Hadoop自动化安装及单节点方式运行

本文尝试使用shell脚本来自动化安装配置Hadoop。使用的操作系统为CentOS，Hadoop版本为?1.x，jdk版本?1.7，其他版本未测试，可能有未知bug。 Hadoop安装脚本 Hadoop安装分为3步，首先安装jdk，然后安装Hadoop，接着配置ssh免密码登陆(非必须)。[1] #!/bin/ba

本文尝试使用shell脚本来自动化安装配置Hadoop。使用的操作系统为CentOS，Hadoop版本为?1.x，jdk版本?1.7，其他版本未测试，可能有未知bug。

Hadoop安装脚本

Hadoop安装分为3步，首先安装jdk，然后安装Hadoop，接着配置ssh免密码登陆(非必须)。[1]

#!/bin/bash
# Usage: Hadoop自动配置脚本
# History: 
#	20140425  annhe  基本功能
#Hadoop版本
HADOOP_VERSION=1.2.1
#Jdk版本，Oracle官方无直链下载，请自备rpm包并设定版本号
JDK_VESION=7u51
#Hadoop下载镜像，默认为北理(bit)
MIRRORS=mirror.bit.edu.cn
#操作系统版本
OS=`uname -a |awk '{print $13}'`
# Check if user is root
if [ $(id -u) != "0" ]; then
    printf "Error: You must be root to run this script!\n"
    exit 1
fi
# 检查是否是Centos
cat /etc/issue|grep CentOS && r=0 || r=1
if [ $r -eq 1 ]; then
	echo "This script can only run on CentOS!"
	exit 1
fi
#软件包
HADOOP_FILE=hadoop-$HADOOP_VERSION-1.$OS.rpm
if [ "$OS"x = "x86_64"x ]; then
	JDK_FILE=jdk-$JDK_VESION-linux-x64.rpm
else
	JDK_FILE=jdk-$JDK_VESION-linux-i586.rpm
fi
function Install ()
{
	#卸载已安装版本
	rpm -qa |grep hadoop
	rpm -e hadoop
	rpm -qa | grep jdk
	rpm -e jdk
	#恢复/etc/profile备份文件
	mv /etc/profile.bak /etc/profile
	#准备软件包
	if [ ! -f $HADOOP_FILE ]; then
		wget "http://$MIRRORS/apache/hadoop/common/stable1/$HADOOP_FILE" && r=0 || r=1
		[ $r -eq 1 ] && { echo "download error, please check your mirrors or check your network....exit"; exit 1; }
	fi
	[ ! -f $JDK_FILE ] && { echo "$JDK_FILE not found! Please download yourself....exit"; exit 1; }
	#开始安装
	rpm -ivh $JDK_FILE && r=0 || r=1
	if [ $r -eq 1 ]; then
		echo "$JDK_FILE install failed, please verify your rpm file....exit"
		exit 1
	fi
	rpm -ivh $HADOOP_FILE && r=0 || r=1
	if [ $r -eq 1 ]; then
		echo "$HADOOP_FILE install failed, please verify your rpm file....exit"
		exit 1
	fi
	#备份/etc/profile
	cp /etc/profile /etc/profile.bak
	#配置java环境变量
	cat >> /etc/profile > ~/.ssh/authorized_keys
	chmod 644 ~/.ssh/authorized_keys
}
Install 2>&1 | tee -a hadoop_install.log
SSHlogin 2>&1 | tee -a hadoop_install.log
#修改HADOOP_CLIENT_OPTS后需要重启 
shutdown -r now

单节点运行自带示例

默认情况下，Hadoop被配置成以非分布式模式运行的一个独立Java进程。这对调试非常有帮助。新建测试文本

[root@linux hadoop]# echo "hello world" >input/hello.txt
[root@linux hadoop]# echo "hello hadoop" >input/hadoop.txt

