·声明
1，已官网中文教程为基础，边看边学，结合环境现状搭建。
2，哥对Ruby不热爱、不熟悉、不感冒，所述内容如有疑义请谅解。
3，3.0官说集群还在测试中，其实用用也还算马马虎虎，对外集群API真心少，望有识之士能够出力。
·准备材料
VM9，CentOS 6.4_x86_64（2.6+），SecureCRT，CentOS-6.4-x86_64-bin-DVD1.iso
redis官网：redis-3.0.0-beta1.tar.gz
ruby官网：rubygems-2.0.7.zip
rubygem官网：redis-3.0.7.gem
·下锅
1，把一些基础包装一下，后面会用到
service iptables stop
放入DVD1-
yum install –y gcc*
yum install –y ruby
yum install –y ruby-rdoc
2，装redis
tar -xvf redis-3.0.0-beta1.tar.gz
cd redis-3.0.0-beta
make
make install
ll /usr/local/bin
3，配置集群（假定 $REDIS_HOME=/root/soft/redis，就像ORACLE_HOME一样）
cd $REDIS_HOME
mkdir cluster-test
cd cluster-test
mkdir 7000 7001 7002 7003 7004 7005
vim redis.conf，修改内容如下：
port 7000 个性化
cluster-enabled yes
cluster-config-file nodes..conf 个性化
cluster-node-timeout
appendonly yes
appendfilename "appendonly.7000.aof" 个性化
deamonize yes
pidfile 个性化
log 个性化
cp ../redis.conf ./7000
cp ../redis.conf ./7001
cp ../redis.conf ./7002
cp ../redis.conf ./7003
cp ../redis.conf ./7004
cp ../redis.conf ./7005
cp /usr/local/bin/redis-server ./7000
cp /usr/local/bin/redis-server ./7001
cp /usr/local/bin/redis-server ./7002
cp /usr/local/bin/redis-server ./7003
cp /usr/local/bin/redis-server ./7004
cp /usr/local/bin/redis-server ./7005
/root/soft/redis/cluster-test/redis-server /root/soft/redis/cluster-test/7000/redis.conf
/root/soft/redis/cluster-test/redis-server /root/soft/redis/cluster-test/7001/redis.conf
/root/soft/redis/cluster-test/redis-server /root/soft/redis/cluster-test/7002/redis.conf
/root/soft/redis/cluster-test/redis-server /root/soft/redis/cluster-test/7003/redis.conf
/root/soft/redis/cluster-test/redis-server /root/soft/redis/cluster-test/7004/redis.conf
/root/soft/redis/cluster-test/redis-server /root/soft/redis/cluster-test/7005/redis.conf
如果需要看一看：ps –ef|grep redis
如果需要杀一杀：ps -ef|grep redis|egrep -v grep|awk -F ' ' '{print $2}'|xargs kill -9
4，配置ruby集群脚本
cd /root/soft/rubygems
ruby setup.rb
到redis-3.0.7.gem路径下
gem install -l redis（不用L的话，走ruby官网，95%被WALL，你懂得，自己改source吧）
./redis-trib.rb create --replicas 1 127.0.0.1:7000 127.0.0.1:7002 127.0.0.3:7003 127.0.0.1:7004 127.0.0.1:7005 127.0.0.1:7005（半路记得“yes"一下）
>>> Creating cluster
Connecting to node 127.0.0.1:: OK
Connecting to node 127.0.0.1:: OK
Connecting to node 127.0.0.1:: OK
Connecting to node 127.0.0.1:: OK
Connecting to node 127.0.0.1:: OK
Connecting to node 127.0.0.1:: OK
>>> Performing hash slots allocation on nodes...
Using masters:
127.0.0.1:
127.0.0.1:
127.0.0.1:
127.0.0.1: replica # is 127.0.0.1:
127.0.0.1: replica # is 127.0.0.1:
127.0.0.1: replica # is 127.0.0.1:
M: bc66d90ca24eb6b69a3b375a4e242fef00de2052 127.0.0.1:
slots:- ( slots) master
M: bc66d90ca24eb6b69a3b375a4e242fef00de2052 127.0.0.1:
slots:- ( slots) master
M: bc66d90ca24eb6b69a3b375a4e242fef00de2052 127.0.0.1:
slots:- ( slots) master
S: bc66d90ca24eb6b69a3b375a4e242fef00de2052 127.0.0.1:
replicates bc66d90ca24eb6b69a3b375a4e242fef00de2052
S: bc66d90ca24eb6b69a3b375a4e242fef00de2052 127.0.0.1:
replicates bc66d90ca24eb6b69a3b375a4e242fef00de2052
S: bc66d90ca24eb6b69a3b375a4e242fef00de2052 127.0.0.1:
replicates bc66d90ca24eb6b69a3b375a4e242fef00de2052
Can I set the above configuration? (type 'yes' to accept): yes
>>> Nodes configuration updated
>>> Sending CLUSTER MEET messages to join the cluster
Waiting for the cluster to join...
>>> Performing Cluster Check (using node 127.0.0.1:)
M: 75bab28893c3536aecdb9879df97db89b24ce21e 127.0.0.1:
slots:- ( slots) master
M: cfc344721cad8a5ddb7a8d675eb3a4d6e5b9d1c1 127.0.0.1:
slots:- ( slots) master
M: 01b3977a3f592fd1538e8c83f86bbba37d0c1058 127.0.0.3:
slots:- ( slots) master
M: df59fce238a1c538933bdc170edb27835d0b7cc7 127.0.0.1:
slots: ( slots) master
replicates 01b3977a3f592fd1538e8c83f86bbba37d0c1058
M: b6bdb2d309aa7f0b810288f24e975230419f25b1 127.0.0.1:
slots: ( slots) master
replicates 75bab28893c3536aecdb9879df97db89b24ce21e
M: b6bdb2d309aa7f0b810288f24e975230419f25b1 127.0.0.1:
slots: ( slots) master
replicates cfc344721cad8a5ddb7a8d675eb3a4d6e5b9d1c1
[OK] All nodes agree about slots configuration.
>>> Check for open slots...
>>> Check slots coverage...
[OK] All slots covered.
redis-cli -c -p 7000
127.0.0.1:> set b c
-> Redirected to slot [] located at 127.0.0.1:
OK
127.0.0.1:> set hello world
OK
127.0.0.1:> get foo
-> Redirected to slot [] located at 127.0.0.1:
(nil)
127.0.0.1:> get hello
-> Redirected to slot [] located at 127.0.0.1:
"world"
127.0.0.1:> get b
"c"
127.0.0.1:>
·FAQ
*** ERROR: Invalid configuration for cluster creation. *** Redis Cluster requires at least master nodes. *** This is not possible with nodes and replicas per node. *** At least nodes are required.
·Redis官说，至少3+3，还延续着Master-Slave的设计理念，这点上个人觉得Cassandra的一致性确实了得。
Either the node already knows other nodes
·因为报错是ruby报的，不太能够理解其意思，因为我之前做了些单机压测，可能直接叠上集群会有垃圾数据，所以把/var/db/和/var/log/下能清空的都清空了。
[] Mar ::24.290 # Handshake error: we already know node bc66d90ca24eb6b69a3b375a4e242fef00de2052, updating the address if needed.
·同样道理，ruby报的错，一开始一直在join被我ctrl+c之后开始报错，用上面的方法无果。把所有app目录清空然后redis重装。大家测试时，最后事先vmware快照一下或者redis现网数据备份一下。
·OK，自己爽一下吧