>(46) | Snappy native library not loaded
INFO - JobClient.monitorAndPrintJob(1380) | Running job: job_local1242054158_0001
INFO - LocalJobRunner$Job.run(340) | Waiting for map tasks
INFO - LocalJobRunner$Job$MapTaskRunnable.run(204) | Starting task: attempt_local1242054158_0001_m_000000_0
INFO - Task.initialize(534) |  Using ResourceCalculatorPlugin : null
INFO - MapTask.runNewMapper(729) | Processing split: hdfs://192.168.75.130:9000/root/output/sort.txt:0+28
INFO - MapTask$MapOutputBuffer.<init>(949) | io.sort.mb = 100
INFO - MapTask$MapOutputBuffer.<init>(961) | data buffer = 79691776/99614720
INFO - MapTask$MapOutputBuffer.<init>(962) | record buffer = 262144/327680
INFO - MapTask$MapOutputBuffer.flush(1289) | Starting flush of map output
INFO - MapTask$MapOutputBuffer.sortAndSpill(1471) | Finished spill 0
INFO - Task.done(858) | Task:attempt_local1242054158_0001_m_000000_0 is done. And is in the process of commiting
INFO - LocalJobRunner$Job.statusUpdate(466) | 
INFO - Task.sendDone(970) | Task 'attempt_local1242054158_0001_m_000000_0' done.
INFO - LocalJobRunner$Job$MapTaskRunnable.run(229) | Finishing task: attempt_local1242054158_0001_m_000000_0
INFO - LocalJobRunner$Job.run(348) | Map task executor complete.
INFO - Task.initialize(534) |  Using ResourceCalculatorPlugin : null
INFO - LocalJobRunner$Job.statusUpdate(466) | 
INFO - Merger$MergeQueue.merge(408) | Merging 1 sorted segments
INFO - Merger$MergeQueue.merge(491) | Down to the last merge-pass, with 1 segments left of total size: 35 bytes
INFO - LocalJobRunner$Job.statusUpdate(466) | 
INFO - Task.done(858) | Task:attempt_local1242054158_0001_r_000000_0 is done. And is in the process of commiting
INFO - LocalJobRunner$Job.statusUpdate(466) | 
INFO - Task.commit(1011) | Task attempt_local1242054158_0001_r_000000_0 is allowed to commit now
INFO - FileOutputCommitter.commitTask(173) | Saved output of task 'attempt_local1242054158_0001_r_000000_0' to hdfs://192.168.75.130:9000/root/outputsort
INFO - LocalJobRunner$Job.statusUpdate(466) | reduce > reduce
INFO - Task.sendDone(970) | Task 'attempt_local1242054158_0001_r_000000_0' done.
INFO - JobClient.monitorAndPrintJob(1393) |  map 100% reduce 100%
INFO - JobClient.monitorAndPrintJob(1448) | Job complete: job_local1242054158_0001
INFO - Counters.log(585) | Counters: 19
INFO - Counters.log(587) |   File Output Format Counters 
INFO - Counters.log(589) |     Bytes Written=26
INFO - Counters.log(587) |   File Input Format Counters 
INFO - Counters.log(589) |     Bytes Read=28
INFO - Counters.log(587) |   FileSystemCounters
INFO - Counters.log(589) |     FILE_BYTES_READ=393
INFO - Counters.log(589) |     HDFS_BYTES_READ=56
INFO - Counters.log(589) |     FILE_BYTES_WRITTEN=135742
INFO - Counters.log(589) |     HDFS_BYTES_WRITTEN=26
INFO - Counters.log(587) |   Map-Reduce Framework
INFO - Counters.log(589) |     Map output materialized bytes=39
INFO - Counters.log(589) |     Map input records=4
INFO - Counters.log(589) |     Reduce shuffle bytes=0
INFO - Counters.log(589) |     Spilled Records=8
INFO - Counters.log(589) |     Map output bytes=25
INFO - Counters.log(589) |     Total committed heap usage (bytes)=455475200
INFO - Counters.log(589) |     Combine input records=0
INFO - Counters.log(589) |     SPLIT_RAW_BYTES=112
INFO - Counters.log(589) |     Reduce input records=4
INFO - Counters.log(589) |     Reduce input groups=4
INFO - Counters.log(589) |     Combine output records=0
INFO - Counters.log(589) |     Reduce output records=4
INFO - Counters.log(589) |     Map output records=4

