redis集群

最新推荐文章于 2024-09-27 21:26:22 发布

Starts️

最新推荐文章于 2024-09-27 21:26:22 发布

阅读量677

点赞数 31

文章标签： redis 数据库缓存

本文链接：https://blog.csdn.net/zbw0323/article/details/136038619

版权

1.4Redis-Cluster的主从复制模型

1.2分别创建/data/application目录用于存放redis

1.8复制/root/redis目录下的所有数据到创建的目录

1.9 redis.conf已经修改过只需修改端口即可

一、redis集群

1.1介绍

1、Redis集群是一个提供在多个Redis间节点间共享数据的程序集
2、Redis集群并不支持处理多个keys的命令,因为这需要在不同的节点间移动数据,从而达不到像Redis那样的性能,在高负载的情况下可能会导致不可预料的错误
3、Redis集群通过分区来提供一定程度的可用性,在实际环境中当某个节点宕机或者不可达的情况下可继续处理命令

1.2优势

自动分割数据到不同的节点上
整个集群的部分节点失败或者不可达的情况下能够继续处理命令

1.3Redis-Cluster数据分片

Redis集群没有使用一致性hash，而是引入了哈希槽概念
Redis集群有16384个哈希槽
每个key通过CRC16校验后对16384取模来决定放置槽
集群的每个节点负责一部分哈希槽

以3个节点组成的集群为例

节点A包含О到5500号哈希槽
节点B包含5501到11000号哈希槽
节点C包含11001到16384号哈希槽

支持添加或者删除节点

添加删除节点无需停止服务

1.4Redis-Cluster的主从复制模型

集群中具有A，B，C三个节点,如果节点B失败了，整个集群就会因缺少5501-11000这个范围的槽而不可用

为每个节点添加一个从节点A1，B1，C1,整个集群便有三个master节点和三个slave节点组成，在节点B失败后，集群便会选举B1为新的主节点继续服务

当B和B1都失败后，集群将不可用

1.5原理

一、搭建redis集群

准备服务器

这里我准备4台

先关防火墙、selinx（4台都执行）

 [root@cluster1 ~]# systemctl stop firewalld && setenforce 0

 [root@cluster1 ~]# 192.168.239.145
 [root@cluster2 ~]# 192.168.239.146
 [root@cluster3 ~]# 192.168.239.141
 [root@cluster4 ~]# 192.168.239.139

1、安装redis

1.1官网下载

[root@cluster1 ~]# wget https://github.com/redis/redis/archive/7.2.4.tar.gz

1.2分别创建/data/application目录用于存放redis

[root@cluster1 ~]# mkdir -p /data/application/{7001,7002}
[root@cluster2 ~]# mkdir -p /data/application/{7003,7004}
[root@cluster3 ~]# mkdir -p /data/application/{7005,7006}
[root@cluster4 ~]# mkdir -p /data/application/{8001,8002}#这台服务器用于后面测试添加主、从节点

1.3解压redis源码包并改名

[root@cluster1 ~]# tar xf 7.2.4.tar.gz 
[root@cluster1 ~]# mv redis-7.2.4 redis

1.4安装编译工具

[root@cluster1 ~]# yum -y install gcc make

1.5编译

[root@cluster1 ~]# cd /root/redis/
[root@cluster1 redis]# make

1.6修改redis的配置文件

[root@cluster1 ~]# cd /root/redis
[root@cluster1 redis]# cp redis.conf redis.conf.bak #备份文件

[root@cluster1 redis]# vim redis.conf     ---修改如下
bind 0.0.0.0　　#所有IP都可访问
daemonize yes　　　　　#开启后台模式将on改为yes
timeout 300　　　　　　#连接超时时间
port 6379                      #端口号
dir /data/application/6379/data　　#本地数据库存放持久化数据的目录该目录-----需要存在 这里改为6379方便后面cp
pidfile /var/run/redis_6379.pid　　#定义pid文件
logfile /var/log/redis.log　　#定义log文件

1.7创建存储数据的目录

[root@cluster1 ~]# mkdir /root/redis/data

1.8复制/root/redis目录下的所有数据到创建的目录

这里选择cp、不需要每台服务器都编译安装redis 有闲时间的可全部编译。

[root@cluster1 ~]# cp -rf /root/redis/* /data/application/7001
[root@cluster1 ~]# cp -rf /root/redis/* /data/application/7002
[root@cluster1 ~]# scp -r /root/redis/* 192.168.239.146:/data/application/7003
[root@cluster1 ~]# scp -r /root/redis/* 192.168.239.146:/data/application/7004
[root@cluster1 ~]# scp -r /root/redis/* 192.168.239.141:/data/application/7005
[root@cluster1 ~]# scp -r /root/redis/* 192.168.239.141:/data/application/7006
[root@cluster1 ~]# scp -r /root/redis/* 192.168.239.139:/data/application/8002
[root@cluster1 ~]# scp -r /root/redis/* 192.168.239.139:/data/application/8003

