引文地址:http://www.cnblogs.com/welbeckxu/archive/2011/12/29/2306757.html
接上一篇《ubuntu部署hadoop之单机模式》。
已经安装好了jdk和hadoop-0.20.2。
首先在hadoop目录下新建几个文件夹
xukangde@xukangde-01:~/hadoop$ mkdir tmp xukangde@xukangde-01:~/hadoop$ mkdir hdfs xukangde@xukangde-01:~/hadoop$ mkdir hdfs/name xukangde@xukangde-01:~/hadoop$ mkdir hdfs/data
core-site.xml: Hadoop Core的配置项,例如HDFS和MapReduce常用的I/O设置等。
hdfs-site.xml: Hadoop 守护进程的配置项,包括namenode,辅助namenode和datanode等。
mapred-site.xml: MapReduce 守护进程的配置项,包括jobtracker和tasktracker。
在conf目录中,编辑以下文件:(以xukangde用户登录,则目录path/to/your/hadoop应该为/home/xukangde/hadoop)
core-site.xml,hdfs-site.xml,mapred-site.xml
xukangde@xukangde-01:~/hadoop$ sudo gedit conf/core-site.xml
<configuration> <property> <name>fs.default.name</name> <value>hdfs://localhost:9000</value> </property> <property> <name>hadoop.tmp.dir</name> <value>/path/to/your/hadoop/tmp</value> </property> </configuration>
xukangde@xukangde-01:~/hadoop$ sudo gedit conf/hdfs-site.xml
<configuration> <property> <name>dfs.replication</name> <value>1</value> </property> <property> <name>dfs.name.dir</name> <value>/path/to/your/hadoop/hdfs/name</value> </property> <property> <name>dfs.data.dir</name> <value>/path/to/your/hadoop/hdfs/data</value> </property> </configuration>
xukangde@xukangde-01:~/hadoop$ sudo gedit conf/mapred-site.xml
<configuration>
<property>
<name>mapred.job.tracker</name>
<value>localhost:9001</value>
</property>
</configuration>
配置SSH
安装一个Hadoop集群时,需要专门指定一个服务器作为主节点。这个服务器通常会驻留NameNode和JobTracker的守护进程。它也将作为一个基站,负责联络并激活所有从节点上的DataNode和TaskTracker守护进程。因此,我们需要为主节点定制一种手段,使它能够远程地访问到集群中的每个节点。为此,Hadoop使用了无口令的(passphraseless)SSH协议。SSH采用标准的公钥加密来生成一对用户验证密钥---- 一个公钥,一个私钥。公钥被本地存储在集群的每个节点上,私钥则由主节点在试图访问远端节点时发送过来。结合这两段信息,目标机可以对这次登录尝试进行验证。
判断是否安装ssh服务,可以通过如下命令进行:
xukangde@xukangde-01:~$ ssh localhost
ssh: connect to host localhost port 22: Connection refused
如上所示,表示没有还没有安装,可以通过apt安装,命令如下:
xukangde@xukangde-01:~$ sudo apt-get install openssh-server
系统将自动进行安装,安装完成以后,先启动服务:
xukangde@xukangde-01:~$ sudo /etc/init.d/ssh start Rather than invoking init scripts through /etc/init.d, use the service(8) utility, e.g. service ssh start Since the script you are attempting to invoke has been converted to an Upstart job, you may also use the start(8) utility, e.g. start ssh
启动后,可以通过如下命令查看服务是否正确启动
xukangde@xukangde-01:~$ ps -e|grep ssh 1168 ? 00:00:00 ssh-agent 6524 ? 00:00:00 sshd
如上表示启动ok。注意,ssh默认的端口是22。
输入命令
xukangde@xukangde-01:~$ ssh localhost
运行后需要输入密码,做以下处理
xukangde@xukangde-01:~$ ssh-keygen -t rsa -P '' -f ~/.ssh/id_rsa xukangde@xukangde-01:~$ cat ~/.ssh/id_rsa.pub >> ~/.ssh/authorized_keys
再次输入命令
xukangde@xukangde-01:~$ ssh localhost
这时不再需要输入密码。
第一次运行
首先需要格式化dfs布式系统:
xukangde@xukangde-01:~/hadoop$ bin/hadoop namenode -format
(如果需要重新格式化,出现(Y / N )提示时记得输入大写的Y)
使用start-all.sh脚本来装载守护进程
xukangde@xukangde-01:~/hadoop$ bin/start-all.sh
用Java的jps命令列出所有守护进程来验证安装成功
xukangde@xukangde-01:~/hadoop$ jps
会见到下面类似的列表
