测试前环境准备：
PostgreSQL主备搭建参考下面三篇文章
PostgreSQL高可用之Repmgr两节点搭建及switchover操作
PostgreSQL高可用之Repmgr自动failover
PostgreSQL高可用之Repmgr见证服务器搭建

应用层JDBC配置
jdbc.url=jdbc:postgresql://192.168.99.200:6000,192.168.99.202:6000/DB
jdbc.username=your user name
jdbc.password=your password

node1

[postgres@ha1 ~]$ psql -h192.168.99.200 -p6000 -Upostgres
psql (12.1)
Type "help" for help.

postgres=# \conninfo
You are connected to database "postgres" as user "postgres" on host "192.168.99.200" at port "6000".
复制

node2

[postgres@ha2 ~]$ psql -h192.168.99.202 -p6000 -Upostgres
psql (12.1)
Type "help" for help.

postgres=# \conninfo
You are connected to database "postgres" as user "postgres" on host "192.168.99.202" at port "6000".
复制

Repmgr显示集群状态

$ /opt/pgsql/bin/repmgr -f /opt/repmgr.conf cluster show
 ID | Name  | Role    | Status    | Upstream | Location | Priority | Timeline | Connection string                                                        
----+-------+---------+-----------+----------+----------+----------+----------+---------------------------------------------------------------------------
 1  | node1 | primary | * running |          | default  | 100      | 21       | host=192.168.99.200 port=6000 user=repmgr dbname=repmgr connect_timeout=2
 2  | node2 | standby |   running | node1    | default  | 100      | 21       | host=192.168.99.202 port=6000 user=repmgr dbname=repmgr connect_timeout=2
 3  | node3 | witness | * running | node1    | default  | 0        | 1        | host=192.168.99.222 port=6000 user=repmgr dbname=repmgr connect_timeout=2
复制

测试一：模拟node2故障

[postgres@ha2 ~]$ /opt/pgsql/bin/pg_ctl -D /opt/data6000/ stop
复制

测试完成后rejoin恢复集群

[postgres@ha2 ~]$ /opt/pgsql/bin/repmgr -f /opt/repmgr.conf \
node rejoin -d'host=192.168.99.200 port=6000 user=repmgr dbname=repmgr connect_timeout=2'
复制

测试二：模拟node1故障

[postgres@ha1 ~]$ /opt/pgsql/bin/pg_ctl -D /opt/data6000/ stop
复制

测试完成后rejoin恢复集群

[postgres@ha1 ~]$ /opt/pgsql/bin/repmgr -f /opt/repmgr.conf \
node rejoin -d'host=192.168.99.202 port=6000 user=repmgr dbname=repmgr connect_timeout=2'
复制

本文基于Java Web应用进行node1和node2交替测试，应用层数据读写一切正常，不过基于数据库序列主键有跳变现象。