Redis Sentinel 介绍与部署

DBA小记 2020-10-27

1794

Redis Sentinel 介绍与部署

一、Sentinel介绍

主从复制的问题

Redis主从复制可将主节点数据同步给从节点，从节点此时有两个作用：

·一旦主节点宕机，从节点作为主节点的备份可以随时顶上来。

·扩展主节点的读能力，分担主节点读压力。

但是问题来了：

·一旦主节点宕机，从节点晋升成主节点，同时需要修改应用方的主节点地址，还需要命令所有从节点去复制新的主节点，整个过程需要人工干预。

·主节点的写能力受到单机的限制。

·主节点的存储能力受到单机的限制。

第一个问题，我们接下来讲的Sentinel就可以解决。而后两个问题，Redis也给出了方案Redis Cluster。

Redis Sentinel的高可用

Redis Sentinel是一个分布式架构，包含若干个Sentinel节点和Redis数据节点，每个Sentinel节点会对数据节点和其余Sentinel节点进行监控，当发现节点不可达时，会对节点做下线标识。

如果被标识的是主节点，他还会选择和其他Sentinel节点进行“协商”，当大多数的Sentinel节点都认为主节点不可达时，他们会选举出一个Sentinel节点来完成自动故障转移工作，同时将这个变化通知给Redis应用方。

整个过程完全自动，不需要人工介入，所以可以很好解决Redis的高可用问题。

接下来就通过部署一个Redis Sentinel实例来了解整体框架。

二、部署

环境介绍

分别有3个Sentinel节点，1个主节点，2个从节点组成一个Redis Sentinel。

集群机器                 redis服务                 哨兵(sentinel)               节点描述
虚拟机：172.16.86.79    安装redis服务,端口6379     安装sentinel服务,端口26379     主节点
虚拟机：172.16.86.80    安装redis服务,端口6379     安装sentinel服务,端口26379     从节点
虚拟机：172.16.86.81    安装redis服务, 端口6379    安装sentinel服务,端口26379     从节点

主节点（79）安装redis

1）安装路径

/app/redis

2）安装过程

上传文件redis-3.2.1.tar.gz到/app/tools-conf下

[root@DFJK-TEST-25 ~]# mkdir -p app/redis/
[root@DFJK-TEST-25 ~]# cd app/tools-conf/
[root@DFJK-TEST-25 tools-conf]# tar zxvf redis-3.2.1.tar.gz
[root@DFJK-TEST-25 tools-conf]# cd redis-3.2.1
[root@DFJK-TEST-25 redis-3.2.1]# make
[root@DFJK-TEST-25 redis-3.2.1]# make PREFIX=/app/redis install

3）更改配置文件

[root@DFJK-TEST-25 ~]# mkdir -p etc/redis
[root@DFJK-TEST-25 ~]# cd etc/red
[root@DFJK-TEST-25 ~]# cd etc/redis/
[root@DFJK-TEST-25 redis]# vi redis.conf
[root@DFJK-TEST-25 redis]# cat redis.conf
################################ GENERAL  #####################################


daemonize yes
pidfile "/var/run/redis.pid"
port 6379
tcp-backlog 511
timeout 0
tcp-keepalive 60
loglevel notice
logfile "/app/redis/redis_log"
databases 16


################################ SNAPSHOTTING  ################################


save 600 5000
stop-writes-on-bgsave-error no
rdbcompression no
rdbchecksum yes
dbfilename "dump.rdb"
dir "/app/redis/dump"


################################# REPLICATION #################################


# slaveof
masterauth "dfjk_qwe"
slave-serve-stale-data yes
slave-read-only no
repl-diskless-sync no
repl-diskless-sync-delay 5
repl-ping-slave-period 10
repl-disable-tcp-nodelay no
# repl-backlog-size 1mb
# repl-backlog-ttl 3600
slave-priority 100
# min-slaves-to-write 3
# min-slaves-max-lag 10
# min-slaves-max-lag is set to 10.


################################## SECURITY ###################################


requirepass "dfjk_qwe"


################################### LIMITS ####################################


# maxclients 10000
maxmemory 5859375kb
# maxmemory-policy noeviction
# maxmemory-samples 5


############################## APPEND ONLY MODE ###############################


appendonly no
appendfilename "appendonly.aof"
# appendfsync always
appendfsync everysec
# appendfsync no
no-appendfsync-on-rewrite no
auto-aof-rewrite-percentage 100
auto-aof-rewrite-min-size 64mb
aof-load-truncated yes
################################ LUA SCRIPTING  ###############################


lua-time-limit 5000


################################## SLOW LOG ###################################


slowlog-log-slower-than 10000
slowlog-max-len 128


################################ LATENCY MONITOR ##############################


latency-monitor-threshold 0


############################# EVENT NOTIFICATION ##############################


notify-keyspace-events ""