2407 Jps 1914 DataNode 1751 NameNode 2075 SecondaryNameNode 2290 TaskTracker 2139 JobTracker
运行WordCount
首先在dfs中创建input目录
xukangde@xukangde-01:~/hadoop$ bin/hadoop dfs -mkdir input
将conf中的文件拷贝到dfs中的input
xukangde@xukangde-01:~/hadoop$ bin/hadoop dfs -copyFromLocal conf/* input
在伪分布式模式下运行WordCount
xukangde@xukangde-01:~/hadoop$ bin/hadoop jar hadoop-0.20.2-examples.jar wordcount input output
可看到以下过程
11/12/29 21:48:47 INFO input.FileInputFormat: Total input paths to process : 13 11/12/29 21:48:47 INFO mapred.JobClient: Running job: job_201112292145_0001 11/12/29 21:48:48 INFO mapred.JobClient: map 0% reduce 0% 11/12/29 21:48:57 INFO mapred.JobClient: map 15% reduce 0% 11/12/29 21:49:00 INFO mapred.JobClient: map 30% reduce 0% 11/12/29 21:49:03 INFO mapred.JobClient: map 46% reduce 0% 11/12/29 21:49:06 INFO mapred.JobClient: map 61% reduce 10% 11/12/29 21:49:09 INFO mapred.JobClient: map 76% reduce 10% 11/12/29 21:49:12 INFO mapred.JobClient: map 92% reduce 10% 11/12/29 21:49:15 INFO mapred.JobClient: map 100% reduce 20% 11/12/29 21:49:25 INFO mapred.JobClient: map 100% reduce 100% 11/12/29 21:49:27 INFO mapred.JobClient: Job complete: job_201112292145_0001 11/12/29 21:49:27 INFO mapred.JobClient: Counters: 17 11/12/29 21:49:27 INFO mapred.JobClient: Job Counters 11/12/29 21:49:27 INFO mapred.JobClient: Launched reduce tasks=1 11/12/29 21:49:27 INFO mapred.JobClient: Launched map tasks=13 11/12/29 21:49:27 INFO mapred.JobClient: Data-local map tasks=13 11/12/29 21:49:27 INFO mapred.JobClient: FileSystemCounters 11/12/29 21:49:27 INFO mapred.JobClient: FILE_BYTES_READ=15969 11/12/29 21:49:27 INFO mapred.JobClient: HDFS_BYTES_READ=18611 11/12/29 21:49:27 INFO mapred.JobClient: FILE_BYTES_WRITTEN=32426 11/12/29 21:49:27 INFO mapred.JobClient: HDFS_BYTES_WRITTEN=10932 11/12/29 21:49:27 INFO mapred.JobClient: Map-Reduce Framework 11/12/29 21:49:27 INFO mapred.JobClient: Reduce input groups=591 11/12/29 21:49:27 INFO mapred.JobClient: Combine output records=777 11/12/29 21:49:27 INFO mapred.JobClient: Map input records=567 11/12/29 21:49:27 INFO mapred.JobClient: Reduce shuffle bytes=16041 11/12/29 21:49:27 INFO mapred.JobClient: Reduce output records=591 11/12/29 21:49:27 INFO mapred.JobClient: Spilled Records=1554 11/12/29 21:49:27 INFO mapred.JobClient: Map output bytes=24746 11/12/29 21:49:27 INFO mapred.JobClient: Combine input records=1780 11/12/29 21:49:27 INFO mapred.JobClient: Map output records=1780 11/12/29 21:49:27 INFO mapred.JobClient: Reduce input records=777
显示输出结果
xukangde@xukangde-01:~/hadoop$ bin/hadoop dfs -cat output/*
当Hadoop结束时,可以通过stop-all.sh脚本来关闭Hadoop的守护进程
xukangde@xukangde-01:~/hadoop$ bin/stop-all.sh
在Hadoop中用于监控集群健康状态的Web界面
- http://localhost:50030/ – web UI for MapReduce job tracker(s)
- http://localhost:50060/ – web UI for task tracker(s)
- http://localhost:50070/ – web UI for HDFS name node(s)
单机模式和伪分布模式均用于开发和调试的目的。真实Hadoop集群的运行采用的是第三种模式,即全分布模式。待续。