Redis集群是一个由多个主从节点群组成的分布式服务集群，它具有复制、高可用和分片特性。Redis集群不需要sentinel哨兵也能完成节点移除和故障转移的功能。需要将每个节点设置成集群模式，这种集群模式没有中心节点，可水平扩展，据官方文档称可以线性扩展到上万个节点(官方推荐不超过1000个节点)。redis集群的性能和高可用性均优于之前版本的哨兵模式，且集群配置非常简单。

系统：centos7.4

服务器金山云

[root@jsy-bj-test00~]#yuminstall-yrubyrubygems复制6份redis服务

[work@jsy-bj-test00~]$cp-rpredisredis1[work@jsy-bj-test00~]$cp-rpredisredis2[work@jsy-bj-test00~]$cp-rpredisredis3redis配置文件修改

#六个节点需做如下更改[work@jsy-bj-test00~]$vimredis1/etc/redis.conf[work@jsy-bj-test00~]$sed-i's/port6379/port6380/g'redis5/etc/redis.conf#修改端口port6380#打开注释，开启集群模式cluster-enabledyes#集群的配置文件cluster-config-filenodes-6380.conf[work@jsy-bj-test00~]$sed-i's/cluster-config-filenodes-6379.conf/cluster-config-filenodes-6380.conf/g'redis5/etc/redis.conf#pidfile文件pidfile/var/run/redis_6380.pid[work@jsy-bj-test00~]$sed-i's/pidfile\/var\/run\/redis_6380.pid/pidfile\/var\/run\/redis_6380.pid/g'redis5/etc/redis.conf#日志文件logfile"/home/work/logs/redis/6380.log"[work@jsy-bj-test00~]$sed-i's/logfile"\/home\/work\/logs\/redis\/6379.log"/logfile"\/home\/work\/logs\/redis\/6380.log"/g'redis5/etc/redis.conf#rdb持久化文件dbfilenamedump6380.rdb[work@jsy-bj-test00~]$sed-i's/dbfilenamedump6379.rdb/dbfilenamedump6380.rdb/g'redis5/etc/redis.conf#请求超时，单位毫秒cluster-node-timeout5000#开启aof持久化方式appendonlyyes#配置持久化文件appendfilename"appendonly6379.aof"[work@jsy-bj-test00~]$sed-i's/appendfilename"appendonly6379.aof"/appendfilename"appendonly6383.aof"/g'redis5/etc/redis.conf编写集群启动脚本和停止脚本

启动脚本start_all.sh

/home/work/redis/bin/redis-server/home/work/redis/etc/redis.conf&/home/work/redis/bin/redis-server/home/work/redis/etc/redis6380.conf&/home/work/redis/bin/redis-server/home/work/redis/etc/redis6381.conf&/home/work/redis/bin/redis-server/home/work/redis/etc/redis6382.conf&/home/work/redis/bin/redis-server/home/work/redis/etc/redis6383.conf&/home/work/redis/bin/redis-server/home/work/redis/etc/redis6384.conf&

关闭脚本stop_all.sh

/home/work/redis/bin/redis-cli-p6379-atestshutdown/home/work/redis/bin/redis-cli-p6380-atestshutdown/home/work/redis/bin/redis-cli-p6381-atestshutdown/home/work/redis/bin/redis-cli-p6382-atestshutdown/home/work/redis/bin/redis-cli-p6383-atestshutdown/home/work/redis/bin/redis-cli-p6384-atestshutdown执行创建集群命令

有如下报错

[work@jsy-bj-test00src]$./redis-trib.rbcreate--replicas1127.0.0.1:6379127.0.0.1:6380127.0.0.1:6381127.0.0.1:6382127.0.0.1:6383127.0.0.1:6384/usr/share/rubygems/rubygems/core_ext/kernel_require.rb:55:in`require':cannotloadsuchfile--redis(LoadError)from/usr/share/rubygems/rubygems/core_ext/kernel_require.rb:55:in`require'from./redis-trib.rb:25:in`

‘

原因：安装redis接口报错ruby版本过低

[root@jsy-bj-test00~]#geminstallredisFetching:redis-4.1.3.gem(100%)ERROR:Errorinstallingredis:redisrequiresRubyversion>=2.3.0.

