centos6搭建redis集群搭建（单机多节点）

一、安装redis

1、安装gcc环境

yum install gcc-c++

2、下载源码包并解压

wget http://download.redis.io/releases/redis-3.2.4.tar.gz
tar -zxvf redis-3.2..tar.gz
cd redis-3.2.

3、安装ruby2.4.1并编译（一定要安装大于2.2.2的ruby，要不然redis在make的时候会报错）

wget http://ftp.ruby-lang.org/pub/ruby/2.4/ruby-2.4.1.tar.gz
tar -zxvf ruby-2.4.1.tar.gz
cd ruby-2.4.1
mkdir -p /usr/local/ruby
./configure --prefix=/usr/local/ruby
make && make install
ln -s /usr/local/ruby/bin/ruby /usr/bin/ruby

4、安装

make
make install PREFIX=/usr/local/redis
cp redis.conf /usr/local/redis/bin/redis.conf
chmod /usr/local/redis/bin/redis.conf

5、启动redis

./redis-server redis.conf

6、验证

[root@host---- bin]# ./redis-cli -p
127.0.0.1:> get
(error) ERR wrong number of arguments for 'get' command
127.0.0.1:> set aa hzb
OK
127.0.0.1:> get aa
"hzb"

二、搭建redis集群（集群模式）

1、在/usr/local/下面创建redis-cluster目录

mkdir -p /usr/local/redis-cluster

2、复制/usr/local/redis/bin目录到redis-cluster里面并重命名为redis1

cd /usr/local
cp -r redis/bin redis-cluster/redis1

3、将redis1在同目录复制3份分别起名redis2,redis3,redis4,redis5,redis6（必须6个节点以上才能创建集群）

[root@host---- redis-cluster]# pwd
/usr/local/redis-cluster
[root@host---- redis-cluster]# cp -r redis1 redis2
[root@host---- redis-cluster]# cp -r redis1 redis3
[root@host---- redis-cluster]# cp -r redis1 redis4
[root@host---- redis-cluster]# cp -r redis1 redis5
[root@host---- redis-cluster]# cp -r redis1 redis6

分别修改为redis-7001.conf,redis-7002.conf 后面依次到redis-7006.conf

以redis-7001为例：

daemonize yes
Port
logfile "./redis-7001.log"
protected-mode no
pidfile /var/run/redis_7001.pid
cluster-enabled yes

4、将redis-trib.rb复到制redis-cluster目录里面

[root@host---- redis-cluster]# cp /root/redis-3.2./src/redis-trib.rb .
[root@host---- redis-cluster]# ll
total
drwxr-xr-x  root root   Oct  : redis1
drwxr-xr-x  root root   Oct  : redis2
drwxr-xr-x  root root   Oct  : redis3
drwxr-xr-x  root root   Oct  : redis4
drwxr-xr-x  root root   Oct  : redis5
drwxr-xr-x  root root   Oct  : redis6
-rwxr-xr-x  root root  Oct  : redis-trib.rb

5、安装gem

 yum install rubygems -y

6、安装ruby的redis包

gem install redis

上面这一步可能会出现：

ERROR: Loading command: install (LoadError)
cannot load such file -- zlib
ERROR: While executing gem ... (NoMethodError)
undefined method `invoke_with_build_args' for nil:NilClass

解决办法是：

1. 进入ruby源码文件夹

1. 安装ruby自身提供的zlib包

#cd ext/zlib
#ruby ./extconf.rb
#make
#make install

1. 进入ruby源码文件夹

1. 安装ruby自身提供的openssl包

#cd ext/openssl
#ruby ./extconf.rb
#make
#make install

如果执行make的时候报错：make: *** No rule to make target ，

错误原因：makefile里面的文件依赖关系有问题，导致头文件找不到。

解决办法：打开编译相关模块的makefile文件，找到.o文件的依赖关系，找到.h文件的路径，查看是否出错并改正。

原因分析：这种情况有时候是因为工程所在的目录改变，例如原来在C盘，现在移到了D盘，而makefile里面的.h文件路径没有改变

以上问题解决后再执行：

[root@host---- include]# gem install redis
Fetching: redis-4.0..gem (%)
Successfully installed redis-4.0.
Parsing documentation for redis-4.0.
Installing ri documentation for redis-4.0.
Done installing documentation for redis after  seconds
 gem installed

表示ruby的redis依赖包安装成功。

7、分别进入redis1,redis2,redis3,redis4,redis5,redis6里面执行

./redis-server redis-7001.conf
......

