Hadoop动态添加删除节点datanode及恢复

1. 配置系统环境

主机名,ssh互信,环境变量等

本文略去jdk安装，请将datanode的jdk安装路径与/etc/Hadoop/hadoop-evn.sh中的java_home保持一致，版本hadoop2.7.5

修改/etc/sysconfig/network

然后执行命令 hostname 主机名 这个时候可以注销一下系统，再重登录之后就行了

[root@localhost ~]# hostname localhost.localdomain [root@localhost ~]# hostname -i ::1 127.0.0.1 [root@localhost ~]# [root@localhost ~]# cat /etc/sysconfig/network # Created by anaconda NETWORKING=yes HOSTNAME=slave2 GATEWAY=192.168.48.2 # Oracle-rdbms-server-11gR2-preinstall : Add NOZEROCONF=yes NOZEROCONF=yes [root@localhost ~]# hostname slave2 [root@localhost ~]# hostname slave2 [root@localhost ~]# su - hadoop Last login: Sat Feb 24 14:25:48 CST 2018 on pts/1 [hadoop@slave2 ~]$ su - root

建datanode目录并改所有者

（此处的具体路径值，请参照namenode中/usr/hadoop/hadoop-2.7.5/etc/hadoop/hdfs-site.xml,core-site.xml中的dfs.name.dir，dfs.data.dir,dfs.tmp.dir等）

Su - root

# mkdir -p /usr/local/hadoop-2.7.5/tmp/dfs/data

# chmod -R 777 /usr/local/hadoop-2.7.5/tmp

# chown -R hadoop:hadoop /usr/local/hadoop-2.7.5

[root@slave2 ~]# mkdir -p /usr/local/hadoop-2.7.5/tmp/dfs/data [root@slave2 ~]# chmod -R 777 /usr/local/hadoop-2.7.5/tmp  [root@slave2 ~]# chown -R hadoop:hadoop /usr/local/hadoop-2.7.5  [root@slave2 ~]# pwd /root [root@slave2 ~]# cd /usr/local/ [root@slave2 local]# ll total 0 drwxr-xr-x. 2 root  root  46 Mar 21  2017 bin drwxr-xr-x. 2 root  root    6 Jun 10  2014 etc drwxr-xr-x. 2 root  root    6 Jun 10  2014 games drwxr-xr-x  3 hadoop hadoop 16 Feb 24 18:18 hadoop-2.7.5 drwxr-xr-x. 2 root  root    6 Jun 10  2014 include drwxr-xr-x. 2 root  root    6 Jun 10  2014 lib drwxr-xr-x. 2 root  root    6 Jun 10  2014 lib64 drwxr-xr-x. 2 root  root    6 Jun 10  2014 libexec drwxr-xr-x. 2 root  root    6 Jun 10  2014 sbin drwxr-xr-x. 5 root  root  46 Dec 17  2015 share drwxr-xr-x. 2 root  root    6 Jun 10  2014 src [root@slave2 local]#

ssh互信，即实现 master-->slave2免密码

master:

[root@hadoop-master ~]# cat /etc/hosts

127.0.0.1  localhost localhost.localdomain localhost4 localhost4.localdomain4

::1        localhost localhost.localdomain localhost6 localhost6.localdomain6

192.168.48.129    hadoop-master

192.168.48.132    slave1

192.168.48.131    slave2

[hadoop@hadoop-master ~]$ scp /usr/hadoop/.ssh/authorized_keys hadoop@slave2:/usr/hadoop/.ssh

The authenticity of host 'slave2 (192.168.48.131)' can't be established.

ECDSA key fingerprint is 1e:cd:d1:3d:b0:5b:62:45:a3:63:df:c7:7a:0f:b8:7c.

Are you sure you want to continue connecting (yes/no)? yes

Warning: Permanently added 'slave2,192.168.48.131' (ECDSA) to the list of known hosts.

hadoop@slave2's password:

authorized_keys       

[hadoop@hadoop-master ~]$ ssh hadoop@slave2

Last login: Sat Feb 24 18:27:33 2018

[hadoop@slave2 ~]$

[hadoop@slave2 ~]$ exit

logout

Connection to slave2 closed.

