Hadoop伪分布式运行

Hadoop可以在单节点上以所谓的伪分布式模式运行，此时每一个Hadoop守护进程都作为一个独立的Java进程运行。本文通过自动化脚本配置Hadoop伪分布式模式。测试环境为VMware中的Centos 6.3, Hadoop 1.2.1.其他版本未测试。 伪分布式配置脚本 包括配置core-site.

Hadoop可以在单节点上以所谓的伪分布式模式运行，此时每一个Hadoop守护进程都作为一个独立的Java进程运行。本文通过自动化脚本配置Hadoop伪分布式模式。测试环境为VMware中的Centos 6.3, Hadoop 1.2.1.其他版本未测试。

伪分布式配置脚本

包括配置core-site.xml，hdfs-site.xml及mapred-site.xml，配置ssh免密码登陆。[1]

#!/bin/bash
# Usage: Hadoop伪分布式配置
# History:
#	20140426  annhe  完成基本功能
# Check if user is root
if [ $(id -u) != "0" ]; then
    printf "Error: You must be root to run this script!\n"
    exit 1
fi
#同步时钟
rm -rf /etc/localtime
ln -s /usr/share/zoneinfo/Asia/Shanghai /etc/localtime
#yum install -y ntp
ntpdate -u pool.ntp.org &>/dev/null
echo -e "Time: `date` \n"
#默认为单网卡结构，多网卡的暂不考虑
IP=`ifconfig eth0 |grep "inet\ addr" |awk '{print $2}' |cut -d ":" -f2`
#伪分布式
function PseudoDistributed ()
{
	cd /etc/hadoop/
	#恢复备份
	mv core-site.xml.bak core-site.xml
	mv hdfs-site.xml.bak hdfs-site.xml
	mv mapred-site.xml.bak mapred-site.xml
	#备份
	mv core-site.xml core-site.xml.bak
	mv hdfs-site.xml hdfs-site.xml.bak
	mv mapred-site.xml mapred-site.xml.bak
	#使用下面的core-site.xml
	cat > core-site.xml 
<?xml-stylesheet type="text/xsl" href="http://www.annhe.net/configuration.xsl"?>
<!-- Put site-specific property overrides in this file. -->
<configuration>
	<property>
		<name>fs.default.name</name>
		<value>hdfs://$IP:9000</value>
	</property>
</configuration>
eof
	#使用下面的hdfs-site.xml
	cat > hdfs-site.xml 
<?xml-stylesheet type="text/xsl" href="http://www.annhe.net/configuration.xsl"?>
<!-- Put site-specific property overrides in this file. -->
<configuration>
	<property>
		<name>dfs.replication</name>
		<value>1</value>
	</property>
</configuration>	
eof
	#使用下面的mapred-site.xml
	cat > mapred-site.xml 
<?xml-stylesheet type="text/xsl" href="http://www.annhe.net/configuration.xsl"?>
<!-- Put site-specific property overrides in this file. -->
<configuration>
	<property>
		<name>mapred.job.tracker</name>
		<value>$IP:9001</value>
	</property>
</configuration>
eof
}
#配置ssh免密码登陆
function PassphraselessSSH ()
{
	#不重复生成私钥
	[ ! -f ~/.ssh/id_dsa ] && ssh-keygen -t dsa -P '' -f ~/.ssh/id_dsa
	cat ~/.ssh/authorized_keys |grep "`cat ~/.ssh/id_dsa.pub`" &>/dev/null && r=0 || r=1
	#没有公钥的时候才添加
	[ $r -eq 1 ] && cat ~/.ssh/id_dsa.pub >> ~/.ssh/authorized_keys
	chmod 644 ~/.ssh/authorized_keys
}
#执行
function Execute ()
{
	#格式化一个新的分布式文件系统
	hadoop namenode -format
	#启动Hadoop守护进程
	start-all.sh
	echo -e "\n========================================================================"
	echo "hadoop log dir : $HADOOP_LOG_DIR"
	echo "NameNode - http://$IP:50070/"
	echo "JobTracker - http://$IP:50030/"
	echo -e "\n========================================================================="
}
PseudoDistributed 2>&1 | tee -a pseudo.log
PassphraselessSSH 2>&1 | tee -a pseudo.log
Execute 2>&1 | tee -a pseudo.log

