hadoop datanode running as process 16774. Stop it first.

我有一台机器173，这是一台虚拟机，某一天这台虚拟机所在的物理机器内存条出现问题，导致机器宕机，于是换了内存后，我在173上执行命令 start-dfs.sh,提示下面的异常

Starting namenodes on [dashuju174]
dashuju174: namenode running as process 8363. Stop it first.
The authenticity of host '[dashuju173]:2222 ([192.168.5.173]:2222)' can't be established.
RSA key fingerprint is a4:91:9f:e5:f9:5d:28:e6:18:04:cc:1d:4f:6d:94:9f.
Are you sure you want to continue connecting (yes/no)? dashuju172: datanode running as process 16774. Stop it first.

hadoop出现namenode running as process 18472. Stop it first.这篇文章告诉我们，我们需要将先将其他的hadoop节点停掉，就可以了。
如果是生产环境出现硬件的问题，也就意味着所有数据都没法直接写入hadoop了。如果hadoop是这样的机制，那么就要求hadoop前面一定要有缓冲。否则真是灾难性的。
我只好先执行stop-all.sh,再执行start-all.sh
这里注意，按照下面的顺序

对于hbase启动顺序：hadoop-->zookeeper-->hbase
关闭顺序：hbase-->zookeeper-->hadoop

如果这个时候再173上单独执行start-hbase.sh,可以发现出现了两个HMaster