hadopp 环境搭建

前序：

首先准备三个虚拟机节点。

 配置hosts文件：每个节点都 如下配置：

vi /etc/hosts

1、 每个结点分别产生公私密钥

1. ssh-keygen -t dsa -P '' -f ~/.ssh/id_dsa

            

以上命令是产生公私密钥，产生目录在用户主目录下的.ssh目录中。

Id_dsa.pub为公钥，id_dsa为私钥，紧接着将公钥文件复制成authorized_keys文件，这个步骤是必须的，过程如下

1. [root@master .ssh]# cat ~/.ssh/id_dsa.pub >> ~/.ssh/authorized_keys

 

单机回环ssh免密码登录测试

如果ssh localhost 返回 

 则用yum下载SSH ：

1. yum -y install openssh-clients

 

以下信息表示操作成功，单点回环SSH登录及注销成功，这将为后续跨子结点SSH远程免密码登录作好准备。

 

2、其他节点也是如此操作即可。

3、让主结点(master)能通过SSH免密码登录两个子结点（slave）

     为了实现这个功能，两个slave结点的公钥文件中必须要包含主结点的公钥信息，这样

当master就可以顺利安全地访问这两个slave结点了。操作过程如下：

1. [root@node1 ~]# scp root@master:~/.ssh/id_dsa.pub ./master_dsa.pub
2. The authenticity of host 'master (30.96.76.220)' can't be established.
3. RSA key fingerprint is ae:8c:7f:00:df:40:b8:ec:20:4b:53:78:98:46:8a:c5.
4. Are you sure you want to continue connecting (yes/no)? yes
5. Warning: Permanently added 'master,30.96.76.220' (RSA) to the list of known hosts.
6. root@master's password:
7. id_dsa.pub 100% 601 0.6KB/s 00:00
8. [root@node1 .ssh]# cat master_dsa.pub >> authorized_keys


如上过程显示了node1结点通过scp命令远程登录master结点，并复制master的公钥文件到当前的目录下，这一过程需要密码验证。接着，将master结点的公钥文件追加至authorized_keys文件中，通过这步操作，如果不出问题，master结点就可以通过ssh远程免密码连接node1结点了。在master结点中操作如下：

1. [root@master .ssh]# ssh node1
2. The authenticity of host 'node1 (30.96.76.221)' can't be established.
3. RSA key fingerprint is ae:8c:7f:00:df:40:b8:ec:20:4b:53:78:98:46:8a:c5.
4. Are you sure you want to continue connecting (yes/no)? yes
5. Warning: Permanently added 'node1,30.96.76.221' (RSA) to the list of known hosts.
6. root@node1's password:
7. Last login: Fri Sep 9 17:42:15 2016 from localhost
8. [root@node1 ~]# ll
9. 总用量 24
10. -rw-------. 1 root root 1144 9月 9 18:11 anaconda-ks.cfg
11. -rw-r--r--. 1 root root 13231 9月 9 18:10 install.log
12. -rw-r--r--. 1 root root 3482 9月 9 18:09 install.log.syslog
13. [root@node1 ~]# pwd
14. /root
15. [root@node1 ~]# exit
16. logout
17. Connection to node1 closed.
18. [root@master .ssh]#

表面上看，这两个结点的ssh免密码登录已经配置成功，但是我们还需要对主结点master也要进行上面的同样工作，这一步有点让人困惑，但是这是有原因的，具体原因现在也说不太好，据说是真实物理结点时需要做这项工作，因为jobtracker有可能会分布在其它结点上，jobtracker有不存在master结点上的可能性。

1. [root@master .ssh]# scp root@master:~/.ssh/id_dsa.pub ./master_dsa.pub
2. The authenticity of host 'master (30.96.76.220)' can't be established.
3. RSA key fingerprint is ae:8c:7f:00:df:40:b8:ec:20:4b:53:78:98:46:8a:c5.
4. Are you sure you want to continue connecting (yes/no)? yes
5. Warning: Permanently added 'master,30.96.76.220' (RSA) to the list of known hosts.
6. id_dsa.pub 100% 601 0.6KB/s 00:00
7. [root@master .ssh]# ssh master
8. Last login: Fri Sep 9 17:42:39 2016 from localhost
9. [root@master ~]# hostname
10. master
11. [root@master ~]# exit
12. logout
13. Connection to master closed.