############################### ADVANCED CONFIG ###############################


hash-max-ziplist-entries 512
hash-max-ziplist-value 64
list-max-ziplist-entries 512
list-max-ziplist-value 64
set-max-intset-entries 512
zset-max-ziplist-entries 128
zset-max-ziplist-value 64
hll-sparse-max-bytes 3000
activerehashing yes
client-output-buffer-limit normal 0 0 0
client-output-buffer-limit slave 256mb 64mb 60
client-output-buffer-limit pubsub 32mb 8mb 60
hz 150
aof-rewrite-incremental-fsync yes
# Generated by CONFIG REWRITE

4）编辑系统启动文件

[root@DFJK-TEST-25 redis]## vi etc/profile
添加：
export PATH="$PATH:/app/redis/bin" 


[root@DFJK-TEST-25 redis]# cat etc/init.d/redis 
#!/bin/sh
#chkconfig: 2345 80 90 
# Simple Redis init.d script conceived to work on Linux systems
# as it does use of the proc filesystem.
PASSWD=dfjk_qwe
REDISPORT=6379
EXEC=/app/redis/bin/redis-server
CLIEXEC=/app/redis/bin/redis-cli


PIDFILE=/var/run/redis.pid
CONF="/etc/redis/redis.conf"


case "$1" in
    start)
        if [ -f $PIDFILE ]
        then
                echo "$PIDFILE exists, process is already running or crashed"
        else
                echo "Starting Redis server..."
                $EXEC $CONF &
        fi
        ;;
    stop)
        if [ ! -f $PIDFILE ]
        then
                echo "$PIDFILE does not exist, process is not running"
        else
                PID=$(cat $PIDFILE)
                echo "Stopping ..."
                $CLIEXEC -p $REDISPORT -a $PASSWD shutdown
                while [ -x proc/${PID} ]
                do
                    echo "Waiting for Redis to shutdown ..."
                    sleep 1
                done
                echo "Redis stopped"
        fi
        ;;
    *)
        echo "Please use start or stop as first argument"
        ;;
Esac

5）sentinel配置

cat /etc/redis/sentinel.conf
#  sentinel.conf
protected-mode no #关闭保护模式
daemonize yes
logfile "/app/redis/sentinel.log"
port 26379
#myid是自动生成（配置时不需要填写）
sentinel myid 6b3609b66b93922a036927591eba3622f39a5f6c
#自动生成
sentinel monitor mymaster 172.16.86.79 6379 2
#2表示在sentinel集群中只要有两个节点检测到redis主节点出故障就进行切换，单sentinel节点无效(自己测试发现的)   
sentinel down-after-milliseconds mymaster 5000
#如果5s内mymaster无响应，则认为mymaster宕机了  
sentinel failover-timeout mymaster 15000
#如果15秒后,mysater仍没活过来，则启动failover
sentinel auth-pass mymaster dfjk_qwe
#redis主节点密码
dir "/app/redis/tmp"
#指定工作目录


# Generated by CONFIG REWRITE
sentinel config-epoch mymaster 38
sentinel leader-epoch mymaster 2134
#以下是自动生成（配置时不需要填写）
sentinel known-slave mymaster 172.16.86.81 6379
sentinel known-slave mymaster 172.16.86.80 6379
sentinel known-sentinel mymaster 172.16.86.80 26379 4a1c268f5be39d175d78b583cbfaaae7cdcc71d5
sentinel known-sentinel mymaster 172.16.86.81 26379 b2022311b7e37e3fac9d404a3918b08d1bbea140 
sentinel current-epoch 2134

从节点80安装

1）安装路径

/app/redis

2）安装过程

上传文件redis-3.2.1.tar.gz到/app/tools-conf下

[root@DFJK-TEST-26 ~]# mkdir -p app/redis/
[root@DFJK-TEST-26 ~]# cd app/tools-conf/
[root@DFJK-TEST-26 tools-conf]# tar zxvf redis-3.2.1.tar.gz 
[root@DFJK-TEST-26 tools-conf]# cd redis-3.2.1
[root@DFJK-TEST-26 redis-3.2.1]# make  
[root@DFJK-TEST-26 redis-3.2.1]# make PREFIX=/app/redis install

3）更改配置文件

[root@DFJK-TEST-26 ~]# mkdir -p etc/redis
[root@DFJK-TEST-26 ~]# cd etc/redis/
[root@DFJK-TEST-26 redis]# vi redis.conf
################################ GENERAL  #####################################


daemonize yes
pidfile "/var/run/redis.pid"
port 6379
tcp-backlog 511
timeout 0
tcp-keepalive 60
loglevel notice
logfile "/app/redis/redis_log"
databases 16


################################ SNAPSHOTTING  ################################


save 600 5000
stop-writes-on-bgsave-error no
rdbcompression no
rdbchecksum yes
dbfilename "dump.rdb"
dir "/app/redis/dump"


################################# REPLICATION #################################


# slaveof <masterip> <masterport>
masterauth "dfjk_qwe"
slave-serve-stale-data yes
slave-read-only no
repl-diskless-sync no
repl-diskless-sync-delay 5
repl-ping-slave-period 10
repl-disable-tcp-nodelay no
# repl-backlog-size 1mb
# repl-backlog-ttl 3600
slave-priority 100
# min-slaves-to-write 3
# min-slaves-max-lag 10
# min-slaves-max-lag is set to 10.