将四个节点启动

[root@host---- redis4]# ps -ef|grep redis
root              : ?        :: ./redis-server 127.0.0.1: [cluster]
root              : ?        :: ./redis-server 127.0.0.1: [cluster]
root              : ?        :: ./redis-server 127.0.0.1: [cluster]
root              : ?        :: ./redis-server 127.0.0.1: [cluster]
root              : ?        :: ./redis-server 127.0.0.1: [cluster]
root              : ?        :: ./redis-server 127.0.0.1: [cluster]
root           : pts/    :: grep redis

8、用redis-trib.rb构建集群

cd /usr/local/redis-cluster/

[root@host---- redis-cluster]# ./redis-trib.rb create --replicas  172.16.80.177: 172.16.80.177: 172.16.80.177: 172.16.80.177:7004 172.16.80.177:7005 172.16.80.177:7006
>>> Creating cluster
[ERR] Sorry, can't connect to node 172.16.80.177:7001

以上错误原因是redis.conf里面未开起远程访问

打开每个redis里面的.conf文件，找到

# JUST COMMENT THE FOLLOWING LINE.
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
bind 127.0.0.1

# Protected mode is a layer of security protection, in order to avoid that
# Redis instances left open on the internet are accessed and exploited.

将上面的bind 127.0.0.1改成bind 172.16.80.177(外部可以访问的ip)，然后重启各服务

root              : ?        :: ./redis-server 172.16.80.177: [cluster]
root              : ?        :: ./redis-server 172.16.80.177: [cluster]
root              : ?        :: ./redis-server 172.16.80.177: [cluster]
root              : ?        :: ./redis-server 172.16.80.177: [cluster]
root              : ?        :: ./redis-server 172.16.80.177: [cluster]
root              : ?        :: ./redis-server 172.16.80.177: [cluster]
root           : pts/    :: grep redis

再用redis-trib.rb创建集群：

[root@host---- redis-cluster]# ./redis-trib.rb create --replicas  172.16.80.177: 172.16.80.177: 172.16.80.177: 172.16.80.177: 172.16.80.177: 172.16.80.177:
>>> Creating cluster
>>> Performing hash slots allocation on  nodes...
Using  masters:
172.16.80.177:
172.16.80.177:
172.16.80.177:
Adding replica 172.16.80.177: to 172.16.80.177:
Adding replica 172.16.80.177: to 172.16.80.177:
Adding replica 172.16.80.177: to 172.16.80.177:
M: 40bbe8a979fe39f1f0941cc8e118443436fc71d8 172.16.80.177:
   slots:- ( slots) master
M: e6b76ee1540c9be6c44abc3dec42378f8cfd0191 172.16.80.177:
   slots:- ( slots) master
M: fdc248e24bd9f8690c652213a5781f4e567208e8 172.16.80.177:
   slots:- ( slots) master
S: 38faede9dd348983b1b8b80edc975e8d34105e5c 172.16.80.177:
   replicates 40bbe8a979fe39f1f0941cc8e118443436fc71d8
S: 38faede9dd348983b1b8b80edc975e8d34105e5c 172.16.80.177:
   replicates e6b76ee1540c9be6c44abc3dec42378f8cfd0191
S: 38faede9dd348983b1b8b80edc975e8d34105e5c 172.16.80.177:
   replicates fdc248e24bd9f8690c652213a5781f4e567208e8
Can I set the above configuration? (type 'yes' to accept): yes
>>> Nodes configuration updated
>>> Assign a different config epoch to each node
>>> Sending CLUSTER MEET messages to join the cluster
Waiting for the cluster to join...
>>> Performing Cluster Check (using node 172.16.80.177:)
M: 40bbe8a979fe39f1f0941cc8e118443436fc71d8 172.16.80.177:
   slots:- ( slots) master
    additional replica(s)
M: e6b76ee1540c9be6c44abc3dec42378f8cfd0191 172.16.80.177:
   slots:- ( slots) master
    additional replica(s)
S: 38faede9dd348983b1b8b80edc975e8d34105e5c 172.16.80.177:
   slots: ( slots) slave
   replicates 40bbe8a979fe39f1f0941cc8e118443436fc71d8
M: fdc248e24bd9f8690c652213a5781f4e567208e8 172.16.80.177:
   slots:- ( slots) master
    additional replica(s)
[OK] All nodes agree about slots configuration.
>>> Check for open slots...
>>> Check slots coverage...
[OK] All  slots covered.