脚本测试结果

[root@hadoop hadoop]# ./pseudo.sh
14/04/26 23:52:30 INFO namenode.NameNode: STARTUP_MSG:
/************************************************************
STARTUP_MSG: Starting NameNode
STARTUP_MSG:   host = hadoop/216.34.94.184
STARTUP_MSG:   args = [-format]
STARTUP_MSG:   version = 1.2.1
STARTUP_MSG:   build = https://svn.apache.org/repos/asf/hadoop/common/branches/branch-1.2 -r 1503152; compiled by 'mattf' on Mon Jul 22 15:27:42 PDT 2013
STARTUP_MSG:   java = 1.7.0_51
************************************************************/
Re-format filesystem in /tmp/hadoop-root/dfs/name ? (Y or N) y
Format aborted in /tmp/hadoop-root/dfs/name
14/04/26 23:52:40 INFO namenode.NameNode: SHUTDOWN_MSG:
/************************************************************
SHUTDOWN_MSG: Shutting down NameNode at hadoop/216.34.94.184
************************************************************/
starting namenode, logging to /var/log/hadoop/root/hadoop-root-namenode-hadoop.out
localhost: starting datanode, logging to /var/log/hadoop/root/hadoop-root-datanode-hadoop.out
localhost: starting secondarynamenode, logging to /var/log/hadoop/root/hadoop-root-secondarynamenode-hadoop.out
starting jobtracker, logging to /var/log/hadoop/root/hadoop-root-jobtracker-hadoop.out
localhost: starting tasktracker, logging to /var/log/hadoop/root/hadoop-root-tasktracker-hadoop.out
========================================================================
hadoop log dir : /var/log/hadoop/root
NameNode - http://192.168.60.128:50070/
JobTracker - http://192.168.60.128:50030/
=========================================================================

通过宿主机上的浏览器访问NameNode和JobTracker的网络接口

浏览器访问namenode的网络接口

浏览器访问jobtracker网络接口

运行测试程序

将输入文件拷贝到分布式文件系统：

$ hadoop fs -put input input

通过网络接口查看hdfs

通过NameNode网络接口查看hdfs文件系统

运行示例程序

[root@hadoop hadoop]# hadoop jar /usr/share/hadoop/hadoop-examples-1.2.1.jar wordcount input output

通过JobTracker网络接口查看执行状态

Wordcount执行状态

执行结果

[root@hadoop hadoop]# hadoop jar /usr/share/hadoop/hadoop-examples-1.2.1.jar wordcount input out2
14/04/27 03:34:56 INFO input.FileInputFormat: Total input paths to process : 2
14/04/27 03:34:56 INFO util.NativeCodeLoader: Loaded the native-hadoop library
14/04/27 03:34:56 WARN snappy.LoadSnappy: Snappy native library not loaded
14/04/27 03:34:57 INFO mapred.JobClient: Running job: job_201404270333_0001
14/04/27 03:34:58 INFO mapred.JobClient:  map 0% reduce 0%
14/04/27 03:35:49 INFO mapred.JobClient:  map 100% reduce 0%
14/04/27 03:36:16 INFO mapred.JobClient:  map 100% reduce 100%
14/04/27 03:36:19 INFO mapred.JobClient: Job complete: job_201404270333_0001
14/04/27 03:36:19 INFO mapred.JobClient: Counters: 29
14/04/27 03:36:19 INFO mapred.JobClient:   Job Counters
14/04/27 03:36:19 INFO mapred.JobClient:     Launched reduce tasks=1
14/04/27 03:36:19 INFO mapred.JobClient:     SLOTS_MILLIS_MAPS=72895
14/04/27 03:36:19 INFO mapred.JobClient:     Total time spent by all reduces waiting after reserving slots (ms)=0
14/04/27 03:36:19 INFO mapred.JobClient:     Total time spent by all maps waiting after reserving slots (ms)=0
14/04/27 03:36:19 INFO mapred.JobClient:     Launched map tasks=2
14/04/27 03:36:19 INFO mapred.JobClient:     Data-local map tasks=2
14/04/27 03:36:19 INFO mapred.JobClient:     SLOTS_MILLIS_REDUCES=24880
14/04/27 03:36:19 INFO mapred.JobClient:   File Output Format Counters
14/04/27 03:36:19 INFO mapred.JobClient:     Bytes Written=25
14/04/27 03:36:19 INFO mapred.JobClient:   FileSystemCounters
14/04/27 03:36:19 INFO mapred.JobClient:     FILE_BYTES_READ=55
14/04/27 03:36:19 INFO mapred.JobClient:     HDFS_BYTES_READ=260
14/04/27 03:36:19 INFO mapred.JobClient:     FILE_BYTES_WRITTEN=164041
14/04/27 03:36:19 INFO mapred.JobClient:     HDFS_BYTES_WRITTEN=25
14/04/27 03:36:19 INFO mapred.JobClient:   File Input Format Counters
14/04/27 03:36:19 INFO mapred.JobClient:     Bytes Read=25
14/04/27 03:36:19 INFO mapred.JobClient:   Map-Reduce Framework
14/04/27 03:36:19 INFO mapred.JobClient:     Map output materialized bytes=61
14/04/27 03:36:19 INFO mapred.JobClient:     Map input records=2
14/04/27 03:36:19 INFO mapred.JobClient:     Reduce shuffle bytes=61
14/04/27 03:36:19 INFO mapred.JobClient:     Spilled Records=8
14/04/27 03:36:19 INFO mapred.JobClient:     Map output bytes=41
14/04/27 03:36:19 INFO mapred.JobClient:     Total committed heap usage (bytes)=414441472
14/04/27 03:36:19 INFO mapred.JobClient:     CPU time spent (ms)=2910
14/04/27 03:36:19 INFO mapred.JobClient:     Combine input records=4
14/04/27 03:36:19 INFO mapred.JobClient:     SPLIT_RAW_BYTES=235
14/04/27 03:36:19 INFO mapred.JobClient:     Reduce input records=4
14/04/27 03:36:19 INFO mapred.JobClient:     Reduce input groups=3
14/04/27 03:36:19 INFO mapred.JobClient:     Combine output records=4
14/04/27 03:36:19 INFO mapred.JobClient:     Physical memory (bytes) snapshot=353439744
14/04/27 03:36:19 INFO mapred.JobClient:     Reduce output records=3
14/04/27 03:36:19 INFO mapred.JobClient:     Virtual memory (bytes) snapshot=2195972096
14/04/27 03:36:19 INFO mapred.JobClient:     Map output records=4