################################## SECURITY ###################################
requirepass "dfjk_qwe"
################################### LIMITS ####################################
# maxclients 10000
maxmemory 5859375kb
# maxmemory-policy noeviction
# maxmemory-samples 5
############################## APPEND ONLY MODE ###############################
appendonly no
appendfilename "appendonly.aof"
# appendfsync always
appendfsync everysec
# appendfsync no
no-appendfsync-on-rewrite no
auto-aof-rewrite-percentage 100
auto-aof-rewrite-min-size 64mb
aof-load-truncated yes
################################ LUA SCRIPTING  ###############################
lua-time-limit 5000
################################## SLOW LOG ###################################
slowlog-log-slower-than 10000
slowlog-max-len 128
################################ LATENCY MONITOR ##############################
latency-monitor-threshold 0
############################# EVENT NOTIFICATION ##############################
notify-keyspace-events ""
############################### ADVANCED CONFIG ###############################
hash-max-ziplist-entries 512
hash-max-ziplist-value 64
list-max-ziplist-entries 512
list-max-ziplist-value 64
set-max-intset-entries 512
zset-max-ziplist-entries 128
zset-max-ziplist-value 64
hll-sparse-max-bytes 3000
activerehashing yes
client-output-buffer-limit normal 0 0 0
client-output-buffer-limit slave 256mb 64mb 60
client-output-buffer-limit pubsub 32mb 8mb 60
hz 150
aof-rewrite-incremental-fsync yes
# Generated by CONFIG REWRITE
slaveof 172.16.86.79 6379

4）编辑系统启动文件

[root@DFJK-TEST-26 redis]# vi etc/profile
添加：
export PATH="$PATH:/app/redis/bin" 
[root@DFJK-TEST-26 redis]# . etc/profile


[root@DFJK-TEST-26 redis]# cat etc/init.d/redis 
#!/bin/sh
#chkconfig: 2345 80 90 
# Simple Redis init.d script conceived to work on Linux systems
# as it does use of the proc filesystem.
PASSWD=dfjk_qwe
REDISPORT=6379
EXEC=/app/redis/bin/redis-server
CLIEXEC=/app/redis/bin/redis-cli


PIDFILE=/var/run/redis.pid
CONF="/etc/redis/redis.conf"


case "$1" in
    start)
        if [ -f $PIDFILE ]
        then
                echo "$PIDFILE exists, process is already running or crashed"
        else
                echo "Starting Redis server..."
                $EXEC $CONF &
        fi
        ;;
    stop)
        if [ ! -f $PIDFILE ]
        then
                echo "$PIDFILE does not exist, process is not running"
        else
                PID=$(cat $PIDFILE)
                echo "Stopping ..."
                $CLIEXEC -p $REDISPORT -a $PASSWD shutdown
                while [ -x proc/${PID} ]
                do
                    echo "Waiting for Redis to shutdown ..."
                    sleep 1
                done
                echo "Redis stopped"
        fi
        ;;
    *)
        echo "Please use start or stop as first argument"
        ;;
Esac

5）sentinel配置

[root@DFJK-TEST-26 ~]# vi etc/redis/sentinel.conf 
#  sentinel.conf
protected-mode no #关闭保护模式
daemonize yes
logfile "/app/redis/sentinel.log"
port 26379
#myid是自动生成（配置时不需要填写）
sentinel myid 4a1c268f5be39d175d78b583cbfaaae7cdcc71d5
sentinel monitor mymaster 172.16.86.79 6379 2
#2表示在sentinel集群中只要有两个节点检测到redis主节点出故障就进行切换，单sentinel节点无效(自己测试发现的)


sentinel down-after-milliseconds mymaster 5000
#如果5s内mymaster无响应，则认为mymaster宕机了


#redis主节点密码
dir "/app/redis/tmp"
#指定工作目录


# Generated by CONFIG REWRITE
sentinel failover-timeout mymaster 15000
#如果15秒后,mysater仍没活过来，则启动failover


sentinel auth-pass mymaster dfjk_qwe
sentinel config-epoch mymaster 38


sentinel leader-epoch mymaster 38
#以下是自动生成（配置时不需要填写）


sentinel known-slave mymaster 172.16.86.81 6379
sentinel known-slave mymaster 172.16.86.80 6379
sentinel known-sentinel mymaster 172.16.86.81 26379 b2022311b7e37e3fac9d404a3918b08d1bbea140
sentinel known-sentinel mymaster 172.16.86.79 26379 6b3609b66b93922a036927591eba3622f39a5f6c 
sentinel current-epoch 2134