[hadoop@hadoop-master ~]$

2. 修改namenode节点的slave文件,增加新节点信息

[hadoop@hadoop-master hadoop]$ pwd

/usr/hadoop/hadoop-2.7.5/etc/hadoop

[hadoop@hadoop-master hadoop]$ vi slaves

slave1

slave2

3. 在namenode节点上,将hadoop-2.7.3复制到新节点上,并在新节点上删除data和logs目录中的文件

Master

[hadoop@hadoop-master ~]$ scp -R hadoop-2.7.5 hadoop@slave2:/usr/hadoop

Slave2

[hadoop@slave2 hadoop-2.7.5]$ ll

total 124

drwxr-xr-x 2 hadoop hadoop  4096 Feb 24 14:29 bin

drwxr-xr-x 3 hadoop hadoop    19 Feb 24 14:30 etc

drwxr-xr-x 2 hadoop hadoop  101 Feb 24 14:30 include

drwxr-xr-x 3 hadoop hadoop    19 Feb 24 14:29 lib

drwxr-xr-x 2 hadoop hadoop  4096 Feb 24 14:29 libexec

-rw-r--r-- 1 hadoop hadoop 86424 Feb 24 18:44 LICENSE.txt

drwxrwxr-x 2 hadoop hadoop  4096 Feb 24 14:30 logs

-rw-r--r-- 1 hadoop hadoop 14978 Feb 24 18:44 NOTICE.txt

-rw-r--r-- 1 hadoop hadoop  1366 Feb 24 18:44 README.txt

drwxr-xr-x 2 hadoop hadoop  4096 Feb 24 14:29 sbin

drwxr-xr-x 4 hadoop hadoop    29 Feb 24 14:30 share

[hadoop@slave2 hadoop-2.7.5]$ pwd

/usr/hadoop/hadoop-2.7.5

[hadoop@slave2 hadoop-2.7.5]$ rm -R logs/*

4. 启动新datanode的datanode和nodemanger进程

先确认namenode和当前的datanode中，etc/hoadoop/excludes文件中无待加入的主机，再进行下面操作

[hadoop@slave2 hadoop-2.7.5]$ sbin/hadoop-daemon.sh start datanode starting datanode, logging to /usr/hadoop/hadoop-2.7.5/logs/hadoop-hadoop-datanode-slave2.out [hadoop@slave2 hadoop-2.7.5]$ sbin/yarn-daemon.sh start nodemanager starting datanode, logging to /usr/hadoop/hadoop-2.7.5/logs/yarn-hadoop-datanode-slave2.out [hadoop@slave2 hadoop-2.7.5]$ [hadoop@slave2 hadoop-2.7.5]$ jps 3897 DataNode 6772 NodeManager 8189 Jps [hadoop@slave2 ~]$

5、在NameNode上刷新节点

[hadoop@hadoop-master ~]$ hdfs dfsadmin -refreshNodes Refresh nodes successful [hadoop@hadoop-master ~]$sbin/start-balancer.sh

6. 在namenode查看当前集群情况,

确认信节点已经正常加入

[hadoop@hadoop-master hadoop]$ hdfs dfsadmin -report Configured Capacity: 58663657472 (54.63 GB) Present Capacity: 15487176704 (14.42 GB) DFS Remaining: 15486873600 (14.42 GB) DFS Used: 303104 (296 KB) DFS Used%: 0.00% Under replicated blocks: 5 Blocks with corrupt replicas: 0 Missing blocks: 0 Missing blocks (with replication factor 1): 0

------------------------------------------------- Live datanodes (2):

Name: 192.168.48.131:50010 (slave2) Hostname: 183.221.250.11 Decommission Status : Normal Configured Capacity: 38588669952 (35.94 GB) DFS Used: 8192 (8 KB) Non DFS Used: 36887191552 (34.35 GB) DFS Remaining: 1701470208 (1.58 GB) DFS Used%: 0.00% DFS Remaining%: 4.41% Configured Cache Capacity: 0 (0 B) Cache Used: 0 (0 B) Cache Remaining: 0 (0 B) Cache Used%: 100.00% Cache Remaining%: 0.00% Xceivers: 1 Last contact: Thu Mar 01 19:36:33 PST 2018

Name: 192.168.48.132:50010 (slave1) Hostname: slave1 Decommission Status : Normal Configured Capacity: 20074987520 (18.70 GB) DFS Used: 294912 (288 KB) Non DFS Used: 6289289216 (5.86 GB) DFS Remaining: 13785403392 (12.84 GB) DFS Used%: 0.00% DFS Remaining%: 68.67% Configured Cache Capacity: 0 (0 B) Cache Used: 0 (0 B) Cache Remaining: 0 (0 B) Cache Used%: 100.00% Cache Remaining%: 0.00% Xceivers: 1 Last contact: Thu Mar 01 19:36:35 PST 2018

[hadoop@hadoop-master hadoop]$

7动态删除datanode

7.1配置NameNode的hdfs-site.xml，

适当减小dfs.replication副本数，增加dfs.hosts.exclude配置

[hadoop@hadoop-master hadoop]$ pwd /usr/hadoop/hadoop-2.7.5/etc/hadoop [hadoop@hadoop-master hadoop]$ cat hdfs-site.xml <configuration> <property>       <name>dfs.replication</name>       <value>3</value> </property>   <property>       <name>dfs.name.dir</name>       <value>/usr/local/hadoop-2.7.5/tmp/dfs/name</value> </property>     <property>       <name>dfs.data.dir</name>       <value>/usr/local/hadoop-2.7.5/tmp/dfs/data</value>     </property> <property>     <name>dfs.hosts.exclude</name>     <value>/usr/hadoop/hadoop-2.7.5/etc/hadoop/excludes</value>   </property>

</configuration>

7.2在namenode对应路径（/etc/hadoop/）下新建excludes文件，

并写入待删除DataNode的ip或域名

[hadoop@hadoop-master hadoop]$ pwd /usr/hadoop/hadoop-2.7.5/etc/hadoop [hadoop@hadoop-master hadoop]$ vi excludes ####slave2 192.168.48.131[hadoop@hadoop-master hadoop]$

7.3在NameNode上刷新所有DataNode

hdfs dfsadmin -refreshNodes sbin/start-balancer.sh

7.4在namenode查看当前集群情况,

确认信节点已经正常删除,结果中已无slave2

[hadoop@hadoop-master hadoop]$ hdfs dfsadmin -report

或者可以在web检测界面（ip:50070）上可以观测到DataNode逐渐变为Dead。

http://192.168.48.129:50070/

在datanode项,Admin state已经由“In Service“变为”Decommissioned“，则表示删除成功

7.5停止已删除的节点相关进程

[hadoop@slave2 hadoop-2.7.5]$ jps 9530 Jps 3897 DataNode 6772 NodeManager [hadoop@slave2 hadoop-2.7.5]$ sbin/hadoop-daemon.sh stop datanode stopping datanode [hadoop@slave2 hadoop-2.7.5]$ sbin/yarn-daemon.sh stop nodemanager stopping nodemanager [hadoop@slave2 hadoop-2.7.5]$ jps 9657 Jps [hadoop@slave2 hadoop-2.7.5]$

8恢复已删除节点

执行7.2 中删除相关信息，然后4，5，6即可。