1.zookeeper配置正常
关闭yarn
stop-yarn.sh
在yarn-site.xml中加上如下配置（将原先的resourcemanager.hostname注释掉）
<property>
<name>yarn.resourcemanager.ha.enabled</name>
<value>true</value>
</property>
<property>
<name>yarn.resourcemanager.cluster-id</name>
<value>cluster1</value>
<description>集群唯一标识</description>
</property>
<property>
<name>yarn.resourcemanager.ha.rm-ids</name>
<value>rm1,rm2</value>
<description>两个RM的唯一标识</descritption>
</property>
<property>
<name>yarn.resourcemanager.hostname.rm1</name>
<value>192-168-100-142</value>
<description>第一个RM部署在的机器名</descritption>
</property>
<property>
<name>yarn.resourcemanager.hostname.rm2</name>
<value>192-168-100-217</value>
<description>第二个RM部署在的机器名</descritption>
</property>
<property>
<name>yarn.resourcemanager.webapp.address.rm1</name>
<value>192-168-100-142:8088</value>
<description>第一个RM的web ui的端口</descritption>
</property>
<property>
<name>yarn.resourcemanager.webapp.address.rm2</name>
<value>192-168-100-217:8088</value>
<description>第二个RM的web ui的端口</descritption>
</property>
<property>
<name>yarn.resourcemanager.zk-address</name>
<value>192-168-100-142:2181,192-168-100-217:2181,192-168-100-34:2181</value>
<description>zk的部署的主机名和端口</descritption>
</property>




2.同步配置文件
   scp yarn-site.xml root@192-168-100-217:/usr/local/hadoop-2.7.6/etc/hadoop/
   scp yarn-site.xml root@192-168-100-225:/usr/local/hadoop-2.7.6/etc/hadoop/
   scp yarn-site.xml root@192-168-100-34:/usr/local/hadoop-2.7.6/etc/hadoop/

3.在192-168-100-142上启动yarn， start-yarn.sh
  在192-168-100-217上启动resourcemanager: yarn-daemon.sh start resourcemanager
4.查看每个节点的状态
 yarn rmadmin -getServiceState rm1
 active
 yarn rmadmin -getServiceState rm2
 standby

5.HA failover切换

 杀掉一个rm1节点的resourcemanager,另外一个节点rm2的自动切换到active

[root@192-168-100-142 hadoop]# yarn rmadmin -getServiceState rm1
active
[root@192-168-100-142 hadoop]# yarn rmadmin -getServiceState rm2
standby
[root@192-168-100-142 hadoop]# jps

26148 QuorumPeerMain
6967 Jps
6104 ResourceManager
21641 NameNode
You have new mail in /var/spool/mail/root
[root@192-168-100-142 hadoop]# 
[root@192-168-100-142 hadoop]# kill -9 6104
[root@192-168-100-142 hadoop]# yarn rmadmin -getServiceState rm1
18/07/02 13:46:57 INFO ipc.Client: Retrying connect to server: 192-168-100-142/192.168.100.142:8033. Already tried 0 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=1, sleepTime=1000 MILLISECONDS)
Operation failed: Call From 192-168-100-142.sky-mobi.com/192.168.100.142 to 192-168-100-142:8033 failed on connection exception: java.net.ConnectException: Connection refused; For more details see:  http://wiki.apache.org/hadoop/ConnectionRefused
[root@192-168-100-142 hadoop]# yarn rmadmin -getServiceState rm2
active

如果要恢复到一个节点的yarn,先关闭yarn,然后把配置文件改回到原来一个节点的配置然后，启动yarn即可。