从节点81安装

1）安装路径

/app/redis

2）安装过程

上传文件redis-3.2.1.tar.gz到/app/tools-conf下

[root@DFJK-TEST-27 ~]# mkdir -p app/redis/
[root@DFJK-TEST-27 ~]# cd /app/tools-conf/
[root@DFJK-TEST-27 tools-conf]# tar zxvf redis-3.2.1.tar.gz
[root@DFJK-TEST-27 tools-conf]# cd redis-3.2.1
[root@DFJK-TEST-27 redis-3.2.1]# make
[root@DFJK-TEST-27 redis-3.2.1]# make PREFIX=/app/redis install

3）更改配置文件

[root@DFJK-TEST-27 ~]# mkdir -p etc/redis
[root@DFJK-TEST-27 ~]# cd /etc/redis/
[root@DFJK-TEST-27 redis]# vi redis.conf
################################ GENERAL  #####################################
daemonize yes
pidfile "/var/run/redis.pid"
port 6379
tcp-backlog 511
timeout 0
tcp-keepalive 60
loglevel notice
logfile "/app/redis/redis_log"
databases 16


################################ SNAPSHOTTING  ################################
save 600 5000
stop-writes-on-bgsave-error no
rdbcompression no
rdbchecksum yes
dbfilename "dump.rdb"
dir "/app/redis/dump"
################################# REPLICATION #################################
# slaveof
masterauth "dfjk_qwe"
slave-serve-stale-data yes
slave-read-only no
repl-diskless-sync no
repl-diskless-sync-delay 5
repl-ping-slave-period 10
repl-disable-tcp-nodelay no
# repl-backlog-size 1mb
# repl-backlog-ttl 3600
slave-priority 100
# min-slaves-to-write 3
# min-slaves-max-lag 10
# min-slaves-max-lag is set to 10.
################################## SECURITY ###################################
requirepass "dfjk_qwe"
################################### LIMITS ####################################
# maxclients 10000
maxmemory 5859375kb
# maxmemory-policy noeviction
# maxmemory-samples 5
############################## APPEND ONLY MODE ###############################
appendonly no
appendfilename "appendonly.aof"
# appendfsync always
appendfsync everysec
# appendfsync no
no-appendfsync-on-rewrite no
auto-aof-rewrite-percentage 100
auto-aof-rewrite-min-size 64mb
aof-load-truncated yes
################################ LUA SCRIPTING  ###############################
lua-time-limit 5000
################################## SLOW LOG ###################################
slowlog-log-slower-than 10000
slowlog-max-len 128
################################ LATENCY MONITOR ##############################
latency-monitor-threshold 0
############################# EVENT NOTIFICATION ##############################
notify-keyspace-events ""
############################### ADVANCED CONFIG ###############################
hash-max-ziplist-entries 512
hash-max-ziplist-value 64
list-max-ziplist-entries 512
list-max-ziplist-value 64
set-max-intset-entries 512
zset-max-ziplist-entries 128
zset-max-ziplist-value 64
hll-sparse-max-bytes 3000
activerehashing yes
client-output-buffer-limit normal 0 0 0
client-output-buffer-limit slave 256mb 64mb 60
client-output-buffer-limit pubsub 32mb 8mb 60
hz 150
aof-rewrite-incremental-fsync yes
# Generated by CONFIG REWRITE
slaveof 172.16.86.79 6379

4）编辑系统启动文件

[root@DFJK-TEST-27 redis]# vi etc/profile
添加：
export PATH="$PATH:/app/redis/bin"
[root@DFJK-TEST-27 redis]# . etc/profile
[root@DFJK-TEST-27 redis]# cat etc/init.d/redis
#!/bin/sh
#chkconfig: 2345 80 90
# Simple Redis init.d script conceived to work on Linux systems
# as it does use of the proc filesystem.
PASSWD=dfjk_qwe
REDISPORT=6379
EXEC=/app/redis/bin/redis-server
CLIEXEC=/app/redis/bin/redis-cli
PIDFILE=/var/run/redis.pid
CONF="/etc/redis/redis.conf"
case "$1" in
    start)
        if [ -f $PIDFILE ]
        then
                echo "$PIDFILE exists, process is already running or crashed"
        else
                echo "Starting Redis server..."
                $EXEC $CONF &
        fi
        ;;
    stop)
        if [ ! -f $PIDFILE ]
        then
                echo "$PIDFILE does not exist, process is not running"
        else
                PID=$(cat $PIDFILE)
                echo "Stopping ..."
                $CLIEXEC -p $REDISPORT -a $PASSWD shutdown
                while [ -x proc/${PID} ]
                do
                    echo "Waiting for Redis to shutdown ..."
                    sleep 1
                done
                echo "Redis stopped"
        fi
        ;;
    *)
        echo "Please use start or stop as first argument"
        ;;
esac

