mkdir -p /app/hadoop
groupadd hadoop
useradd licz -g hadoop -d /app/hadoop
chown licz:hadoop /app/hadoop
passwd licz

[root@dbserver22 ~]# su - licz
-bash-3.2$

cp -a /etc/skel/. /app/hadoop

[licz@server123 ~]$ vi .bash_profile
PATH=$PATH:$HOME/bin
export LANG=zh_CN
export PATH
unset USERNAME
export HADOOP_HOME=/app/hadoop/hadoop-1.2.1
export JAVA_HOME=/usr/java/jdk1.6.0_18
export HIVE_HOME=/app/hadoop/hive-0.11.0
export PIG_HOME=/app/hadoop/pig-0.12.0
export PIG_CLASSPATH=/app/hadoop/pig-0.12.0/conf
PATH=$JAVA_HOME/bin:$PATH:$HOME/bin:$HADOOP_HOME/bin:$PIG_HOME/bin:$HIVE_HOME/bin
export PATH
export HADOOP_HOME_WARN_SUPPRESS=1

[root@server123 ~]# vi /etc/hosts
10.1.32.91             nticket1
10.1.32.93             nticket2
10.1.32.95             nticket3
10.1.5.123             server123

--首先，为了避免误操作，操作之前要先备份原集群的密钥文件
[licz@nticket1 .ssh]$ cp authorized_keys authorized_keys.bak
[licz@server123 ~]$ ssh-keygen -t rsa
[licz@server123 ~]$ cat ~/.ssh/id_rsa.pub >>~/.ssh/authorized_keys
[licz@server123 ~]$ ssh nticket1 cat ~/.ssh/authorized_keys >> ~/.ssh/authorized_keys
[licz@server123 ~]$ scp ~/.ssh/authorized_keys nticket1:~/.ssh/authorized_keys
[licz@server123 ~]$ ssh nticket1 date

[licz@nticket1 .ssh]$ ssh server123 date

--同样把新新密钥复制（覆盖）到原集群中的新节点
[licz@server123 ~]$ scp ~/.ssh/authorized_keys nticket2:~/.ssh/authorized_keys
[licz@server123 ~]$ scp ~/.ssh/authorized_keys nticket3:~/.ssh/authorized_keys

--在各节点修改hadoop的配置文件
[licz@nticket1 conf]$ vi slaves
nticket2
nticket3
server123

--把集群中的hadoop复制到新节点
[licz@nticket2~]$ scp -r hadoop-1.2.1/server123:/app/hadoop

[licz@server123~]$ hadoop-daemon.sh start datanode
startingdatanode, logging to /app/hadoop/hadoop-1.2.1/libexec/../logs/hadoop-licz-datanode-server123.out
[licz@server123~]$ hadoop-daemon.sh start tasktracker
startingtasktracker, logging to /app/hadoop/hadoop-1.2.1/libexec/../logs/hadoop-licz-tasktracker-server123.out
--测试安装成功
[licz@server123 ~]$ jps
18356 DataNode
18517 TaskTracker
18780 Jps

--在hdfs-site.xml中增加设置balance的带宽，默认只有1M：
<property>
<name>dfs.balance.bandwidthPerSec</name>
<value>10485760</value>
<description>
Specifies the maximum bandwidth thateach datanode can utilize for the balancing purpose in term of the number ofbytes per second.
</description>
</property>

[licz@server123conf]$ start-balancer.sh -threshold 5
startingbalancer, logging to /app/hadoop/hadoop-1.2.1/libexec/../logs/hadoop-licz-balancer-server123.out
Time Stamp               Iteration#  Bytes Already Moved  Bytes Left To Move  Bytes Being Moved
2014-2-20 17:55:14                0                 0 KB            14.12 GB           14.06 GB
--测试
[licz@server123~]$ hadoop dfs -ls /user/hive
Found 1 items
drwxr-xr-x   - licz supergroup          0 2014-02-10 11:25/user/hive/warehouse
[licz@nticket1 ~]$ hadoop dfsadmin -report
Configured Capacity: 2588293705728 (2.35 TB)
Present Capacity: 2027166097408 (1.84 TB)
DFS Remaining: 2026681536512 (1.84 TB)
DFS Used: 484560896 (462.11 MB)
DFS Used%: 0.02%
Under replicated blocks: 9
Blocks with corrupt replicas: 0
Missing blocks: 0
-------------------------------------------------
Datanodes available: 3 (3 total, 0 dead)
Name: 10.1.32.95:50010
Decommission Status : Normal
Configured Capacity: 1041225043968 (969.72 GB)
DFS Used: 242110464 (230.89 MB)
Non DFS Used: 102109831168 (95.1 GB)
DFS Remaining: 938873102336(874.39 GB)
DFS Used%: 0.02%
DFS Remaining%: 90.17%
Last contact: Fri Feb 14 09:49:02 CST 2014
Name: 10.1.32.93:50010
Decommission Status : Normal
Configured Capacity: 1041225043968 (969.72 GB)
DFS Used: 242143232 (230.93 MB)
Non DFS Used: 57774628864 (53.81 GB)
DFS Remaining: 983208271872(915.68 GB)
DFS Used%: 0.02%
DFS Remaining%: 94.43%
Last contact: Fri Feb 14 09:49:02 CST 2014
Name: 10.1.5.123:50010
Decommission Status : Normal
Configured Capacity: 505843617792 (471.1 GB)
DFS Used: 307200 (300 KB)
Non DFS Used: 401243148288 (373.69 GB)
DFS Remaining: 104600162304(97.42 GB)
DFS Used%: 0%
DFS Remaining%: 20.68%
Last contact: Fri Feb 14 09:49:03 CST 2014