2.2 启动服务
前面各节点参数配置好之后，接下来要先对namenode进行格式化，这个操作仍是在hdnode1节点上执行：
[grid@hdnode1 ~]$ hadoop namenode -format
13/01/30 10:14:36 INFO namenode.NameNode: STARTUP_MSG:
/************************************************************
STARTUP_MSG: Starting NameNode
STARTUP_MSG: host = hdnode1/192.168.30.203
STARTUP_MSG: args = [-format]
STARTUP_MSG: version = 0.20.2
STARTUP_MSG: build = https://svn.apache.org/repos/asf/hadoop/common/branches/branch-0.20 -r 911707; compiled by ¨chrisdo¨ on Fri Feb 19 08:07:34 UTC 2010
************************************************************/
Re-format filesystem in /data2/hadoop/name ? (Y or N) y
Format aborted in /data2/hadoop/name
13/01/30 10:14:39 INFO namenode.NameNode: SHUTDOWN_MSG:
/************************************************************
SHUTDOWN_MSG: Shutting down NameNode at hdnode1/192.168.30.203
************************************************************/在执行格式化时，三思本地的测试环境反复尝试都报错。不过如果我们把hdfs-site.xml中dfs.name.dir那段property给删除掉，那么就能正常执行。可是，我实在是希望能够自定义namenode的存储路径，经过对比hadoop权威指南，该段参数写的也没有问题。
不过最终这个问题还是被解决了，知道怎么解决的吗？泥嘛呀，坑爹了。注意前面输出信息中加粗的部分。最终发现，居然是由于选择Y or N那里，我输入的是小写y，它默认理解非Y就是N，于是就没格式化。
重新执行hadoop namenode -format：
[grid@hdnode1 ~]$ hadoop namenode -format
13/01/30 105:20:07 INFO namenode.NameNode: STARTUP_MSG:
/************************************************************
STARTUP_MSG: Starting NameNode
STARTUP_MSG: host = hdnode1/192.168.30.203
STARTUP_MSG: args = [-format]
STARTUP_MSG: version = 0.20.2
STARTUP_MSG: build = https://svn.apache.org/repos/asf/hadoop/common/branches/branch-0.20 -r 911707; compiled by ¨chrisdo¨ on Fri Feb 19 08:07:34 UTC 2010
************************************************************/
Re-format filesystem in /data2/hadoop/name ? (Y or N)Y
13/01/30 10:20:08 INFO namenode.FSNamesystem: fsOwner=grid,grid
13/01/30 10:20:08 INFO namenode.FSNamesystem: supergroup=supergroup
13/01/30 10:20:08 INFO namenode.FSNamesystem: isPermissionEnabled=true
13/01/30 10:20:08 INFO common.Storage: Image file of size 94 saved in 0 seconds.
13/01/30 10:20:08 INFO common.Storage: Storage directory /data2/hadoop/name has been successfully formatted.
13/01/30 10:20:08 INFO namenode.NameNode: SHUTDOWN_MSG:
/************************************************************
SHUTDOWN_MSG: Shutting down NameNode at hdnode1/192.168.30.203
************************************************************/注意加粗字体，然后就正常了。
激动人心的时刻到了，启动hadoop集群，让我们见识见识传说中的Hadoop到底是怎么个情况吧。
直接执行start-all.sh脚本即可：
[grid@hdnode1 hadoop]$ start-all.sh
starting namenode, logging to /usr/local/hadoop-0.20.2/bin/../logs/hadoop-grid-namenode-hdnode1.out
hdnode3: starting datanode, logging to /usr/local/hadoop-0.20.2/bin/../logs/hadoop-grid-datanode-hdnode3.out
hdnode2: starting datanode, logging to /usr/local/hadoop-0.20.2/bin/../logs/hadoop-grid-datanode-hdnode2.out
hdnode1: starting secondarynamenode, logging to /usr/local/hadoop-0.20.2/bin/../logs/hadoop-grid-secondarynamenode-hdnode1.out
starting jobtracker, logging to /usr/local/hadoop-0.20.2/bin/../logs/hadoop-grid-jobtracker-hdnode1.out
hdnode3: starting tasktracker, logging to /usr/local/hadoop-0.20.2/bin/../logs/hadoop-grid-tasktracker-hdnode3.out
hdnode2: starting tasktracker, logging to /usr/local/hadoop-0.20.2/bin/../logs/hadoop-grid-tasktracker-hdnode2.out这个脚本做了什么呢，查看这个脚本的内容，会发现它其实是调用了start-dfs.sh和start-mapred.sh两个脚本。如果你认真看了屏幕输出的信息，即使不看脚本的内容，也会发现输出的日志已经把什么都告诉我们了。
Start-all首先会启动namenode，而后按照conf/slaves文件中指定的节点各启动一个DataNode，在conf/masters文件指定的节点各启动一个SecondaryNameNode；这时start-dfs.sh的工具就结束了，而后执行的start-mapred.sh与start-dfs.sh类似，它会在本地机器上启动一个JobTracker，在conf/slaves文件中指定的节点上各启动一个TaskTracker。
提示：有start就有stop，若要关闭资源，则有与之对应的stop-mapred.sh、stop-dfs.sh及stop-all.sh用于停止这些服务。
验证服务端节点上的服务进程(有namenode但没有datanode)：
[grid@hdnode1 ~]$ jps
7727 Jps
6588 JobTracker
6404 NameNode
6524 SecondaryNameNode验证客户端节点上的进程(有datanode没有namenode)：
[grid@hdnode2 ~]$ jps
9581 Jps
9337 DataNode
9409 TaskTracker查看HDFS存储信息：
[grid@hdnode1 ~]$ hadoop dfsadmin -report
Configured Capacity: 211376005120 (196.86 GB)
Present Capacity: 200244994048 (186.49 GB)
DFS Remaining: 200244920320 (186.49 GB)
DFS Used: 73728 (72 KB)
DFS Used%: 0%
Under replicated blocks: 1
Blocks with corrupt replicas: 0
Missing blocks: 0
-------------------------------------------------
Datanodes available: 2 (2 total, 0 dead)
Name: 192.168.30.204:50010
Decommission Status : Normal
Configured Capacity: 105688002560 (98.43 GB)
DFS Used: 36864 (36 KB)
Non DFS Used: 5565505536 (5.18 GB)
DFS Remaining: 100122460160(93.25 GB)
DFS Used%: 0%
DFS Remaining%: 94.73%
Last contact: Tue Feb 31 10:33:51 CST 2013
Name: 192.168.30.205:50010
Decommission Status : Normal
Configured Capacity: 105688002560 (98.43 GB)
DFS Used: 36864 (36 KB)
Non DFS Used: 5565505536 (5.18 GB)
DFS Remaining: 100122460160(93.25 GB)
DFS Used%: 0%
DFS Remaining%: 94.73%
Last contact: Tue Feb 31 10:33:51 CST 2013此外也可以通过浏览器获取服务的信息：
- http://192.168.30.203:50030：Namenode状态信息；
- http://192.168.30.203:50070：浏览HDFS文件系统；
- http://192.168.30.203:50090：Secondary Namenode状态信息；