至此，redis集群创建成功

9、验证集群是否成功

[root@host---- redis2]# ./redis-cli -h 172.16.80.177 -c -p
172.16.80.177:> set hello helloword
-> Redirected to slot [] located at 172.16.80.177:
OK

172.16.80.177:> get hello
-> Redirected to slot [] located at 172.16.80.177:
"helloword"

三、搭建redis集群（哨兵模式）

1、在/usr/local/下面创建redis-sentinel目录

mkdir -p /usr/local/redis-sentinel

2、将一大节中的/usr/local/redis/bin复制到redis-sentinel目录底下起名叫redis1,redis2,redis3,redis4,sentinel1,sentinel2

修改4个redis的配置文件（以redis1为例，将redis.conf重命名为redis-7001.conf）

daemonize yes
Port
#Bind 127.0.0.1
logfile "./redis-7001.log"
protected-mode no
cluster-config-file nodes-7001.conf

其他三个redis的端口号为7002,7003,7004

3、将redis解压包里面的sentinel.conf复制到sentinel1和sentinel2里面并重命名为sentinel-7005.conf,sentinel-7006.conf，并赋予可执行权限

修改配置（以sentinel-7005.conf为例）

port
sentinel monitor mymaster 172.16.80.177  2
protected-mode no

注：我们稍后要启动四个redis实例，其中端口为7001的redis设为master，其他三个设为slave 。所以my mymaster 后跟的是master的ip和端口，最后一个’2’代表我要启动只要有2个sentinel认为master下线，就认为该master客观下线，启动failover并选举产生新的master。通常最后一个参数不能多于启动的sentinel实例数。

4.启动4个redis实例

./redis-server redis-.conf
......

[root@host---- redis4]# ps -ef|grep redis
root             : ?        :: ./redis-server 172.16.80.177:
root             : ?        :: ./redis-server 172.16.80.177:
root             : ?        :: ./redis-server 172.16.80.177:
root             : ?        :: ./redis-server 172.16.80.177:

然后分别登陆7002 7003 7004三个实例，动态改变主从关系，成为7001的slave:

./redis-cli -h 172.16.80.177 -p
172.16.80.177:> SLAVEOF 172.16.80.177 

5、以后台启动模式启动两个sentinel（哨兵）:

