hdfs ha脚本检测failover时，发送邮件预警

1.启动集群

[root@ruozedata002 ~]# su - hadoop
[hadoop@ruozedata001 ~]$ cd script/
[hadoop@ruozedata001 script]$ ./start_cluster.sh

---

2.检查集群进程

[hadoop@ruozedata001 script]$ ./jps.sh

---

3.检查 HA 中两个NameNode节点状态

[hadoop@ruozedata001 script]$ hdfs haadmin -getServiceState nn1
standby
[hadoop@ruozedata001 script]$ hdfs haadmin -getServiceState nn2
active

---

4.修改get_hdfs_ha_state.sh中的发件人邮箱和收件人邮箱及hadoop bin目录的路径

---

5.执行get_hdfs_ha_state.sh 脚本

[hadoop@ruozedata001 script]$ ./get_hdfs_ha_state.sh 

6.手动执行故障转移

---

7.执行get_hdfs_ha_state.sh 脚本