5）sentinel配置

[root@DFJK-TEST-26 ~]# vi etc/redis/sentinel.conf
cat /etc/redis/sentinel.conf
#  sentinel.conf
protected-mode no #关闭保护模式
daemonize yes
logfile "/app/redis/sentinel.log"
port 26379
sentinel monitor mymaster 172.16.86.79 6379 2
#2表示在sentinel集群中只要有两个节点检测到redis主节点出故障就进行切换，单sentinel节点无效(自己测试发现的)  
sentinel down-after-milliseconds mymaster 5000
#如果5s内mymaster无响应，则认为mymaster宕机了  
sentinel failover-timeout mymaster 15000
#如果15秒后,mysater仍没活过来，则启动failover
sentinel auth-pass mymaster dfjk_qwe
#redis主节点密码
dir "/app/redis/tmp"
#指定工作目录


# Generated by CONFIG REWRITE
sentinel config-epoch mymaster 38
sentinel leader-epoch mymaster 38
sentinel known-slave mymaster 172.16.86.80 6379
sentinel known-slave mymaster 172.16.86.81 6379
sentinel known-sentinel mymaster 172.16.86.80 26379 4a1c268f5be39d175d78b583cbfaaae7cdcc71d5
sentinel known-sentinel mymaster 172.16.86.79 26379 6b3609b66b93922a036927591eba3622f39a5f6c
sentinel current-epoch 2134

启动并检测

1）主redis（79）启动

[root@DFJK-TEST-25 ~]# chmod +x etc/init.d/redis
[root@DFJK-TEST-25 ~]# mkdir -p app/redis/dump
[root@DFJK-TEST-25 redis]# mkdir -p app/redis/tmp
[root@DFJK-TEST-25 redis]# service redis start
Starting Redis server...
[root@DFJK-TEST-25 redis]# ps -ef |grep -v grep |grep redis
root     27816     1  0 09:53 ?        00:00:00 app/redis/bin/redis-server *:6379

2）主redis（79）启动sentinel

[root@DFJK-TEST-25 redis]# redis-sentinel etc/redis/sentinel.conf
[root@DFJK-TEST-25 redis]# ps -ef |grep -v grep |grep sentinel
root     27823     1  0 09:53 ?        00:00:00 redis-sentinel *:26379 [sentinel]

3）从redis（80）启动

[root@DFJK-TEST-26 ~]# chmod +x etc/init.d/redis
[root@DFJK-TEST-26 ~]# mkdir -p app/redis/dump
[root@DFJK-TEST-26 redis]# mkdir -p app/redis/tmp
[root@DFJK-TEST-26 redis]# service redis start
Starting Redis server...
[root@DFJK-TEST-26 redis]# ps -ef |grep -v grep |grep redis
root     20247     1  0 09:54 ?        00:00:00 app/redis/bin/redis-server *:6379

4）从redis（80）启动sentinel

[root@DFJK-TEST-26 redis]# redis-sentinel etc/redis/sentinel.conf
[root@DFJK-TEST-26 redis]# ps -ef |grep -v grep |grep sentinel
root     20261     1  0 09:55 ?        00:00:00 redis-sentinel *:26379 [sentinel]

5）从redis（81）启动

[root@DFJK-TEST-27 ~]# chmod +x etc/init.d/redis
[root@DFJK-TEST-27 ~]# mkdir -p app/redis/dump
[root@DFJK-TEST-27 redis]# mkdir -p app/redis/tmp
[root@DFJK-TEST-27 redis]# service redis start
Starting Redis server...
[root@DFJK-TEST-27 ~]# ps -ef |grep -v grep |grep redis
root     10869     1  0 09:55 ?        00:00:00 /app/redis/bin/redis-server *:6379

6）从redis（81）启动sentinel

[root@DFJK-TEST-27 ~]# redis-sentinel /etc/redis/sentinel.conf
[root@DFJK-TEST-27 ~]# ps -ef |grep -v grep |grep sentinel
root     10882     1  0 09:56 ?        00:00:00 redis-sentinel *:26379 [sentinel]

7）ping命令查看是否启动

[root@DFJK-TEST-25 ~]# redis-cli  -h 172.16.86.80 -p 6379 -a "dfjk_qwe" ping
PONG
[root@DFJK-TEST-25 ~]# redis-cli  -h 172.16.86.81 -p 6379 -a "dfjk_qwe" ping
PONG
[root@DFJK-TEST-25 ~]# redis-cli  -h 172.16.86.79 -p 6379 -a "dfjk_qwe" ping
PONG

确定主从关系

部署三个Sentinel节点之后，真个拓扑结构如图所示：

验证复制

1）主节点视角

[root@DFJK-TEST-25 ~]# redis-cli  -h 172.16.86.79 -p 6379 -a "dfjk_qwe" INFO replication
# Replication
role:master
connected_slaves:2
slave0:ip=172.16.86.80,port=6379,state=online,offset=57709,lag=1
slave1:ip=172.16.86.81,port=6379,state=online,offset=57995,lag=0
master_repl_offset:58138
repl_backlog_active:1
repl_backlog_size:1048576
repl_backlog_first_byte_offset:2
repl_backlog_histlen:58137

