redis-3.28 一主二从模式介绍、主从从模式介绍、sentinel模式一主两从高可用

一、redis-3.28部署

yum install gcc tcl
wget http://download.redis.io/releases/redis-3.2.8.tar.gz
tar -zxvf redis-3.2.8.tar.gz
cd redis-3.2.8
mkdir -p /roobo/server/redis-3.2.8
make install PREFIX=/roobo/server/redis-3.2.8
ln -s /roobo/server/redis-3.2.8 /roobo/server/redis
cd /roobo/server/redis/bin

#启动redis
./redis-server

#!/bin/bash
cd /roobo/server/redis
mkdir -p /roobo/server/redis/6379
mkdir -p /roobo/server/redis/6380
mkdir -p /roobo/server/redis/6381
sleep 5
echo "##############6379配置文件####################"
cat > /roobo/server/redis/6379/6379.conf <<EOF
# Redis使用后台模式
daemonize yes

# 关闭保护模式
protected-mode no

# 注释以下内容开启远程访问
# bind 127.0.0.1

# 修改启动端口为6379
port 6379

# 修改pidfile指向路径
pidfile /roobo/server/redis/6379/redis_6379.pid
EOF
sleep 2
echo "##############6380配置文件####################"
cat > /roobo/server/redis/6380/6380.conf <<EOF
# Redis使用后台模式
daemonize yes

# 关闭保护模式
protected-mode no

# 注释以下内容开启远程访问
# bind 127.0.0.1

# 修改启动端口为6380
port 6380

# 修改pidfile指向路径
pidfile /roobo/server/redis/6380/redis_6380.pid
EOF
sleep 2
echo "##############6381配置文件####################"
cat > /roobo/server/redis/6381/6381.conf <<EOF
# Redis使用后台模式
daemonize yes

# 关闭保护模式
protected-mode no

# 注释以下内容开启远程访问
# bind 127.0.0.1

# 修改启动端口为6381
port 6381

# 修改pidfile指向路径
pidfile /roobo/server/redis/6381/redis_6381.pid
EOF

#启动三个实例
/roobo/server/redis/bin/redis-server /roobo/server/redis/6379/6379.conf
/roobo/server/redis/bin/redis-server /roobo/server/redis/6380/6380.conf
/roobo/server/redis/bin/redis-server /roobo/server/redis/6381/6381.conf

二、主从模式
一主--->二从

介绍在Redis中设置主从有2种方式：
1.在redis.conf中设置slaveof
a) slaveof <masterip> <masterport>

2、使用redis-cli客户端连接到redis服务，执行slaveof命令
a) slaveof <masterip> <masterport>

我们演示：第2种方式在重启后将失去主从复制关系。

主库 6371
从库 6380，6381

#设置6380和6381为6379的从库
#redis-cli
[root@instance-m4p7hj0t bin]# /roobo/server/redis/bin/redis-cli -h 172.16.0.4 -p 6380
172.16.0.4:6380> slaveof 172.16.0.4 6379
OK

[root@instance-m4p7hj0t bin]# /roobo/server/redis/bin/redis-cli -h 172.16.0.4 -p 6381
172.16.0.4:6381> slaveof 172.16.0.4 6379
OK

#查看主从，主库：6379 ，从库：6380，6381
[root@instance-m4p7hj0t ~]# /roobo/server/redis/bin/redis-cli -h 172.16.0.4 -p 6379
172.16.0.4:6379> info replication
# Replication
role:master
connected_slaves:2
slave0:ip=172.16.0.4,port=6380,state=online,offset=687,lag=1
slave1:ip=172.16.0.4,port=6381,state=online,offset=687,lag=1
master_repl_offset:687
repl_backlog_active:1
repl_backlog_size:1048576
repl_backlog_first_byte_offset:2
repl_backlog_histlen:686
172.16.0.4:6379> exit

role:角色信息
slaveX:从库信息
connected_slaves:从库数量

三、测试
在主库写入数据
[root@instance-m4p7hj0t bin]# /roobo/server/redis/bin/redis-cli -h 172.16.0.4 -p 6379
172.16.0.4:6379> set nulige 123
OK

在从库读取数据

#6380
[root@instance-m4p7hj0t ~]# /roobo/server/redis/bin/redis-cli -h 172.16.0.4 -p 6380
172.16.0.4:6380> get nulige
"123"