运行Wordcount

[root@linux hadoop]# hadoop jar /usr/share/hadoop/hadoop-examples-1.2.1.jar wordcount input output
14/04/26 02:56:23 INFO util.NativeCodeLoader: Loaded the native-hadoop library
14/04/26 02:56:23 INFO input.FileInputFormat: Total input paths to process : 2
14/04/26 02:56:24 WARN snappy.LoadSnappy: Snappy native library not loaded
14/04/26 02:56:24 INFO mapred.JobClient: Running job: job_local275273933_0001
14/04/26 02:56:24 INFO mapred.LocalJobRunner: Waiting for map tasks
14/04/26 02:56:24 INFO mapred.LocalJobRunner: Starting task: attempt_local275273933_0001_m_000000_0
14/04/26 02:56:25 INFO util.ProcessTree: setsid exited with exit code 0
14/04/26 02:56:25 INFO mapred.Task:  Using ResourceCalculatorPlugin : org.apache.hadoop.util.LinuxResourceCalculatorPlugin@7e86fe3a
14/04/26 02:56:25 INFO mapred.MapTask: Processing split: file:/root/hadoop/input/hadoop.txt:0+13
14/04/26 02:56:25 INFO mapred.MapTask: io.sort.mb = 100
14/04/26 02:56:25 INFO mapred.MapTask: data buffer = 79691776/99614720
14/04/26 02:56:25 INFO mapred.MapTask: record buffer = 262144/327680
14/04/26 02:56:25 INFO mapred.MapTask: Starting flush of map output
14/04/26 02:56:25 INFO mapred.MapTask: Finished spill 0
14/04/26 02:56:25 INFO mapred.Task: Task:attempt_local275273933_0001_m_000000_0 is done. And is in the process of commiting
14/04/26 02:56:25 INFO mapred.LocalJobRunner:
14/04/26 02:56:25 INFO mapred.Task: Task 'attempt_local275273933_0001_m_000000_0' done.
14/04/26 02:56:25 INFO mapred.LocalJobRunner: Finishing task: attempt_local275273933_0001_m_000000_0
14/04/26 02:56:25 INFO mapred.LocalJobRunner: Starting task: attempt_local275273933_0001_m_000001_0
14/04/26 02:56:25 INFO mapred.Task:  Using ResourceCalculatorPlugin : org.apache.hadoop.util.LinuxResourceCalculatorPlugin@16ed889d
14/04/26 02:56:25 INFO mapred.MapTask: Processing split: file:/root/hadoop/input/hello.txt:0+12
14/04/26 02:56:25 INFO mapred.MapTask: io.sort.mb = 100
14/04/26 02:56:25 INFO mapred.MapTask: data buffer = 79691776/99614720
14/04/26 02:56:25 INFO mapred.MapTask: record buffer = 262144/327680
14/04/26 02:56:25 INFO mapred.MapTask: Starting flush of map output
14/04/26 02:56:25 INFO mapred.MapTask: Finished spill 0
14/04/26 02:56:25 INFO mapred.Task: Task:attempt_local275273933_0001_m_000001_0 is done. And is in the process of commiting
14/04/26 02:56:25 INFO mapred.LocalJobRunner:
14/04/26 02:56:25 INFO mapred.Task: Task 'attempt_local275273933_0001_m_000001_0' done.
14/04/26 02:56:25 INFO mapred.LocalJobRunner: Finishing task: attempt_local275273933_0001_m_000001_0
14/04/26 02:56:25 INFO mapred.LocalJobRunner: Map task executor complete.
14/04/26 02:56:25 INFO mapred.Task:  Using ResourceCalculatorPlugin : org.apache.hadoop.util.LinuxResourceCalculatorPlugin@42701c57
14/04/26 02:56:25 INFO mapred.LocalJobRunner:
14/04/26 02:56:25 INFO mapred.Merger: Merging 2 sorted segments
14/04/26 02:56:25 INFO mapred.Merger: Down to the last merge-pass, with 2 segments left of total size: 53 bytes
14/04/26 02:56:25 INFO mapred.LocalJobRunner:
14/04/26 02:56:25 INFO mapred.Task: Task:attempt_local275273933_0001_r_000000_0 is done. And is in the process of commiting
14/04/26 02:56:25 INFO mapred.LocalJobRunner:
14/04/26 02:56:25 INFO mapred.Task: Task attempt_local275273933_0001_r_000000_0 is allowed to commit now
14/04/26 02:56:25 INFO output.FileOutputCommitter: Saved output of task 'attempt_local275273933_0001_r_000000_0' to output
14/04/26 02:56:25 INFO mapred.LocalJobRunner: reduce > reduce
14/04/26 02:56:25 INFO mapred.Task: Task 'attempt_local275273933_0001_r_000000_0' done.
14/04/26 02:56:25 INFO mapred.JobClient:  map 100% reduce 100%
14/04/26 02:56:25 INFO mapred.JobClient: Job complete: job_local275273933_0001
14/04/26 02:56:25 INFO mapred.JobClient: Counters: 20
14/04/26 02:56:25 INFO mapred.JobClient:   File Output Format Counters
14/04/26 02:56:25 INFO mapred.JobClient:     Bytes Written=37
14/04/26 02:56:25 INFO mapred.JobClient:   FileSystemCounters
14/04/26 02:56:25 INFO mapred.JobClient:     FILE_BYTES_READ=429526
14/04/26 02:56:25 INFO mapred.JobClient:     FILE_BYTES_WRITTEN=586463
14/04/26 02:56:25 INFO mapred.JobClient:   File Input Format Counters
14/04/26 02:56:25 INFO mapred.JobClient:     Bytes Read=25
14/04/26 02:56:25 INFO mapred.JobClient:   Map-Reduce Framework
14/04/26 02:56:25 INFO mapred.JobClient:     Reduce input groups=3
14/04/26 02:56:25 INFO mapred.JobClient:     Map output materialized bytes=61
14/04/26 02:56:25 INFO mapred.JobClient:     Combine output records=4
14/04/26 02:56:25 INFO mapred.JobClient:     Map input records=2
14/04/26 02:56:25 INFO mapred.JobClient:     Reduce shuffle bytes=0
14/04/26 02:56:25 INFO mapred.JobClient:     Physical memory (bytes) snapshot=0
14/04/26 02:56:25 INFO mapred.JobClient:     Reduce output records=3
14/04/26 02:56:25 INFO mapred.JobClient:     Spilled Records=8
14/04/26 02:56:25 INFO mapred.JobClient:     Map output bytes=41
14/04/26 02:56:25 INFO mapred.JobClient:     CPU time spent (ms)=0
14/04/26 02:56:25 INFO mapred.JobClient:     Total committed heap usage (bytes)=480915456
14/04/26 02:56:25 INFO mapred.JobClient:     Virtual memory (bytes) snapshot=0
14/04/26 02:56:25 INFO mapred.JobClient:     Combine input records=4
14/04/26 02:56:25 INFO mapred.JobClient:     Map output records=4
14/04/26 02:56:25 INFO mapred.JobClient:     SPLIT_RAW_BYTES=197
14/04/26 02:56:25 INFO mapred.JobClient:     Reduce input records=