解决办法到官网下载最新稳定版源代码 http://www.ruby-lang.org/en/downloads/ 进行编译安装

[work@jsy-bj-test00soft]$tarzxvfruby-2.7.0.tar.gz[work@jsy-bj-test00soft]$cdruby-2.7.0[work@jsy-bj-test00ruby-2.7.0]$./configure--prefix=/home/work/ruby&&make&&makeinstall

安装redis接口

[work@jsy-bj-test00bin]$geminstallredis

再次执行创建集群命令

[work@jsy-bj-test00src]$./redis-trib.rbcreate--replicas1127.0.0.1:6379127.0.0.1:6380127.0.0.1:6381127.0.0.1:6382127.0.0.1:6383127.0.0.1:6384>>>Creatingcluster>>>Performinghashslotsallocationon6nodes.........>>>Nodesconfigurationupdated>>>Assignadifferentconfigepochtoeachnode>>>SendingCLUSTERMEETmessagestojointheclusterWaitingfortheclustertojoin...>>>PerformingClusterCheck(usingnode127.0.0.1:6379)......[OK]Allnodesagreeaboutslotsconfiguration.>>>Checkforopenslots...>>>Checkslotscoverage...[OK]All16384slotscovered.

创建成功,查看集群状态

[work@jsy-bj-test00~]$./redis/bin/redis-cli-h127.0.0.1-p6379-c127.0.0.1:6379>clusterinfocluster_state:okcluster_slots_assigned:16384cluster_slots_ok:16384cluster_slots_pfail:0cluster_slots_fail:0cluster_known_nodes:6cluster_size:3cluster_current_epoch:6cluster_my_epoch:1cluster_stats_messages_sent:627cluster_stats_messages_received:627

查看集群节点信息

127.0.0.1:6379>clusternodes565246bf31d8e05e464db7455521b1a9f165a9cd127.0.0.1:6380master-015784477762302connected5461-109221b99b2a1e4b530501476ab48422c75f30423fd19127.0.0.1:6383slave565246bf31d8e05e464db7455521b1a9f165a9cd015784477782335connected735ad5778458059316794b9378d4b81aaff20322127.0.0.1:6379myself,master-001connected0-5460757c2c11ecebfc607aa10a6877e348d0e2da484f127.0.0.1:6381master-015784477777323connected10923-16383f22efc2bcfcd11cee6487ebc9c75de3b59f5e1d0127.0.0.1:6382slave735ad5778458059316794b9378d4b81aaff20322015784477762304connectedb8bba94b9647caa8600363144fd7108082e45f56127.0.0.1:6384slave757c2c11ecebfc607aa10a6877e348d0e2da484f015784477772326connected#这是很重要的命令，我们需要关心的信息有：#第一个参数：节点ID#第二个参数：IP:PORT@TCP这里一个坑，jedis-2.9.0之前的版本解析@出错#第三个参数：标志（Master，Slave，Myself，Fail...）#第四个参数：如果是从机则是主机的节点ID#最后两个参数：连接的状态和槽的位置。集群管理相关命令

#集群增加节点，先复制两个配置文件，并修改配置文件内容[work@jsy-bj-test00etc]$cp-predis.confredis6385.conf[work@jsy-bj-test00etc]$cp-predis.confredis6386.conf[work@jsy-bj-test00etc]$sed-i's/6379/6385/g'redis6385.conf[work@jsy-bj-test00etc]$sed-i's/6379/6386/g'redis6386.conf

启动6385节点

[work@jsy-bj-test00bin]$./redis-server/home/work/redis/etc/redis6385.conf&

将集群管理工具软连到redis的bin下

[work@jsy-bj-test00bin]$ln-s/home/work/soft/redis-3.2.11/src/redis-trib.rb/home/work/redis/bin/redis-trib.rb