#从库6381
[root@instance-m4p7hj0t ~]# /roobo/server/redis/bin/redis-cli -h 172.16.0.4 -p 6381
172.16.0.4:6381> get nulige
"123"

一、需求
（1）、模拟6379宕机，把6380切换为主，6381为从
（2）、当6379修复好后，再把6379挂为6380的从库。

1、开始模拟6379宕机
/roobo/server/redis/bin/redis-cli -h 172.16.0.4 -p 6379

#关机（相当于宕机了）
shutdown

#取消主从关系
slave-read-only no

2、把6380切换为主
/roobo/server/redis/bin/redis-cli -h 172.16.0.4 -p 6380
#把从提升为主
172.16.0.4:6380> slaveof no one
OK
#查看
172.16.0.4:6380> info replication
# Replication
role:master
connected_slaves:0
master_repl_offset:0
repl_backlog_active:0
repl_backlog_size:1048576
repl_backlog_first_byte_offset:0
repl_backlog_histlen:0

3、设置6381为6380的从
/roobo/server/redis/bin/redis-cli -h 172.16.0.4 -p 6381
slaveof 172.16.0.4 6380

#再在6380上面查看（备注：现在6380为主，6381就为从）
/roobo/server/redis/bin/redis-cli -h 172.16.0.4 -p 6380
172.16.0.4:6380> info replication
# Replication
role:master
connected_slaves:1
slave0:ip=172.16.0.4,port=6381,state=online,offset=1,lag=0
master_repl_offset:1
repl_backlog_active:1
repl_backlog_size:1048576
repl_backlog_first_byte_offset:2
repl_backlog_histlen:0
172.16.0.4:6380>

4、6379修好了，再把他加到6380的主下面，变成从库。
[root@instance-m4p7hj0t ~]# /roobo/server/redis/bin/redis-server /roobo/server/redis/6379/6379.conf
[root@instance-m4p7hj0t ~]# /roobo/server/redis/bin/redis-cli -h 172.16.0.4 -p 6379
172.16.0.4:6379> slaveof 172.16.0.4 6380
OK

5、查看6380上面，现在已经是6380为主，6379和6381为从
/roobo/server/redis/bin/redis-cli -h 172.16.0.4 -p 6380
172.16.0.4:6380> info replication
# Replication
role:master
connected_slaves:2
slave0:ip=172.16.0.4,port=6381,state=online,offset=239,lag=1
slave1:ip=172.16.0.4,port=6379,state=online,offset=239,lag=1
master_repl_offset:239
repl_backlog_active:1
repl_backlog_size:1048576
repl_backlog_first_byte_offset:2
repl_backlog_histlen:238
172.16.0.4:6380> exit

四、设置从库只读
默认情况下redis数据库充当slave角色时是只读的不能进行写操作
可以在配置文件中开启非只读:
slave-read-only no

五、取消主从关系命令
SLAVEOF NO ONE
注:如果要取消Redis主从关系,可以在对应的从库执行SLAVEOF NO ONE命令,取消主从关系

六、开启无磁盘复制:
配置文件加参数：
repl-diskless-sync yes

七、主从从架构

后补。。。。。

八、redis3.2.8 sentinel模式一主两从高可用环境搭建测试

原理：

例如下图所示：

在Server1 掉线后：

升级Server2 为新的主服务器：

实现功能：
a、6379为主，6380为从库，6381为从库
b、6379为主，宕机自动切换到6381为主，再挂载6380和6379为从库

（备注：因演示环境只有一台机，所以只配置了一个sentinel.conf,一主两从环境，三台都要配置,并改参数为2，才能实现。）

实施步骤如下：
1、6379服务器为主，配置sentinel.conf
sentinel.conf
[root@instance-m4p7hj0t 6379]# pwd
/roobo/server/redis/6379
[root@instance-m4p7hj0t 6379]# ll
总用量 16
-rw-r--r-- 1 root root 253 8月 15 16:44 6379.conf
-rw-r--r-- 1 root root 91 8月 16 09:30 dump.rdb
-rw-r--r-- 1 root root 6 8月 16 09:30 redis_6379.pid
-rw-r--r-- 1 root root 192 8月 16 09:50 sentinel.conf