下载JDK,hadoop安装包,并配置环境变量自行百度。

 

4、配置hadoop 文件：

     4.1、 Core-site.xml

1. <?xml version="1.0"?>
2. <?xml-stylesheet type="text/xsl" href="configuration.xsl"?>
4. <!-- Put site-specific property overrides in this file. -->
6. <configuration>
7. <property>
8. <name>fs.default.name</name>
9. <value>hdfs://master:9000</value>
10. <final>true></final>
11. </property>
12. <property>
13. <name>hadoop.tmp.dir</name>
14. <value>/urs/hadoop/tmp</value>
15. <description>A base for other temporary directories</description>
16. </property>
17. </configuration>

  4.2、Hdfs-site.xml配置如下：

1. <?xml version="1.0"?>
2. <?xml-stylesheet type="text/xsl" href="configuration.xsl"?>
4. <!-- Put site-specific property overrides in this file. -->
6. <configuration>
7. <property>
8. <name>dfs.name.dir</name>
9. <value>/usr/hadoop-1.2.1/name</value>
10. <final>true</final>
11. </property>
12. <property>
13. <name>dfs.data.dir</name>
14. <value>/usr/hadoop-1.2.1/data</value>
15. <final>true</final>
16. </property>
17. <property>
18. <name>dfs.replication</name>
19. <value>2</value>
20. <final>true</final>
21. </property>
22. </configuration>

  4.3、mapred-site.xml:

1. <?xml version="1.0"?>
2. <?xml-stylesheet type="text/xsl" href="configuration.xsl"?>
4. <!-- Put site-specific property overrides in this file. -->
6. <configuration>
7. <property>
8. <name>mapred.job.tracker</name>
9. <value>30.96.76.220:9001</value>
10. </property>
11. </configuration>

（注：如果后续访问 http://30.96.76.220:50070/ 失败。请关闭防火墙试一下。具体问题还是查看日志比较清楚。）

   4.4、hadoop-env.sh:

     export JAVA_HOME=/usr/jdk/jdk1.7.0_76

  4.5 、masters和slaves文件

       

       

 

注：复制hoodoop 向 node1,node2.

5、格式化namenode

1. [root@master ~]# hadoop namenode -format
2. 16/09/10 17:03:49 INFO namenode.NameNode: STARTUP_MSG:
3. /************************************************************
4. STARTUP_MSG: Starting NameNode
5. STARTUP_MSG: host = master/30.96.76.220
6. STARTUP_MSG: args = [-format]
7. STARTUP_MSG: version = 1.2.1
8. STARTUP_MSG: build = https://svn.apache.org/repos/asf/hadoop/common/branches/branch-1.2 -r 1503152; compiled by 'mattf' on Mon Jul 22 15:23:09 PDT 2013
9. STARTUP_MSG: java = 1.7.0_76
10. ************************************************************/
11. 16/09/10 17:03:49 INFO util.GSet: Computing capacity for map BlocksMap
12. 16/09/10 17:03:49 INFO util.GSet: VM type = 64-bit
13. 16/09/10 17:03:49 INFO util.GSet: 2.0% max memory = 1013645312
14. 16/09/10 17:03:49 INFO util.GSet: capacity = 2^21 = 2097152 entries
15. 16/09/10 17:03:49 INFO util.GSet: recommended=2097152, actual=2097152
16. 16/09/10 17:03:50 INFO namenode.FSNamesystem: fsOwner=root
17. 16/09/10 17:03:50 INFO namenode.FSNamesystem: supergroup=supergroup
18. 16/09/10 17:03:50 INFO namenode.FSNamesystem: isPermissionEnabled=true
19. 16/09/10 17:03:50 INFO namenode.FSNamesystem: dfs.block.invalidate.limit=100
20. 16/09/10 17:03:50 INFO namenode.FSNamesystem: isAccessTokenEnabled=false accessKeyUpdateInterval=0 min(s), accessTokenLifetime=0 min(s)
21. 16/09/10 17:03:50 INFO namenode.FSEditLog: dfs.namenode.edits.toleration.length = 0
22. 16/09/10 17:03:50 INFO namenode.NameNode: Caching file names occuring more than 10 times
23. 16/09/10 17:03:50 INFO common.Storage: Image file /usr/hadoop-1.2.1/name/current/fsimage of size 110 bytes saved in 0 seconds.
24. 16/09/10 17:03:51 INFO namenode.FSEditLog: closing edit log: position=4, editlog=/usr/hadoop-1.2.1/name/current/edits
25. 16/09/10 17:03:51 INFO namenode.FSEditLog: close success: truncate to 4, editlog=/usr/hadoop-1.2.1/name/current/edits
26. 16/09/10 17:03:51 INFO common.Storage: Storage directory /usr/hadoop-1.2.1/name has been successfully formatted.
27. 16/09/10 17:03:51 INFO namenode.NameNode: SHUTDOWN_MSG:
28. /************************************************************
29. SHUTDOWN_MSG: Shutting down NameNode at master/30.96.76.220
30. ************************************************************/

注意：上面只要出现“successfully formatted”就表示成功了。

 6、启动hadoop

有这四项进程表示启动成功。

 

 表示node1.node2 启动成功

7、通过网站查看集群情况

      访问：http://30.96.76.220:50030/jobtracker.jsp 

     

  访问 http://30.96.76.220:50070/dfshealth.jsp

   

   

 

 全部成功。周六没有白加班。

 

最后附上原文连接：

http://blog.csdn.net/ab198604/article/details/8250461