结果

[root@linux hadoop]# cat output/*
hadoop  1
hello   2
world   1

?运行自己编写的Wordcount

package net.annhe.wordcount;
import java.io.IOException;
import java.util.*;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.conf.*;
import org.apache.hadoop.io.*;
import org.apache.hadoop.mapreduce.*;
import org.apache.hadoop.mapreduce.lib.input.*;
import org.apache.hadoop.mapreduce.lib.output.*;
import org.apache.hadoop.util.*;
public class WordCount extends Configured implements Tool {
	public static class Map extends Mapper {
		private final static IntWritable one = new IntWritable(1);
		private Text word = new Text();
		public void map(LongWritable key, Text value, Context context) throws IOException, InterruptedException {
			String line = value.toString();
			StringTokenizer tokenizer = new StringTokenizer(line);
			while (tokenizer.hasMoreTokens()) {
				word.set(tokenizer.nextToken());
				context.write(word,one);
			}
		}
	}
	public static class Reduce extends Reducer {
		public void reduce (Text key, Iterable values, Context context) throws IOException, InterruptedException {
			int sum=0;
			for(IntWritable val : values) {
				sum += val.get();
			}
			context.write(key, new IntWritable(sum));
		}
	}
	public int run(String[] args) throws Exception {
		Job job = new Job(getConf());
		job.setJarByClass(WordCount.class);
		job.setJobName("wordcount");
		job.setOutputKeyClass(Text.class);
		job.setOutputValueClass(IntWritable.class);
		job.setMapperClass(Map.class);
		job.setReducerClass(Reduce.class);
		job.setInputFormatClass(TextInputFormat.class);
		job.setOutputFormatClass(TextOutputFormat.class);
		FileInputFormat.setInputPaths(job, new Path(args[0]));
		FileOutputFormat.setOutputPath(job, new Path(args[1]));
		boolean success = job.waitForCompletion(true);
		return success ? 0 : 1;
	}
	public static void main(String[] args) throws Exception {
		int ret = ToolRunner.run(new WordCount(),args);
		System.exit(ret);
	}
}

编译

javac -classpath /usr/share/hadoop/hadoop-core-1.2.1.jar -d . WordCount.java

打包

jar -vcf wordcount.jar -C demo/ .

运行

hadoop jar wordcount.jar net.annhe.wordcount.WordCount input/ out

结果

[root@linux hadoop]# cat out/*
hadoop  1
hello   2
world   1

?遇到的问题

1. 内存不足

分给虚拟机的内存才180M，运行实例程序时报错：

java.lang.Exception: java.lang.OutOfMemoryError: Java heap space

解决方案：
增加虚拟机内存，并编辑/etc/hadoop/hadoop-env.sh，修改：

export HADOOP_CLIENT_OPTS="-Xmx512m $HADOOP_CLIENT_OPTS" #改成512m

原来启动JVM时配置的最大内存是128m，当运行hadoop的一些自带的实例会报内存溢出，其实这里是可以修改内存大小
如果不需要也不必修改。[2]

?2. 带有包名的类的引用

带有包名的类要按照包层次调用类。如上面的 net.annhe.wordcount.WordCount [3]

3. 带有包名的类的编译

需要打包编译，加-d选项。

java的类文件是应该放入包中的，如package abc;
public class ls {...} 那么这个abc就是就是类ls的包，那么编译的时候就应该创建相应的abc包，具体就是用javac的一个参数，就是这个-d来生成这个类文件的包，例如上面的类在编译时应该写javac -d . ls.java注意javac和-d,-d和后面的．,.和后面的ls.java中间都有空格[4]

参考资料

[1]. 陆嘉桓. Hadoop实战. 第二版. 机械工业出版社

[2]. OSchina博客:http://my.oschina.net/mynote/blog/93340

[3]. CSDN博客：http://blog.csdn.net/xw13106209/article/details/6861855

[4]. 百度知道:http://zhidao.baidu.com/link?url=ND1BWmyGb_5a05Jntd9vGZNWGtmJmcKF1V6dhVNM1eFNuHL6kbQyVrEWtCUmy7KYP5F66R2BumCifCnPQnYdD_

---

本文遵从CC版权协定，转载请以链接形式注明出处。
本文链接地址: http://www.annhe.net/article-2672.html