#sentinel配置文件
[root@instance-m4p7hj0t 6379]# cat sentinel.conf
port 26379
pidfile "/roobo/server/redis/6379/redis-sentinel.pid"
dir "/roobo/server/redis-3.2.8/6379"
daemonize yes
protected-mode no
logfile "/roobo/server/redis/6379/redis-sentinel.log"
#最后一位叁数1表示：sentinel只安装了一台机，多台机设为2
sentinel monitor redisMaster 172.16.0.4 6381 1
sentinel down-after-milliseconds redisMaster 10000
sentinel parallel-syncs redisMaster 1
sentinel failover-timeout redisMaster 60000

2、查看6379配置信息，确认6379为主，其它两节点为从
[root@instance-m4p7hj0t 6379]# /roobo/server/redis/bin/redis-cli -h 172.16.0.4 -p 6379
172.16.0.4:6379> info replication
# Replication
role:master
connected_slaves:2
slave0:ip=172.16.0.4,port=6381,state=online,offset=11016,lag=1
slave1:ip=172.16.0.4,port=6380,state=online,offset=11016,lag=1
master_repl_offset:11016
repl_backlog_active:1
repl_backlog_size:1048576
repl_backlog_first_byte_offset:2
repl_backlog_histlen:11015

3、启动6379节点的sentinel服务
[root@instance-m4p7hj0t bin]# /roobo/server/redis/bin/redis-sentinel /roobo/server/redis/6379/sentinel.conf

#切换目录到6379文件夹中
[root@instance-m4p7hj0t 6379]# cd /roobo/server/redis/6379
[root@instance-m4p7hj0t 6379]# ll
总用量 24
-rw-r--r-- 1 root root 253 8月 15 16:44 6379.conf
-rw-r--r-- 1 root root 91 8月 16 10:10 dump.rdb
-rw-r--r-- 1 root root 6 8月 16 09:30 redis_6379.pid
-rw-r--r-- 1 root root 1690 8月 16 10:14 redis-sentinel.log
-rw-r--r-- 1 root root 6 8月 16 10:14 redis-sentinel.pid
-rw-r--r-- 1 root root 611 8月 16 10:14 sentinel.conf

#查看日志
[root@instance-m4p7hj0t 6379]# tail -f redis-sentinel.log
`-._ `-._`-.__.-'_.-' _.-'
`-._ `-.__.-' _.-'
`-._ _.-'
`-.__.-'

4、演示6379宕机，6381原来是从库，会自动切换为主库（6381）
[root@instance-m4p7hj0t 6379]# /roobo/server/redis/bin/redis-cli -h 172.16.0.4 -p 6379
172.16.0.4:6379>shutdown (模拟宕机)

5、查看6379服务器上面的sentinel日志
[root@instance-m4p7hj0t 6379]# cd /roobo/server/redis/6379
[root@instance-m4p7hj0t 6379]# tail -f redis-sentinel.log
`-._ `-._`-.__.-'_.-' _.-'
`-._ `-.__.-' _.-'
`-._ _.-'
`-.__.-'

