1 服役新数据节点
- 需求:
随着公司业务的增长,数据量越来越大,原有的数据节点的容量已经不能满足存储数据的需求,需要在原有集群基础上动态添加新的数据节点。
1.1 环境准备
(1)克隆一台虚拟机
(2)修改ip地址和主机名称
(3)修改xcall和xsync文件,增加新增节点的同步
(4)删除原来HDFS文件系统留存的文件(还有ssh)
/opt/module/hadoop-2.7.2/data
1.2 服役新节点具体步骤
1.2.1
在namenode的/opt/module/hadoop-2.7.2/etc/hadoop目录下创建dfs.hosts文件
[root@pdc4 hadoop]$ pwd
/opt/module/hadoop-2.7.2/etc/hadoop
[root@pdc4 hadoop]$ touch dfs.hosts
[root@pdc4 hadoop]$ vi dfs.hosts
添加如下主机名称(包含新服役的节点)
pdc
pdc2
pdc3
pdc4
1.2.2
在namenode的hdfs-site.xml配置文件中增加dfs.hosts属性
集群启动的时候,加载slave,为静态,动态的话就需要配置dfs.hsots
<property>
<name>dfs.hosts</name>
<value>/opt/module/hadoop-2.7.2/etc/hadoop/dfs.hosts</value>
</property>
1.2.3
刷新namenode
[root@pdc hadoop-2.7.2]$ hdfs dfsadmin -refreshNodes
Refresh nodes successful
1.2.4
更新resourcemanager节点
[root @pdc hadoop-2.7.2]$ yarn rmadmin -refreshNodes
19/05/07 14:17:11 INFO client.RMProxy: Connecting to ResourceManager at pdc2/192.168.44.131:8033
1.2.5
在namenode的slaves文件中增加新主机名称(也可以不用)
增加pdc3 不需要分发
pdc
pdc2
pdc3
pdc4
1.2.6
单独命令启动新的数据节点和节点管理器
[root@pdc4 hadoop-2.7.2]$ sbin/hadoop-daemon.sh start datanode
starting datanode, logging to /opt/module/hadoop-2.7.2/logs/hadoop-atguigu-datanode-hadoop105.out
[root@pdc4 hadoop-2.7.2]$ sbin/yarn-daemon.sh start nodemanager
starting nodemanager, logging to /opt/module/hadoop-2.7.2/logs/yarn-atguigu-nodemanager-hadoop105.out
1.2.7
在web浏览器上检查是否ok
1.3 如果数据不均衡,可以用命令实现集群的再平衡
[root@pdc sbin]$ ./start-balancer.sh
starting balancer, logging to /opt/module/hadoop-2.7.2/logs/hadoop-atguigu-balancer-hadoop102.out
Time Stamp Iteration# Bytes Already Moved Bytes Left To Move Bytes Being Moved
2 退役旧数据节点
2.1
在namenode的/opt/module/hadoop-2.7.2/etc/hadoop目录下创建dfs.hosts.exclude文件
[root @pdc hadoop]$ pwd
/opt/module/hadoop-2.7.2/etc/hadoop
[root @ pdc hadoop]$ touch dfs.hosts.exclude
[root@ pdc hadoop]$ vi dfs.hosts.exclude
添加如下主机名称(要退役的节点)
pdc4
2.2
在namenode的hdfs-site.xml配置文件中增加dfs.hosts.exclude属性
<property>
<name>dfs.hosts.exclude</name>
<value>/opt/module/hadoop-2.7.2/etc/hadoop/dfs.hosts.exclude</value>
</property>
2.3
刷新namenode、刷新resourcemanager
[root@pdc hadoop-2.7.2]$ hdfs dfsadmin -refreshNodes
Refresh nodes successful
[root@pdc hadoop-2.7.2]$ yarn rmadmin -refreshNodes
17/06/24 14:55:56 INFO client.RMProxy: Connecting to ResourceManager at hadoop103/192.168.1.103:8033
2.4
检查web浏览器,退役节点的状态为decommission in progress(退役中),说明数据节点正在复制块到其他节点。
2.5
等待退役节点状态为decommissioned(所有块已经复制完成),停止该节点及节点资源管理器。注意:如果副本数是3,服役的节点小于等于3,是不能退役成功的,需要修改副本数后才能退役。·
[root@pdc4 hadoop-2.7.2]$ sbin/hadoop-daemon.sh stop datanode
stopping datanode
[root@pdc4 hadoop-2.7.2]$ sbin/yarn-daemon.sh stop nodemanager
stopping nodemanager
2.6
从include文件中删除退役节点,再运行刷新节点的命令
2.6.1
从namenode的dfs.hosts文件中删除退役节点pdc4
pdc
pdc2
pdc3
2.6.2
刷新namenode,刷新resourcemanager
[root@pdc hadoop-2.7.2]$ hdfs dfsadmin -refreshNodes
Refresh nodes successful
[root@pdc hadoop-2.7.2]$ yarn rmadmin -refreshNodes
17/06/24 14:55:56 INFO client.RMProxy: Connecting to ResourceManager at pdc2/192.168.44.131:8033
2.7
从namenode的slave文件中删除退役节点pdc4
pdc
pdc2
pdc3
2.8
如果数据不均衡,可以用命令实现集群的再平衡
[root@pdc hadoop-2.7.2]$ sbin/start-balancer.sh
starting balancer, logging to /opt/module/hadoop-2.7.2/logs/hadoop-atguigu-balancer-hadoop102.out
Time Stamp Iteration# Bytes Already Moved Bytes Left To Move Bytes Being Moved