准备环境：
使用2台centos服务器，每台机器上部署3个实例，集群为三个主节点与三个从节点：
第一台
192.168.1.107:7000
192.168.1.107:7001
192.168.1.107:7002
第二台
192.168.1.108:7003
192.168.1.108:7004
192.168.1.108:7005
安装配置：
192.168.1.107服务器
#yum -y install gcc zlib ruby rubygems
#gem install redis
#wgethttp://download.redis.io/releases/redis-3.2.8.tar.gz
#tar -zxvf redis-3.2.8.tar.gz
#cd redis-3.2.8
#make && make install
将redis-trib.rb 复制到 /usr/local/bin 目录下
#cp src/redis-trib.rb /usr/local/bin/
#mkdir -p /data/redis/{conf,data,logs}
#cp redis.conf/data/redis/conf/redis-7000.conf
#cp redis.conf/data/redis/conf/redis-7001.conf
#cp redis.conf/data/redis/conf/redis-7002.conf
配置内容如下(redis-7000.conf其它类同)：
daemonize yes
pidfile/data/redis/data/redis-7000.pid
port 7000
bind 192.168.1.107
unixsocket/data/redis/data/redis-7000.sock
unixsocketperm 700
timeout 300
loglevel verbose
logfile/data/redis/logs/redis-7000.log
databases 16
dbfilename dump-7000.rdb
dir /data/redis/data/
#aof持久化
appendonly yes
appendfilename appendonly-7000.aof
appendfsync everysec
no-appendfsync-on-rewrite yes
auto-aof-rewrite-percentage 80-100
auto-aof-rewrite-min-size 64mb
lua-time-limit 5000
#集群配置
cluster-enabled yes
cluster-config-file /data/redis/data/nodes-7000.conf
cluster-node-timeout 5000
同样的在192.168.1.108上配置其它端口，配置后我们分别启动
192.168.1.107主机
#redis-server/data/redis/conf/redis-7000.conf
#redis-server/data/redis/conf/redis-7001.conf
#redis-server/data/redis/conf/redis-7002.conf
192.168.1.108主机
#redis-server/data/redis/conf/redis-7003.conf
#redis-server/data/redis/conf/redis-7004.conf
#redis-server/data/redis/conf/redis-7005.conf
分别在两台主机查看启动进程与端口：
# ps -ef|grep redis
root 4663 1 0 10:53 ? 00:00:09 redis-server192.168.1.107:7000 [cluster]
root 4667 1 0 10:53 ? 00:00:09 redis-server192.168.1.107:7001 [cluster]
root 5566 1 0 12:12 ? 00:00:06 redis-server192.168.1.107:7002 [cluster]
root 5611 5313 0 14:49 pts/1 00:00:00 grep redis
# netstat -tnlp | grep redis
tcp 0 0 192.168.1.107:7000 0.0.0.0:* LISTEN 4663/redis-server 1
tcp 0 0 192.168.1.107:7001 0.0.0.0:* LISTEN 4667/redis-server 1
tcp 0 0 192.168.1.107:7002 0.0.0.0:* LISTEN 5566/redis-server 1
tcp 0 0 192.168.1.107:17000 0.0.0.0:* LISTEN 4663/redis-server 1
tcp 0 0 192.168.1.107:17001 0.0.0.0:* LISTEN 4667/redis-server 1
tcp 0 0 192.168.1.107:17002 0.0.0.0:* LISTEN 5566/redis-server 1
创建集群
redis-trib.rb命令介绍可以参考：
http://weizijun.cn/2016/01/08/redis%20cluster%E7%AE%A1%E7%90%86%E5%B7%A5%E5%85%B7redis-trib-rb%E8%AF%A6%E8%A7%A3/
# redis-trib.rb create --replicas 1 192.168.1.107:7000 192.168.1.107:7001 192.168.1.107:7002 192.168.1.108:7003 192.168.1.108:7004 192.168.1.108:7005
>>> Creating cluster
>>> Performing hash slotsallocation on 6 nodes...
