前言：
redis在主从模式下，如果master节点出现故障，redis不会进行切换主节点。如果要测试redis sentinel，首先要启动redis主从，看我的另一篇博客。
redis sentinel主要功能：监控redis的运行状态，若master节点出现问题，能够选举一个slave为主节点，并修改其他从节点的主节点地址。单个sentinel是不可靠的，sentinel也是支持集群的，因为50%以上投票才生效，所以至少要3个sentinel节点。
配置：
port端口号
bind 192.168.94.151 和redis的bind一样
daemonize yes 后台运行
sentinel monitor mymaster 192.168.94.151 7051 2 mymaster为自己起的集群的名字，192.168.94.151 master的ip 7051master的端口号 2 当有两个sentinel节点认为主节点出问题是，就可以发起failover，如果50%以上节点同意，就切换主节点成功
sentinel auth-pass mymaster xuhaixing redis主节点的密码
sentinel down-after-milliseconds <master-name> <milliseconds> 在一定的时间内，没有PING通master节点，这个节点就认为master已经down，将其标记为主观下线
sentinel failover-timeout <master-name> <milliseconds> 在该时间内未完成failover，则failover失败
sentinel parallel-syncs <master-name> <numslaves> 在执行故障转移时最多可以有多少个从节点同步新的主master
sentinel notification-script <master-name> <script-path> 在异常时可以执行调用脚本
启动redis(主从部署看我另一篇文章)
[root@localhost redis-replicate]# redis-cli -h 192.168.94.151 -p 7051 -a xuhaixing
192.168.94.151:7051> info replication
# Replication
role:master
connected_slaves:2
slave0:ip=192.168.94.151,port=7052,state=online,offset=56,lag=1
slave1:ip=192.168.94.151,port=7050,state=online,offset=56,lag=1
master_replid:dd924d2fb0001b41eca962f2030aa4694d572235
master_replid2:0000000000000000000000000000000000000000
master_repl_offset:56
second_repl_offset:-1
repl_backlog_active:1
repl_backlog_size:1048576
repl_backlog_first_byte_offset:1
repl_backlog_histlen:56
192.168.94.151:7051>
启动sentinel
[root@localhost redis-replicate]# cat sentinel-start.sh
cd /usr/local/redis-replicate/7050
./redis-sentinel ./sentinel.conf
cd /usr/local/redis-replicate/7051
./redis-sentinel ./sentinel.conf
cd /usr/local/redis-replicate/7052
./redis-sentinel ./sentinel.conf
[root@localhost redis-replicate]# sh sentinel-start.sh
1764:X 26 Jul 06:08:53.076 # oO0OoO0OoO0Oo Redis is starting oO0OoO0OoO0Oo
1764:X 26 Jul 06:08:53.076 # Redis version=4.0.0, bits=64, commit=00000000, modified=0, pid=1764, just started
1764:X 26 Jul 06:08:53.076 # Configuration loaded
1766:X 26 Jul 06:08:53.099 # oO0OoO0OoO0Oo Redis is starting oO0OoO0OoO0Oo
1766:X 26 Jul 06:08:53.099 # Redis version=4.0.0, bits=64, commit=00000000, modified=0, pid=1766, just started
1766:X 26 Jul 06:08:53.099 # Configuration loaded
1771:X 26 Jul 06:08:53.122 # oO0OoO0OoO0Oo Redis is starting oO0OoO0OoO0Oo
1771:X 26 Jul 06:08:53.122 # Redis version=4.0.0, bits=64, commit=00000000, modified=0, pid=1771, just started
1771:X 26 Jul 06:08:53.122 # Configuration loaded
[root@localhost redis-replicate]# ps -ef|grep redis
root 1005 1 0 04:44 ? 00:00:07 /usr/local/redis/bin/redis-server 0.0.0.0:6379
root 1741 1 0 06:06 ? 00:00:00 ./redis-server 192.168.94.151:7050
root 1746 1 0 06:06 ? 00:00:00 ./redis-server 192.168.94.151:7051
root 1751 1 0 06:06 ? 00:00:00 ./redis-server 192.168.94.151:7052
root 1765 1 0 06:08 ? 00:00:00 ./redis-sentinel 192.168.94.151:27050 [sentinel]
root 1770 1 0 06:08 ? 00:00:00 ./redis-sentinel 192.168.94.151:27051 [sentinel]
root 1775 1 0 06:08 ? 00:00:00 ./redis-sentinel 192.168.94.151:27052 [sentinel]
root 1783 1537 0 06:10 pts/0 00:00:00 grep --color=auto redis
[root@localhost redis-replicate]#
[root@localhost redis-replicate]# redis-cli -h 192.168.94.151 -p 27051 -a xuhaixing
192.168.94.151:27051> info sentinel
# Sentinel
sentinel_masters:1
sentinel_tilt:0
sentinel_running_scripts:0
sentinel_scripts_queue_length:0
sentinel_simulate_failure_flags:0
master0:name=mymaster,status=ok,address=192.168.94.151:7051,slaves=2,sentinels=3
可以通过kill -9 进程号，干掉redis主节点，等一会，再重新查集群信息，会发现主节点已经变了
如果因为网络中断，会出现两个master的原因，可通过该redis配置
min-slaves-to-write 1
min-slaves-max-lag 10min-slaves-to-write表示至少一个从库时主库才可以写，否则返回错误
min-slaves-max-lag表示允许从库失去连接的时间