排序结果如下:

dd	99999
a	784
b	12
c	-11

单机模式调试通过之后，我们就可以考虑采用hadoop集群的模式来跑，这时候有2种方式，可以来完成这件事，第一是，为了方便将整个项目打成一个jar包，上传到Linux上，然后执行shell命令：
bin/hadoop jar tt.jar com.qin.sort.TestSort
来进行测试，注意，散仙是为了方便，路径是写死在程序里面所以后面不用输入，输入和输出路径，正式的开发，为了灵活性，一般会通过外部传产来指定输入和输出路径。

第二种方式，也比较方便，直接在eclipse中提交到hadoop集群作业中，不过即使是使用eclipse来提交作业，还是需要将整个项目打成一个jar包，只不过这时是eclipse帮我们提交作业的，这样我们就可以Win平台上直接提交运行hadoop作业了，但是主流的还是使用上传jar包的方式。关于把整个项目打成一个jar包，散仙在后面会上传一个ant脚本，直接执行它就可以了，这样就可以把有依赖关系的类打在一起，把一整个项目做为一个整体，在hadoop上，只需要指定jar，指定类的全名称，和输入，输出路径即可。ant的脚本内容如下：

<project name="${component.name}" basedir="." default="jar">
	<property environment="env"/>
	<!--
	<property name="hadoop.home" value="${env.HADOOP_HOME}"/>
	-->
	<property name="hadoop.home" value="D:/hadoop-1.2.0"/>
	<!-- 指定jar包的名字 -->
	<property name="jar.name" value="tt.jar"/>
	<path id="project.classpath">
		<fileset dir="lib">
			<include name="*.jar" />
		</fileset>
		<fileset dir="${hadoop.home}">
			<include name="**/*.jar" />
		</fileset>
	</path>
	<target name="clean" >
	 	<delete dir="bin" failonerror="false" />
		<mkdir dir="bin"/>
 	</target>	
	<target name="build" depends="clean">
 		<echo message="${ant.project.name}: ${ant.file}"/>
 		<javac destdir="bin" encoding="utf-8" debug="true" includeantruntime="false" debuglevel="lines,vars,source">
            <src path="src"/>
 			<exclude name="**/.svn" />
            <classpath refid="project.classpath"/>
        </javac>
		<copy todir="bin">
			<fileset dir="src">
				<include name="*config*"/>
			</fileset>
		</copy>
 	</target>
	
 	<target name="jar" depends="build">
 		<copy todir="bin/lib">
 			<fileset dir="lib">
 				<include name="**/*.*"/>
	 		</fileset>
 		</copy>
 		
 		<path id="lib-classpath">
 			<fileset dir="lib" includes="**/*.jar" />
 		</path>
 		
 		<pathconvert property="my.classpath" pathsep=" " >
 			<mapper>
 		    	<chainedmapper>
 		        	<!-- 移除绝对路径 -->
 		        	<flattenmapper />
 		        	<!-- 加上lib前缀 -->
 		        	<globmapper from="*" to="lib/*" />
 		       </chainedmapper>
 		     </mapper>
 		     <path refid="lib-classpath" />
 		</pathconvert>
 		
 		<jar basedir="bin" destfile="${jar.name}" >
 			<include name="**/*"/>
 			<!-- define MANIFEST.MF -->
 			<manifest>
				<attribute name="Class-Path" value="${my.classpath}" />
 			</manifest>
 		</jar>
 	</target>
</project>