[root@host---- sentinel1]# ./redis-sentinel sentinel-7005.conf &
[]
[root@host---- sentinel1]# :X  Oct ::49.712 * Increased maximum number of open files to  (it was originally set to ).
                _._
           _.-``__ ''-._
      _.-``    `.  `_.  ''-._           Redis 3.2. (/)  bit
  .-`` .-```.  ```\/    _.,_ ''-._
 (    '      ,       .-`  | `,    )     Running in sentinel mode
 |`-._`-...-` __...-.``-._|'` _.-'|     Port:
 |    `-._   `._    /     _.-'    |     PID: 30038
  `-._    `-._  `-./  _.-'    _.-'
 |`-._`-._    `-.__.-'    _.-'_.-'|
 |    `-._`-._        _.-'_.-'    |           http://redis.io
  `-._    `-._`-.__.-'_.-'    _.-'
 |`-._`-._    `-.__.-'    _.-'_.-'|
 |    `-._`-._        _.-'_.-'    |
  `-._    `-._`-.__.-'_.-'    _.-'
      `-._    `-.__.-'    _.-'
          `-._        _.-'
              `-.__.-'                                               

:X  Oct ::49.713 # WARNING: The TCP backlog setting of  cannot be enforced because /proc/sys/net/core/somaxconn is set to the lower value of .
:X  Oct ::49.744 # Sentinel ID is cbd44de127d84fd7d21f9e4a73e62b94c3b0ea00
:X  Oct ::49.744 # +monitor master mymaster 172.16.80.177  quorum
:X  Oct ::49.745 * +slave slave 172.16.80.177: 172.16.80.177  @ mymaster 172.16.80.177
:X  Oct ::49.768 * +slave slave 172.16.80.177: 172.16.80.177  @ mymaster 172.16.80.177
:X  Oct ::49.792 * +slave slave 172.16.80.177: 172.16.80.177  @ mymaster 172.16.80.177 

[root@host---- redis-3.2.]# ps -ef|grep sentinel
root         : pts/    :: ./redis-sentinel *: [sentinel]
root         : pts/    :: ./redis-sentinel *: [sentinel] 

6、验证

sentinel的基本命令

① INFO 
sentinel的基本状态信息 
②SENTINEL masters 
列出所有被监视的主服务器，以及这些主服务器的当前状态 
③ SENTINEL slaves 
列出给定主服务器的所有从服务器，以及这些从服务器的当前状态 
④SENTINEL get-master-addr-by-name 
返回给定名字的主服务器的 IP 地址和端口号 
⑤SENTINEL reset 
重置所有名字和给定模式 pattern 相匹配的主服务器。重置操作清除主服务器目前的所有状态， 包括正在执行中的故障转移， 并移除目前已经发现和关联的， 主服务器的所有从服务器和 Sentinel 。 
⑥SENTINEL failover 
当主服务器失效时， 在不询问其他 Sentinel 意见的情况下， 强制开始一次自动故障迁移，但是它会给其他sentinel发送一个最新的配置，其他sentinel会根据这个配置进行更新

[root@host---- redis1]# ./redis-cli -h 172.16.80.177 -p
172.16.80.177:>
172.16.80.177:> set name hzb
OK
172.16.80.177:> get name
"hzb"
172.16.80.177:> exit
[root@host---- redis1]# cd ../redis2
[root@host---- redis2]# ./redis-cli -h 172.16.80.177 -p
172.16.80.177:> get name
"hzb"
172.16.80.177:> set age
(error) READONLY You can't write against a read only slave.

上面可以看到slave是只读的

登录一个sentinel，用sentinel masters命令查看哪个是master

[root@host---- sentinel1]# ./redis-cli -h 172.16.80.177 -p
172.16.80.177:>
172.16.80.177:>
172.16.80.177:7005> sentinel masters
)  ) "name"
    ) "mymaster"
    ) "ip"
    4) "172.16.80.177"
    5) "port"
    6) "7001"
    ) "runid"
    ) "8b5d064e4195a1ee6279f5fb06f3fa88e4605854"
    ) "flags"
   ) "master"
   ) "link-pending-commands"
   ) ""
   ) "link-refcount"
   ) ""
   ) "last-ping-sent"
   ) ""
   ) "last-ok-ping-reply"
   ) ""
   ) "last-ping-reply"
   ) ""
   ) "down-after-milliseconds"
   ) ""
   ) "info-refresh"
   ) ""
   ) "role-reported"
   ) "master"
   ) "role-reported-time"
   ) ""
   ) "config-epoch"
   ) ""
   ) "num-slaves"
   ) ""
   ) "num-other-sentinels"
   ) ""
   ) "quorum"
   ) ""
   ) "failover-timeout"
   ) ""
   ) "parallel-syncs"
   ) ""