28048:X 16 Aug 10:14:00.614 # WARNING: The TCP backlog setting of 511 cannot be enforced because /proc/sys/net/core/somaxconn is set to the lower value of 128.
28048:X 16 Aug 10:14:00.628 # Sentinel ID is 9d235e0e09237ec250f33b00bc7369d080c4703a
28048:X 16 Aug 10:14:00.628 # +monitor master redisMaster 172.16.0.4 6379 quorum 1
28048:X 16 Aug 10:14:00.629 * +slave slave 172.16.0.4:6381 172.16.0.4 6381 @ redisMaster 172.16.0.4 6379
28048:X 16 Aug 10:14:00.632 * +slave slave 172.16.0.4:6380 172.16.0.4 6380 @ redisMaster 172.16.0.4 6379
28048:X 16 Aug 10:16:58.258 # +sdown master redisMaster 172.16.0.4 6379 #显示6379宕机了
28048:X 16 Aug 10:16:58.259 # +odown master redisMaster 172.16.0.4 6379 #quorum 1/1
28048:X 16 Aug 10:16:58.259 # +new-epoch 1
28048:X 16 Aug 10:16:58.260 # +try-failover master redisMaster 172.16.0.4 6379
28048:X 16 Aug 10:16:58.263 # +vote-for-leader 9d235e0e09237ec250f33b00bc7369d080c4703a 1
28048:X 16 Aug 10:16:58.263 # +elected-leader master redisMaster 172.16.0.4 6379
28048:X 16 Aug 10:16:58.264 # +failover-state-select-slave master redisMaster 172.16.0.4 6379
28048:X 16 Aug 10:16:58.326 # +selected-slave slave 172.16.0.4:6381 172.16.0.4 6381 @ redisMaster 172.16.0.4 6379
28048:X 16 Aug 10:16:58.326 * +failover-state-send-slaveof-noone slave 172.16.0.4:6381 172.16.0.4 6381 @ redisMaster4 6379
28048:X 16 Aug 10:16:58.378 * +failover-state-wait-promotion slave 172.16.0.4:6381 172.16.0.4 6381 @ redisMaster 17279
28048:X 16 Aug 10:16:58.722 # +promoted-slave slave 172.16.0.4:6381 172.16.0.4 6381 @ redisMaster 172.16.0.4 6379
28048:X 16 Aug 10:16:58.723 # +failover-state-reconf-slaves master redisMaster 172.16.0.4 6379
28048:X 16 Aug 10:16:58.795 * +slave-reconf-sent slave 172.16.0.4:6380 172.16.0.4 6380 @ redisMaster 172.16.0.4 6379
28048:X 16 Aug 10:16:59.776 * +slave-reconf-inprog slave 172.16.0.4:6380 172.16.0.4 6380 @ redisMaster 172.16.0.4 63
28048:X 16 Aug 10:17:00.786 * +slave-reconf-done slave 172.16.0.4:6380 172.16.0.4 6380 @ redisMaster 172.16.0.4 6379
28048:X 16 Aug 10:17:00.886 # +failover-end master redisMaster 172.16.0.4 6379
28048:X 16 Aug 10:17:00.887 # +switch-master redisMaster 172.16.0.4 6379 172.16.0.4 6381 #切换6381为主
28048:X 16 Aug 10:17:00.887 * +slave slave 172.16.0.4:6380 172.16.0.4 6380 @ redisMaster 172.16.0.4 6381
28048:X 16 Aug 10:17:00.887 * +slave slave 172.16.0.4:6379 172.16.0.4 6379 @ redisMaster 172.16.0.4 6381
28048:X 16 Aug 10:17:10.923 # +sdown slave 172.16.0.4:6379 172.16.0.4 6379 @ redisMaster 172.16.0.4 6381

6、如何确认6381为主，并加入6380为从
[root@instance-m4p7hj0t ~]# /roobo/server/redis/bin/redis-cli -h 172.16.0.4 -p 6381
172.16.0.4:6381> info replication
# Replication
role:master #当前6381为主
connected_slaves:1
slave0:ip=172.16.0.4,port=6380,state=online,offset=1998,lag=0 #6380为从
master_repl_offset:1998
repl_backlog_active:1
repl_backlog_size:1048576
repl_backlog_first_byte_offset:2
repl_backlog_histlen:1997

7、修改6379，再把他加为6381的从
#启动6379服务
[root@instance-m4p7hj0t ~]# /roobo/server/redis/bin/redis-server /roobo/server/redis/6379/6379.conf
#登录6379
[root@instance-m4p7hj0t ~]# /roobo/server/redis/bin/redis-cli -h 172.16.0.4 -p 6379
#加到6381为从
172.16.0.4:6379> slaveof 172.16.0.4 6381
OK

8、查看6381是否，加入6379和8380为从
[root@instance-m4p7hj0t ~]# /roobo/server/redis/bin/redis-cli -h 172.16.0.4 -p 6381
172.16.0.4:6381> info replication
# Replication
role:master
connected_slaves:2
slave0:ip=172.16.0.4,port=6380,state=online,offset=11173,lag=1
slave1:ip=172.16.0.4,port=6379,state=online,offset=11173,lag=1
master_repl_offset:11173
repl_backlog_active:1
repl_backlog_size:1048576
repl_backlog_first_byte_offset:2
repl_backlog_histlen:11172
172.16.0.4:6381> exit

这样就实现了，6379为主，宕机自动切换到6381为主，再挂载6380和6379为从的全部过程。

参考：
一主多从模式和主从从架构
https://blog.csdn.net/RobertoHuang/article/details/70766809

redis3.2.8 sentinel模式一主两从高可用环境搭建测试
https://blog.csdn.net/busbanana/article/details/71213812

redis集群模式 (集群三主三从模式)

https://blog.csdn.net/varyall/article/details/78856450