运行上面的这个ant脚本之后，我们的项目就会被打成一个jar包，截图如下：


jar包有了之后，我们先测试在eclipse上如何把作业提交到hadoop集群上，只要把main方面的代码，稍加改动即可：

 /**
	  * main方法
	  * */
	 public static void main(String[] args) throws Exception{
		 String inputPath="hdfs://192.168.75.130:9000/root/output";	    
		  String outputPath="hdfs://192.168.75.130:9000/root/outputsort";
		  JobConf conf=new JobConf();
		 //Configuration conf=new Configuration();//可以使用这个conf来测试Local模式
		 //如果在src目录下有，mapred-site.xml文件，就不要此行代码
		 //注意此行代码也是在非Local模式下才使用
		 conf.set("mapred.job.tracker","192.168.75.130:9001");
		 // conf.get("mapred.job.tracker");
		 System.out.println("模式：  "+conf.get("mapred.job.tracker"));
		 // conf.setJar("tt.jar"); 非Local模式下使用
		  FileSystem  fs=FileSystem.get(conf);
		  Path pout=new Path(outputPath);
		  if(fs.exists(pout)){
			  fs.delete(pout, true);
			  System.out.println("存在此路径, 已经删除......");
		  } 		  
		  Job job=new Job(conf, "sort123"); 
          job.setJarByClass(TestSort.class);
          job.setOutputKeyClass(IntWritable.class);//告诉map，reduce输出K,V的类型
          FileInputFormat.setInputPaths(job, new Path(inputPath));  //输入路径
          FileOutputFormat.setOutputPath(job, new Path(outputPath));//输出路径  
          job.setMapperClass(SMapper.class);//map类
          job.setReducerClass(SReduce.class);//reduce类
          job.setSortComparatorClass(SSort.class);//排序类
//          job.setInputFormatClass(org.apache.hadoop.mapreduce.lib.input.TextInputFormat.class);
//		  job.setOutputFormatClass(TextOutputFormat.class);
          System.exit(job.waitForCompletion(true) ? 0 : 1);  
		 
		 
	}

运行程序，输出，如下：

模式：  192.168.75.130:9001
存在此路径, 已经删除......
WARN - JobClient.copyAndConfigureFiles(746) | Use GenericOptionsParser for parsing the arguments. Applications should implement Tool for the same.
INFO - FileInputFormat.listStatus(237) | Total input paths to process : 1
WARN - NativeCodeLoader.<clinit>(52) | Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
WARN - LoadSnappy.<clinit>(46) | Snappy native library not loaded
INFO - JobClient.monitorAndPrintJob(1380) | Running job: job_201403252058_0035
INFO - JobClient.monitorAndPrintJob(1393) |  map 0% reduce 0%
INFO - JobClient.monitorAndPrintJob(1393) |  map 100% reduce 0%
INFO - JobClient.monitorAndPrintJob(1393) |  map 100% reduce 33%
INFO - JobClient.monitorAndPrintJob(1393) |  map 100% reduce 100%
INFO - JobClient.monitorAndPrintJob(1448) | Job complete: job_201403252058_0035
INFO - Counters.log(585) | Counters: 29
INFO - Counters.log(587) |   Job Counters 
INFO - Counters.log(589) |     Launched reduce tasks=1
INFO - Counters.log(589) |     SLOTS_MILLIS_MAPS=8498
INFO - Counters.log(589) |     Total time spent by all reduces waiting after reserving slots (ms)=0
INFO - Counters.log(589) |     Total time spent by all maps waiting after reserving slots (ms)=0
INFO - Counters.log(589) |     Launched map tasks=1
INFO - Counters.log(589) |     Data-local map tasks=1
INFO - Counters.log(589) |     SLOTS_MILLIS_REDUCES=9667
INFO - Counters.log(587) |   File Output Format Counters 
INFO - Counters.log(589) |     Bytes Written=26
INFO - Counters.log(587) |   FileSystemCounters
INFO - Counters.log(589) |     FILE_BYTES_READ=39
INFO - Counters.log(589) |     HDFS_BYTES_READ=140
INFO - Counters.log(589) |     FILE_BYTES_WRITTEN=117654
INFO - Counters.log(589) |     HDFS_BYTES_WRITTEN=26
INFO - Counters.log(587) |   File Input Format Counters 
INFO - Counters.log(589) |     Bytes Read=28
INFO - Counters.log(587) |   Map-Reduce Framework
INFO - Counters.log(589) |     Map output materialized bytes=39
INFO - Counters.log(589) |     Map input records=4
INFO - Counters.log(589) |     Reduce shuffle bytes=39
INFO - Counters.log(589) |     Spilled Records=8
INFO - Counters.log(589) |     Map output bytes=25
INFO - Counters.log(589) |     Total committed heap usage (bytes)=176033792
INFO - Counters.log(589) |     CPU time spent (ms)=1140
INFO - Counters.log(589) |     Combine input records=0
INFO - Counters.log(589) |     SPLIT_RAW_BYTES=112
INFO - Counters.log(589) |     Reduce input records=4
INFO - Counters.log(589) |     Reduce input groups=4
INFO - Counters.log(589) |     Combine output records=0
INFO - Counters.log(589) |     Physical memory (bytes) snapshot=259264512
INFO - Counters.log(589) |     Reduce output records=4
INFO - Counters.log(589) |     Virtual memory (bytes) snapshot=1460555776
INFO - Counters.log(589) |     Map output records=4

