13859 DataNode
去data01节点看下，后台的jps进程如下： [hadoop@data01 ~]$ jps 3194 Jps
2863 DataNode
[hadoop@data01 ~]$
再启动yarn
执行命令start-yarn.sh，可以单独启动资源管理器的服务器端和客户端进程，关闭的命令就是stop-yarn.sh
[hadoop@name01 sbin]$ start-yarn.sh starting yarn daemons starting resourcemanager, logging to /home/hadoop/src/hadoop-2.6.0/logs/yarn-hadoop-resourcemanager-name01.out data01: starting nodemanager, logging to /home/hadoop/src/hadoop-2.6.0/logs/yarn-hadoop-nodemanager-data01.out data02: starting nodemanager, logging to /home/hadoop/src/hadoop-2.6.0/logs/yarn-hadoop-nodemanager-data02.out
在name01节点下，查看后台的jps进程，多了一个ResourceManager进程，如下所示： [hadoop@name01 sbin]$ jps 4601 ResourceManager 3800 NameNode 4853 Jps
3977 SecondaryNameNode [hadoop@name01 sbin]$
去data01节点看下，后台的jps进程多了一个NodeManager进程，如下所示： [hadoop@data01 ~]$ jps 3249 NodeManager 2863 DataNode 3365 Jps
[hadoop@data01 ~]$
依次关闭，先关闭yarn再关闭HDFS [hadoop@name01 sbin]$ stop-yarn.sh stopping yarn daemons stopping resourcemanager data01: stopping nodemanager data02: stopping nodemanager no proxyserver to stop [hadoop@name01 sbin]$
[hadoop@name01 sbin]$ stop-dfs.sh Stopping namenodes on [name01] name01: stopping namenode data01: stopping datanode data02: stopping datanode
Stopping secondary namenodes [name01] name01: stopping secondarynamenode [hadoop@name01 sbin]$
PS：当然，也可以先启动MapReduce，再启动HDFS。说明HDFS 和MapReduce的进程之间是互相独立的，没有依赖关系。
4.5.3第三种，分别启动各个进程
[root@book0 bin]# jps 14821 Jps
[root@book0 bin]# hadoop-daemon.sh start namenode [root@book0 bin]# hadoop-daemon.sh start datanode
[root@book0 bin]# hadoop-daemon.sh start secondarynamenode [root@book0 bin]# hadoop-daemon.sh start jobtracker [root@book0 bin]# hadoop-daemon.sh start tasktracker [root@book0 bin]# jps 14855 NameNode 14946 DataNode
15043 SecondaryNameNode 15196 TaskTracker 15115 JobTracker 15303 Jps
执行的命令是“hadoop-daemon.sh start [进程名称]”，这种启动方式适合于单独增加、删除节点的情况，在安装集群环境的时候会看到
4.6另外一种检查状态hadoop集群的状态
用\来查看hadoop集群的状态