2）从节点1视角

[root@DFJK-TEST-26 ~]# redis-cli  -h 172.16.86.80 -p 6379 -a "dfjk_qwe" INFO replication
# Replication
role:slave
master_host:172.16.86.79
master_port:6379
master_link_status:up
master_last_io_seconds_ago:2
master_sync_in_progress:0
slave_repl_offset:62027
slave_priority:100
slave_read_only:0
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）从节点2视角

[root@DFJK-TEST-27 ~]# redis-cli  -h 172.16.86.81 -p 6379 -a "dfjk_qwe" INFO replication
# Replication
role:slave
master_host:172.16.86.79
master_port:6379
master_link_status:up
master_last_io_seconds_ago:1
master_sync_in_progress:0
slave_repl_offset:68075
slave_priority:100
slave_read_only:0
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

主写入：

[root@DFJK-TEST-25 ~]# redis-cli -h 172.16.86.79 -p 6379 -a "dfjk_qwe"

172.16.86.79:6379> set name xiaoming

172.16.86.79:6379> get name

"xiaoming"

从查看：

[root@DFJK-TEST-26 ~]# redis-cli -h 172.16.86.80 -p 6379 -a "dfjk_qwe"

172.16.86.80:6379> get name

"xiaoming"

[root@DFJK-TEST-27 ~]# redis-cli -h 172.16.86.81 -p 6379 -a "dfjk_qwe"

172.16.86.81:6379> get name

"xiaoming"

故障验证

故障转移实验

主停止redis：

[root@DFJK-TEST-25 ~]# service redis stop

主sentinel日志：

27823:X 24 May 10:07:17.912 # +sdown master mymaster 172.16.86.79 6379
27823:X 24 May 10:07:17.912 # +odown master mymaster 172.16.86.79 6379 #quorum 1/1
27823:X 24 May 10:07:17.912 # +new-epoch 2135
27823:X 24 May 10:07:17.912 # +try-failover master mymaster 172.16.86.79 6379
27823:X 24 May 10:07:17.913 # +vote-for-leader 6b3609b66b93922a036927591eba3622f39a5f6c 2135
27823:X 24 May 10:07:28.713 # -failover-abort-not-elected master mymaster 172.16.86.79 6379
27823:X 24 May 10:07:28.771 # Next failover delay: I will not start a failover before Thu May 24 10:07:48 2018
27823:X 24 May 10:07:48.750 # +new-epoch 2136
27823:X 24 May 10:07:48.750 # +try-failover master mymaster 172.16.86.79 6379
27823:X 24 May 10:07:48.751 # +vote-for-leader 6b3609b66b93922a036927591eba3622f39a5f6c 2136
27823:X 24 May 10:07:59.020 # -failover-abort-not-elected master mymaster 172.16.86.79 6379
27823:X 24 May 10:07:59.075 # Next failover delay: I will not start a failover before Thu May 24 10:08:18 2018
27823:X 24 May 10:08:18.996 # +new-epoch 2137
27823:X 24 May 10:08:18.996 # +try-failover master mymaster 172.16.86.79 6379
27823:X 24 May 10:08:18.998 # +vote-for-leader 6b3609b66b93922a036927591eba3622f39a5f6c 2137
27823:X 24 May 10:08:29.936 # -failover-abort-not-elected master mymaster 172.16.86.79 6379
27823:X 24 May 10:08:29.994 # Next failover delay: I will not start a failover before Thu May 24 10:08:49 2018

从1的sentinel日志：

20261:X 24 May 10:07:17.938 # +sdown master mymaster 172.16.86.79 6379
20261:X 24 May 10:07:19.079 # +new-epoch 2135
20261:X 24 May 10:07:50.122 # +new-epoch 2136
20261:X 24 May 10:08:19.166 # +new-epoch 2137
20261:X 24 May 10:08:50.772 # +new-epoch 2138

从2的sentinel日志：

10882:X 24 May 10:07:17.990 # +sdown master mymaster 172.16.86.79 6379
10882:X 24 May 10:07:19.078 # +new-epoch 2135
10882:X 24 May 10:07:50.121 # +new-epoch 2136
10882:X 24 May 10:08:19.165 # +new-epoch 2137
10882:X 24 May 10:08:50.777 # +new-epoch 2138

此时，Redis Sentinel
对主节点进行客观下线（Objectively Down，简称 ODOWN）的判断，确认主节点不可达，则通知从节点中止复制主节点的操作。