我们可以看出，运行正常，排序的内容如下：

dd	99999
a	784
b	12
c	-11

结果和local模式下的一样，还有一个与local模式不同的地方是，我们可以在http://192.168.75.130:50030/jobtracker.jsp的任务页面上找到刚才执行的任务状况，这一点在Local模式下运行程序，是没有的。/size]

[size=large]最后，散仙再来看下，如何将jar包，上传到Linux提交作业到hadoop集群上。刚才，我们已经把jar给打好了，现在只需上传到linux上即可：

然后开始执行shell命令运行程序：



到此，我们已经完美的执行成功，最后一点需要注意的是，在执行排序任务时，如果出现异常：

java.lang.Exception: java.io.IOException: Type mismatch in key from map: expected org.apache.hadoop.io.LongWritable, recieved org.apache.hadoop.io.IntWritable
	at org.apache.hadoop.mapred.LocalJobRunner$Job.run(LocalJobRunner.java:354)
Caused by: java.io.IOException: Type mismatch in key from map: expected org.apache.hadoop.io.LongWritable, recieved org.apache.hadoop.io.IntWritable
	at org.apache.hadoop.mapred.MapTask$MapOutputBuffer.collect(MapTask.java:1019)
	at org.apache.hadoop.mapred.MapTask$NewOutputCollector.write(MapTask.java:690)
	at org.apache.hadoop.mapreduce.TaskInputOutputContext.write(TaskInputOutputContext.java:80)
	at com.qin.sort.TestSort$SMapper.map(TestSort.java:51)
	at com.qin.sort.TestSort$SMapper.map(TestSort.java:1)
	at org.apache.hadoop.mapreduce.Mapper.run(Mapper.java:145)
	at org.apache.hadoop.mapred.MapTask.runNewMapper(MapTask.java:764)
	at org.apache.hadoop.mapred.MapTask.run(MapTask.java:364)
	at org.apache.hadoop.mapred.LocalJobRunner$Job$MapTaskRunnable.run(LocalJobRunner.java:223)
	at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:471)
	at java.util.concurrent.FutureTask$Sync.innerRun(FutureTask.java:334)
	at java.util.concurrent.FutureTask.run(FutureTask.java:166)
	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1110)
	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:603)
	at java.lang.Thread.run(Thread.java:722)

这个异常的出现，多半是因为，我们没有指定输出的Key，或者Value，或者指定的类型不一致，导致，我们只需要正确的设置输出的Key或者Value的类型即可.

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

设置完后，就可以正常测试运行了。