查看结果

[root@hadoop hadoop]# hadoop fs -cat out2/*
hadoop  1
hello   2
world   1

也可以将分布式文件系统上的文件拷贝到本地查看

[root@hadoop hadoop]# hadoop fs -get out2 out4
[root@hadoop hadoop]# cat out4/*
cat: out4/_logs: Is a directory
hadoop  1
hello   2
world   1

完成全部操作后，停止守护进程：

[root@hadoop hadoop]# stop-all.sh
stopping jobtracker
localhost: stopping tasktracker
stopping namenode
localhost: stopping datanode
localhost: stopping secondarynamenode

遇到的问题

宿主机不能访问网络接口

因为开启了iptables，所以需要添加相应端口，当然测试环境也可以直接将iptables关闭。

# Firewall configuration written by system-config-firewall
# Manual customization of this file is not recommended.
*filter
:INPUT ACCEPT [0:0]
:FORWARD ACCEPT [0:0]
:OUTPUT ACCEPT [0:0]
-A INPUT -m state --state ESTABLISHED,RELATED -j ACCEPT
-A INPUT -p icmp -j ACCEPT
-A INPUT -i lo -j ACCEPT
-A INPUT -m state --state NEW -m tcp -p tcp --dport 22 -j ACCEPT
-A INPUT -m state --state NEW -m tcp -p tcp --dport 50070 -j ACCEPT
-A INPUT -m state --state NEW -m tcp -p tcp --dport 50030 -j ACCEPT
-A INPUT -m state --state NEW -m tcp -p tcp --dport 50075 -j ACCEPT
-A INPUT -j REJECT --reject-with icmp-host-prohibited
-A FORWARD -j REJECT --reject-with icmp-host-prohibited
COMMIT

Browse the filesystem跳转地址不对

NameNode网络接口点击Browse the filesystem，跳转到localhost:50075。[2][3]

修改core-site.xml，将hdfs://localhost:9000改成虚拟机ip地址。(上面的脚本已经改写为自动配置为IP)。

根据几次改动的情况，这里也是可以填写域名的，只是要在访问的机器上能解析这个域名。因此公网环境中有DNS服务器的应该是可以设置域名的。

执行reduce的时候卡死

在/etc/hosts中添加主机名对应的ip地址 [4][5]。（已更新Hadoop安装脚本，会自动配置此项）

127.0.0.1   localhost localhost.localdomain localhost4 localhost4.localdomain4
::1         localhost localhost.localdomain localhost6 localhost6.localdomain6
127.0.0.1   hadoop  #添加这一行

参考文献

[1]. Hadoop官方文档.?http://hadoop.apache.org/docs/r1.2.1/single_node_setup.html

[2]. Stackoverflow.?http://stackoverflow.com/questions/15254492/wrong-redirect-from-hadoop-hdfs-namenode-to-localhost50075

[3]. Iteye.?http://yymmiinngg.iteye.com/blog/706909

[4].Stackoverflow.?http://stackoverflow.com/questions/10165549/hadoop-wordcount-example-stuck-at-map-100-reduce-0

[5]. 李俊的博客.?http://www.colorlight.cn/archives/32

 

---

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-2682.html