目前7001是master, 我们强制kill掉 7001 的进程以后，可以看到sentinel打出的信息：

:X  Oct ::24.138 # +sdown master mymaster 172.16.80.177
:X  Oct ::24.557 # +new-epoch
:X  Oct ::24.599 # +vote-for-leader 3948b4c445549c51d67bf829553372bca8bc67c8
:X  Oct ::25.174 # +config-update-from sentinel 3948b4c445549c51d67bf829553372bca8bc67c8 172.16.80.177  @ mymaster 172.16.80.177
:X  Oct ::25.174 # +switch-master mymaster 172.16.80.177  172.16.80.177
:X  Oct ::25.175 * +slave slave 172.16.80.177: 172.16.80.177  @ mymaster 172.16.80.177
:X  Oct ::25.175 * +slave slave 172.16.80.177: 172.16.80.177  @ mymaster 172.16.80.177
:X  Oct ::25.175 * +slave slave 172.16.80.177: 172.16.80.177  @ mymaster 172.16.80.177
30325:X 22 Oct 22:51:55.202 # +sdown slave 172.16.80.177:7001 172.16.80.177 7001 @ mymaster 172.16.80.177 7003

[root@host---- sentinel1]# ./redis-cli -h 172.16.80.177 -p
172.16.80.177:> sentinel masters
)  ) "name"
    ) "mymaster"
    ) "ip"
    ) "172.16.80.177"
    ) "port"
    ) ""
    ) "runid"
    ) "31721c8d1371697d78bf3c6acc2f8d4ba829825a"
    ) "flags"
   ) "master"
   ) "link-pending-commands"
   ) ""
   ) "link-refcount"
   ) ""
   ) "last-ping-sent"
   ) ""
   ) "last-ok-ping-reply"
   ) ""
   ) "last-ping-reply"
   ) ""
   ) "down-after-milliseconds"
   ) ""
   ) "info-refresh"
   ) ""
   ) "role-reported"
   ) "master"
   ) "role-reported-time"
   ) ""
   ) "config-epoch"
   ) ""
   ) "num-slaves"
   ) ""
   ) "num-other-sentinels"
   ) ""
   ) "quorum"
   ) ""
   ) "failover-timeout"
   ) ""
   ) "parallel-syncs"
   ) ""

上面显示了，master自动从7001变成了7003节点

再次将7001启动起来

:X  Oct ::15.544 * +convert-to-slave slave 172.16.80.177: 172.16.80.177  @ mymaster 172.16.80.177 

可以看到自动将7001变成了7003的slave

下面加一个批量启动和停止的脚本：

#!/bin/bash

start(){
  ./redis1/redis-server ./redis1/redis.conf  &
  ./redis2/redis-server ./redis2/redis.conf  &
  ./redis3/redis-server ./redis3/redis.conf  &
  ./redis4/redis-server ./redis4/redis.conf  &
  ./sentinel1/redis-sentinel ./sentinel1/sentinel.conf  &
  ./sentinel2/redis-sentinel ./sentinel2/sentinel.conf  &

}

stop(){
 ps -ef | grep "redis" | grep -v "grep" |awk '{print $2}'| while read pid
 do
    C_PID=$(ps --no-heading $pid | wc -l)
    echo "当前PID=$pid"
    if [[ $C_PID == "" ]]; then
        kill - $pid
        echo "PID=$pid 已经结束"
    else
        echo "PID=$pid 不存在"
    fi
 done

}

case "$1" in
start)
start
;;
stop)
stop
;;
restart)
stop
start
;;
*)
printf 'Usage: %s {start|stop|restart}\n'"$prog"
exit
;;
esac