HBase集群管理

新增Hadoop节点

Master
  • slaves新增node4
  • 配置分发( hosts, slaves )
node4

启动DataNode和NodeManager

1
2
[zhangsan@node4 default]$ sbin/hadoop-daemon.sh start datanode
[zhangsan@node4 default]$ sbin/yarn-daemon.sh start nodemanager
刷新节点状态
1
[zhangsan@node1 ~]$ hdfs dfsadmin -refreshNodes
负载均衡
1
2
3
4
5
6
# 设置负载均衡带宽
[zhangsan@node1 ~]$ hdfs dfsadmin -setBalancerBandwidth bytes
# 磁盘使用率阈值，单位为百分比。此命令会产生一个进程，并在logs目录输出日志；负载均衡完毕，进程结束。
# -threshold
# 默认值:10%。该值确保每个DataNode上的磁盘使用率与集群整体使用率的差距不超过10%。
[zhangsan@node1 ~]$ start-balancer.sh -threshold 1

https://partners-intl.aliyun.com/help/en/e-mapreduce/latest/hdfs-balancer

移除Hadoop节点

Master配置排除节点列表

hdfs-site.xml

1
2
3
4
<property>
    <name>dfs.hosts.exclude</name>
    <value>/opt/bigdata/hadoop/hadoop-2.7.3/etc/hadoop/datanode-excludes</value>
</property>

datanode-excludes (一行一个)

1
node4

yarn-site.xml

1
2
3
4
<property>
    <name>yarn.hosts.exclude</name>
    <value>/opt/bigdata/hadoop/hadoop-2.7.3/etc/hadoop/nodemanager-excludes</value>
</property>

nodemanager-excludes (一行一个)

1
node4
刷新节点状态

Decommission In Progress --> Decommissioned

1
2
[zhangsan@node1 hadoop]$ hdfs dfsadmin -refreshNodes
[zhangsan@node1 hadoop]$ yarn rmadmin -refreshNodes
停止守护进程
1
2
[zhangsan@node4 ~]$ hadoop-daemon.sh stop datanode
[zhangsan@node4 ~]$ yarn-daemon.sh stop nodemanager
查看下线情况
1
2
[zhangsan@node1 ~]$ hdfs dfsadmin -report
[zhangsan@node1 ~]$ yarn node -list