将新节点加入集群master

[work@jsy-bj-test00bin]$./redis-trib.rbadd-node127.0.0.1:6385127.0.0.1:6379>>>Addingnode127.0.0.1:6385tocluster127.0.0.1:6379>>>PerformingClusterCheck(usingnode127.0.0.1:6379)......[OK]Allnodesagreeaboutslotsconfiguration.>>>Checkforopenslots...>>>Checkslotscoverage...[OK]All16384slotscovered.>>>SendCLUSTERMEETtonode127.0.0.1:6385tomakeitjointhecluster.[OK]Newnodeaddedcorrectly.

查看集群状态，未分配槽位

[work@jsy-bj-test00bin]$./redis-cli-h127.0.0.1-p6379-cclusternodesd00d05f601df0b69df0c2cc532b636d2c83347be127.0.0.1:6385master-015784502801910connected

给6385节点分配槽位

[work@jsy-bj-test00bin]$./redis-trib.rbreshard127.0.0.1:6379>>>PerformingClusterCheck(usingnode127.0.0.1:6379)......[OK]Allnodesagreeaboutslotsconfiguration.>>>Checkforopenslots...>>>Checkslotscoverage...[OK]All16384slotscovered.Howmanyslotsdoyouwanttomove(from1to16384)?500WhatisthereceivingnodeID?d00d05f601df0b69df0c2cc532b636d2c83347bePleaseenterallthesourcenodeIDs.Type'all'touseallthenodesassourcenodesforthehashslots.Type'done'onceyouenteredallthesourcenodesIDs.Sourcenode#1:allDoyouwanttoproceedwiththeproposedreshardplan(yes/no)?yes#第一个参数：需要移动槽的个数，#第二个参数：接受槽的节点ID，#第三个参数：输入"all"表示从所有原节点中获取槽，#第四个参数：输入"yes"开始移动槽到目标结点id#查看6385节点信息，已经分配槽位[work@jsy-bj-test00bin]$./redis-cli-h127.0.0.1-p6379-cclusternodesd00d05f601df0b69df0c2cc532b636d2c83347be127.0.0.1:6385master-015784524221677connected0-1655461-562710923-11088

添加从节点，启动6386节点并加入集群

[work@jsy-bj-test00bin]$./redis-server/home/work/redis/etc/redis6386.conf&[work@jsy-bj-test00bin]$./redis-trib.rbadd-node--slave--master-idd00d05f601df0b69df0c2cc532b636d2c83347be127.0.0.1:6386127.0.0.1:6385>>>Addingnode127.0.0.1:6386tocluster127.0.0.1:6385>>>PerformingClusterCheck(usingnode127.0.0.1:6385)......[OK]Allnodesagreeaboutslotsconfiguration.>>>Checkforopenslots...>>>Checkslotscoverage...[OK]All16384slotscovered.>>>SendCLUSTERMEETtonode127.0.0.1:6386tomakeitjointhecluster.Waitingfortheclustertojoin.>>>Configurenodeasreplicaof127.0.0.1:6385.[OK]Newnodeaddedcorrectly.

查看6386节点状态

[work@jsy-bj-test00bin]$./redis-cli-h127.0.0.1-p6379-cclusternodes6845878cbef3fe25f19a70a8db3eb29abb1b9ea6127.0.0.1:6386slaved00d05f601df0b69df0c2cc532b636d2c83347be015784528054397connected

删除节点

[work@jsy-bj-test00bin]$./redis-trib.rbdel-node127.0.0.1:63831b99b2a1e4b530501476ab48422c75f30423fd19>>>Removingnode1b99b2a1e4b530501476ab48422c75f30423fd19fromcluster127.0.0.1:6383>>>SendingCLUSTERFORGETmessagestothecluster...>>>SHUTDOWNthenode.