Using 3 masters:
192.168.1.108:7003
192.168.1.107:7000
192.168.1.108:7004
Adding replica 192.168.1.107:7001 to192.168.12.108:7003
Adding replica 192.168.1.108:7005 to192.168.12.107:7000
Adding replica 192.168.1.107:7002 to192.168.12.108:7004
M: d0ce4d4eb8c503267ffea606e90d1c537a83a22e192.168.1.107:7000
slots:5461-10922 (5462 slots) master
S: 24ca2753885db7f37c0e0077b1f43eb2b1a52e42192.168.1.107:7001
replicates 8ee448821d5e757af73bab65861340e557b8cd14
S: 90c36924c71ae924b1179ff1a384c53e5ea2e484192.168.1.107:7002
replicates 5a93221a3281f88f9984c24b6e1d6f08b3685c89
M: 8ee448821d5e757af73bab65861340e557b8cd14192.168.1.108:7003
slots:0-5460 (5461 slots) master
M: 5a93221a3281f88f9984c24b6e1d6f08b3685c89192.168.1.108:7004
slots:10923-16383 (5461 slots) master
S: a63176f88cbf58ab879421148da3b171f169540d192.168.1.108:7005
replicates d0ce4d4eb8c503267ffea606e90d1c537a83a22e
Can I set the above configuration? (type‘yes‘ to accept): yes #输入yes
>>> Nodes configuration updated
>>> Assign a different configepoch to each node
>>> Sending CLUSTER MEET messagesto join the cluster
Waiting for the cluster to join.....
>>> Performing Cluster Check(using node 192.168.1.107:7000)
M: d0ce4d4eb8c503267ffea606e90d1c537a83a22e192.168.1.107:7000
slots:5461-10922 (5462 slots) master
M: 24ca2753885db7f37c0e0077b1f43eb2b1a52e42192.168.1.107:7001
slots: (0 slots) master
replicates 8ee448821d5e757af73bab65861340e557b8cd14
M: 90c36924c71ae924b1179ff1a384c53e5ea2e484192.168.1.107:7002
slots: (0 slots) master
replicates 5a93221a3281f88f9984c24b6e1d6f08b3685c89
M: 8ee448821d5e757af73bab65861340e557b8cd14192.168.1.108:7003
slots:0-5460 (5461 slots) master
M: 5a93221a3281f88f9984c24b6e1d6f08b3685c89192.168.1.108:7004
slots:10923-16383 (5461 slots) master
M: a63176f88cbf58ab879421148da3b171f169540d192.168.1.108:7005
slots: (0 slots) master
replicates d0ce4d4eb8c503267ffea606e90d1c537a83a22e
[OK] All nodes agree about slotsconfiguration.
>>> Check for open slots...
>>> Check slots coverage...
[OK] All 16384 slots covered. #集群创建成功
集群验证
192.168.1.107的7002端口的节点写入数据，在192.168.1.108的7005端口查询，接方式为
redis-cli -h 192.168.31.245 -c -p 7002 ,加参数 -C 可连接到集群，因为上面redis.conf
将 bind 改为了ip地址，所以 -h 参数不可以省略。
# redis-cli -h 192.168.1.107 -c -p 7002
192.168.1.108:7003> set test01"this is test01"
OK
192.168.1.108:7003>
# redis-cli -h 192.168.1.108 -c -p 7005
192.168.12.108:7005> get test01
-> Redirected to slot [1840] located at192.168.12.108:7003
"this is test01"
说明集群运作正常。
Redis 集群会把数据存在一个 master 节点，然后在这个master 和其对应的salve 之间进行数据同步。当读取数据时，也根据一致性哈希算法到对应的 master 节点获取数据。只有当一个master 挂掉之后，才会启动一个对应的 salve 节点，充当 master 。
需要注意的是：必须要3个或以上的主节点，否则在创建集群时会失败，并且当存活的主节点数小于总节点数的一半时，整个集群就无法提供服务了。
原文地址：http://10158955.blog.51cto.com/10148955/1963390