当主节点下线时长超过配置的下线时长5
000
毫秒
，Redis Sentinel
执行故障转移操作。

查看一下Sentinel节点监控的主节点信息：

[root@DFJK-TEST-25 ~]# redis-cli  -h 172.16.86.79 -p 26379
172.16.86.79:26379> sentinel masters
1)  1) "name"
    2) "mymaster"
    3) "ip"
    4) "172.16.86.80"
    5) "port"
    6) "6379"
    7) "runid"
    8) "1ece1ee74e6a172c4522a34a2de0df8c94f24fed"
    9) "flags"
   10) "master"
   11) "link-pending-commands"
   12) "0"
   13) "link-refcount"
   14) "1"
   15) "last-ping-sent"
   16) "0"
   17) "last-ok-ping-reply"
   18) "231"
   19) "last-ping-reply"
   20) "231"
   21) "down-after-milliseconds"
   22) "5000"
   23) "info-refresh"
   24) "5219"
   25) "role-reported"
   26) "master"
   27) "role-reported-time"
   28) "35728"
   29) "config-epoch"
   30) "2158"
   31) "num-slaves"
   32) "2"
   33) "num-other-sentinels"
   34) "2"
   35) "quorum"
   36) "1"
   37) "failover-timeout"
   38) "15000"
   39) "parallel-syncs"
   40) "1"

看一下Sentinel节点监控的从节点信息：

172.16.86.79:26379> sentinel slaves mymaster
1)  1) "name"
    2) "172.16.86.79:6379"
    3) "ip"
    4) "172.16.86.79"
    5) "port"
    6) "6379"
    7) "runid"
    8) ""
    9) "flags"
   10) "s_down,slave,disconnected"  //端口79：6379的原主节点已经断开了连接
   11) "link-pending-commands"
   12) "2"
   13) "link-refcount"
   14) "1"
   15) "last-ping-sent"
   16) "227506"
   17) "last-ok-ping-reply"
   18) "227506"
   19) "last-ping-reply"
   20) "227506"
   21) "s-down-time"
   22) "222473"
   23) "down-after-milliseconds"
   24) "5000"
   25) "info-refresh"
   26) "1527129371700"
   27) "role-reported"
   28) "slave"
   29) "role-reported-time"
   30) "227506"
   31) "master-link-down-time"
   32) "0"
   33) "master-link-status"
   34) "err"
   35) "master-host"
   36) "?"
   37) "master-port"
   38) "0"
   39) "slave-priority"
   40) "100"
   41) "slave-repl-offset"
   42) "0"
2)  1) "name"
    2) "172.16.86.81:6379"
    3) "ip"
    4) "172.16.86.81"
    5) "port"
    6) "6379"
    7) "runid"
    8) "b8dd9e7be4c1ea13e15f19f50b4a81647568b571"
    9) "flags"
   10) "slave"
   11) "link-pending-commands"
   12) "0"
   13) "link-refcount"
   14) "1"
   15) "last-ping-sent"
   16) "0"
   17) "last-ok-ping-reply"
   18) "338"
   19) "last-ping-reply"
   20) "338"
   21) "down-after-milliseconds"
   22) "5000"
   23) "info-refresh"
   24) "6713"
   25) "role-reported"
   26) "slave"
   27) "role-reported-time"
   28) "227506"
   29) "master-link-down-time"
   30) "0"
   31) "master-link-status"
   32) "ok"
   33) "master-host"
   34) "172.16.86.80"
   35) "master-port"
   36) "6379"
   37) "slave-priority"
   38) "100"
   39) "slave-repl-offset"
   40) "46732"

由以上信息可得，端口为80
的Redis数据节点成为新的主节点，端口为79
的旧主节点断开连接。如图所示：

重启79
的数据节点：

[root@DFJK-TEST-25 ~]# service redis start
Starting Redis server...

查看下变化：

172.16.86.79:26379> sentinel slaves mymaster
1)  1) "name"
    2) "172.16.86.79:6379" //79的节点重启后，变成了"活"的从节点
    3) "ip"
    4) "172.16.86.79"
    5) "port"
    6) "6379"
    7) "runid"
    8) "30876d99252c019af65ba40de62c42c5bc39d0dd"
    9) "flags"
   10) "slave"
   11) "link-pending-commands"
   12) "0"
   13) "link-refcount"
   14) "1"
   15) "last-ping-sent"
   16) "0"
   17) "last-ok-ping-reply"
   18) "605"
   19) "last-ping-reply"
   20) "605"
   21) "down-after-milliseconds"
   22) "5000"
   23) "info-refresh"
   24) "7726"
   25) "role-reported"
   26) "slave"
   27) "role-reported-time"
   28) "37834"
   29) "master-link-down-time"
   30) "0"
   31) "master-link-status"
   32) "ok"
   33) "master-host"
   34) "172.16.86.80"
   35) "master-port"
   36) "6379"
   37) "slave-priority"
   38) "100"
   39) "slave-repl-offset"
   40) "112149"
2)  1) "name"
    2) "172.16.86.81:6379"  //81的节点无变化，还是从节点
    3) "ip"
    4) "172.16.86.81"
    5) "port"
    6) "6379"
    7) "runid"
    8) "b8dd9e7be4c1ea13e15f19f50b4a81647568b571"
    9) "flags"
   10) "slave"
   11) "link-pending-commands"
   12) "0"
   13) "link-refcount"
   14) "1"
   15) "last-ping-sent"
   16) "0"
   17) "last-ok-ping-reply"
   18) "606"
   19) "last-ping-reply"
   20) "606"
   21) "down-after-milliseconds"
   22) "5000"
   23) "info-refresh"
   24) "3912"
   25) "role-reported"
   26) "slave"
   27) "role-reported-time"
   28) "536124"
   29) "master-link-down-time"
   30) "0"
   31) "master-link-status"
   32) "ok"
   33) "master-host"
   34) "172.16.86.80"
   35) "master-port"
   36) "6379"
   37) "slave-priority"
   38) "100"
   39) "slave-repl-offset"
   40) "113033"

