Hdfs datanode启动问题

场景：

项目组6个节点中，某一节点硬盘寿命到了，由于当时硬盘做的raid0，没有备份，更换新硬盘后，启动datanode遇到如下问题：

1. 更换硬盘前，停止datanode，停止在此节点上应用；

2. 停止机器，更换硬盘

3. 重启机器，重启datanode，遇到问题

3.1 

WARN org.apache.hadoop.hdfs.: Invalid /data :

WARN : Invalid .data.dir /data :



EPERM: Operation not permitted



        at $(Native Method)



        at $(:230)



        at (:729)



        at $1.apply(:505)



        at $FsOperation.run(:486)



        at (:502)



        at (:140)



        at (:156)



        at $(:2345)



        at (:2387)



        at (:2369)



        at (:2261)



        at (:2308)



        at (:2485)



        at (:2509)




2019-03-05 21:01:10,707 FATAL : Exception in secureMain



: All directories in .data.dir are invalid: "/data/"




        at (:2396)



        at (:2369)



        at (:2261)



        at (:2308)



        at (:2485)



        at (:2509)




2019-03-05 21:01:10,708 INFO : Exiting with status 1

原因：

由于更换硬盘，针对/data目录的权限变成了root，需要对当前用户赋予权限

解决办法：

chown -R hadoop:hadoop /data

 

3.2 

2019-03-05 21:25:47,891 FATAL .server.: Exception in secureMain
: failed to stat a path component: '/var/run/hdfs-sockets'

2019-03-05 21:25:47,891 FATAL : Exception in secureMain



: failed to stat a path component: '/var/run/hdfs-sockets'.  error code 2 (No such file or directory)



        at 0(Native Method)



        at (:189)



        at .<init>(:40)



        at (:946)



        at (:912)



        at (:1111)



        at .<init>(:429)



        at (:2374)



        at (:2261)



        at (:2308)



        at (:2485)



        at (:2509)




2019-03-05 21:25:47,893 INFO : Exiting with status 1





2019-03-05 21:25:47,894 INFO : SHUTDOWN_MSG:

原因：

没有/var/run/hdfs-sockets 这个路径

解决办法：

创建这个目录即可