79被降为80的从节点：

[root@DFJK-TEST-25 ~]# redis-cli  -h 172.16.86.79 -p 6379 -a "dfjk_qwe"  INFO replication
# Replication
role:slave
master_host:172.16.86.80
master_port:6379
master_link_status:up
master_last_io_seconds_ago:2
master_sync_in_progress:0
slave_repl_offset:133604
slave_priority:100
slave_read_only:0
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


[root@DFJK-TEST-25 ~]# redis-cli  -h 172.16.86.80 -p 6379 -a "dfjk_qwe"  INFO replication
# Replication
role:master
connected_slaves:2
slave0:ip=172.16.86.81,port=6379,state=online,offset=151565,lag=0
slave1:ip=172.16.86.79,port=6379,state=online,offset=151710,lag=0
master_repl_offset:152000
repl_backlog_active:1
repl_backlog_size:1048576
repl_backlog_first_byte_offset:2
repl_backlog_histlen:151999


[root@DFJK-TEST-25 ~]# redis-cli  -h 172.16.86.81 -p 6379 -a "dfjk_qwe"  INFO replication
# Replication
role:slave
master_host:172.16.86.80
master_port:6379
master_link_status:up
master_last_io_seconds_ago:1
master_sync_in_progress:0
slave_repl_offset:152870
slave_priority:100
slave_read_only:0
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

从上面的逻辑架构和故障转移试验中，可以看出Redis Sentinel的以下几个功能。

·监控：Sentinel节点会定期检测Redis数据节点和其余Sentinel节点是否可达。

·通知：Sentinel节点会将故障转移通知给应用方。

·主节点故障转移：实现从节点晋升为主节点并维护后续正确的主从关系。

·配置提供者：在Redis Sentinel结构中，客户端在初始化的时候连接的是Sentinel节点集合，从中获取主节点信息。

Sentinel配置说明

·sentinel monitor mymaster 127.0.0.1 6379 2

·当前Sentinel节点监控 127.0.0.1:6379 这个主节点

·2代表判断主节点失败至少需要2个Sentinel节点节点同意

·mymaster是主节点的别名

·sentinel down-after-milliseconds mymaster 30000

·每个Sentinel节点都要定期PING命令来判断Redis数据节点和其余Sentinel节点是否可达，如果超过30000毫秒且没有回复，则判定不可达

·sentinel parallel-syncs mymaster 1

·当Sentinel节点集合对主节点故障判定达成一致时，Sentinel领导者节点会做故障转移操作，选出新的主节点，原来的从节点会向新的主节点发起复制操作，限制每次向新的主节点发起复制操作的从节点个数为1。

·sentinel failover-timeout mymaster 180000

·故障转移超时时间为180000

·sentinel auth-pass \ \

·如果Sentinel
监控的主节点配置了密码，可以通过sentinel auth-pass
配置通过添加主节点的密码，防止Sentinel
节点无法对主节点进行监控。

·例如：sentinel auth-pass mymaster MySUPER--secret-0123passw0rd

·sentinel notification-script \ \

·在故障转移期间，当一些警告级别的Sentinel
事件发生（指重要事件，如主观下线，客观下线等）时，会触发对应路径的脚本，想脚本发送相应的事件参数。

·例如：sentinel notification-script mymaster /var/redis/notify.sh

·sentinel client-reconfig-script \ \

·在故障转移结束后，触发应对路径的脚本，并向脚本发送故障转移结果的参数。

·例如：sentinel client-reconfig-script mymaster /var/redis/reconfig.sh
。

redis

文章转载自DBA小记，如果涉嫌侵权，请发送邮件至：contact@modb.pro进行举报，并提供相关证据，一经查实，墨天轮将立刻删除相关内容。

Redis Sentinel 介绍与部署

Redis Sentinel 介绍与部署

一、Sentinel介绍

主从复制的问题

Redis Sentinel的高可用

二、部署

环境介绍

主节点（79）安装redis

从节点81安装

确定主从关系

验证复制

故障验证

故障转移实验

Sentinel配置说明

评论