1.9 redis.conf已经修改过只需修改端口即可

[root@cluster1 ~]# sed -i s/6379/7001/g /data/application/7001/redis.conf
[root@cluster1 ~]# sed -i s/6379/7002/g /data/application/7002/redis.conf
[root@cluster2 ~]# sed -i s/6379/7003/g /data/application/7003/redis.conf
[root@cluster2 ~]# sed -i s/6379/7004/g /data/application/7004/redis.conf
[root@cluster3 ~]# sed -i s/6379/7005/g /data/application/7005/redis.conf
[root@cluster3 ~]# sed -i s/6379/7006/g /data/application/7006/redis.conf
[root@cluster4 ~]# sed -i s/6379/8001/g /data/application/8001/redis.conf
[root@cluster4 ~]# sed -i s/6379/8002/g /data/application/8002/redis.conf

1.10启动redis

这里我只演示cluster2，剩余自己启动，方法一样

注意：启动时必须在7002这层目录

端口为7002、17002

root@cluster1 7002]# src/redis-server redis.conf
[root@cluster1 7002]# ss -nplt
State       Recv-Q Send-Q            Local Address:Port                           Peer Address:Port              
LISTEN      0      100                   127.0.0.1:25                                        *:*                   users:(("master",pid=1116,fd=13))
LISTEN      0      128                           *:7002                                      *:*                   users:(("redis-server",pid=3067,fd=7))
LISTEN      0      128                           *:17002                                     *:*                   users:(("redis-server",pid=3067,fd=8))
LISTEN      0      128                           *:22                                        *:*                   users:(("sshd",pid=965,fd=3))
LISTEN      0      100                       [::1]:25                                     [::]:*                   users:(("master",pid=1116,fd=14))
LISTEN      0      128                        [::]:22                                     [::]:*                   users:(("sshd",pid=965,fd=4))
[root@cluster1 7002]#

1.11创建集群

注意IP与端口的对应

[root@cluster1 ~]# /data/application/7001/src/redis-cli --cluster create --cluster-replicas 1 192.168.239.145:7001 192.168.239.145:7002 192.168.239.146:7003 192.168.239.146:7004 192.168.239.141:7005 192.168.239.141:7006

输出结果如下

>>> Performing hash slots allocation on 6 nodes...
Master[0] -> Slots 0 - 5460
Master[1] -> Slots 5461 - 10922
Master[2] -> Slots 10923 - 16383
Adding replica 192.168.239.146:7004 to 192.168.239.145:7001
Adding replica 192.168.239.141:7006 to 192.168.239.146:7003
Adding replica 192.168.239.145:7002 to 192.168.239.141:7005
M: 409b8508a5ea87fdfe7a29c35ad9ed333798ec15 192.168.239.145:7001
   slots:[0-5460] (5461 slots) master
S: 5fb3421f52f1db4d5de9092770a83dfafd91111e 192.168.239.145:7002
   replicates d38b84d902ba175258862bf429af1a133127d495
M: f4ddbf650fa050776daf019471a0e51224b2c503 192.168.239.146:7003
   slots:[5461-10922] (5462 slots) master
S: 71b2a492596aaf8fefa060f3aa8bdb8dc241e982 192.168.239.146:7004
   replicates 409b8508a5ea87fdfe7a29c35ad9ed333798ec15
M: d38b84d902ba175258862bf429af1a133127d495 192.168.239.141:7005
   slots:[10923-16383] (5461 slots) master
S: 52f5b9e838339595a2ca66633464a1ca041deee4 192.168.239.141:7006
   replicates f4ddbf650fa050776daf019471a0e51224b2c503
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 192.168.239.145:7001)
M: 409b8508a5ea87fdfe7a29c35ad9ed333798ec15 192.168.239.145:7001
   slots:[0-5460] (5461 slots) master
   1 additional replica(s)
S: 71b2a492596aaf8fefa060f3aa8bdb8dc241e982 192.168.239.146:7004
   slots: (0 slots) slave
   replicates 409b8508a5ea87fdfe7a29c35ad9ed333798ec15
S: 5fb3421f52f1db4d5de9092770a83dfafd91111e 192.168.239.145:7002
   slots: (0 slots) slave
   replicates d38b84d902ba175258862bf429af1a133127d495
M: f4ddbf650fa050776daf019471a0e51224b2c503 192.168.239.146:7003
   slots:[5461-10922] (5462 slots) master
   1 additional replica(s)
M: d38b84d902ba175258862bf429af1a133127d495 192.168.239.141:7005
   slots:[10923-16383] (5461 slots) master
   1 additional replica(s)
S: 52f5b9e838339595a2ca66633464a1ca041deee4 192.168.239.141:7006
   slots: (0 slots) slave
   replicates f4ddbf650fa050776daf019471a0e51224b2c503
[OK] All nodes agree about slots configuration.
>>> Check for open slots...
>>> Check slots coverage...
[OK] All 16384 slots covered.

1.12测试集群

[root@cluster1 7001]# src/redis-cli -c  -h 192.168.239.145 -p 7001  #'连接master1'   
192.168.239.145:7001> keys *     #查看所有键值
(empty array)
192.168.239.145:7001> set name xiaoming          ##'创建一个键值'
-> Redirected to slot [5798] located at 192.168.239.146:7003  #'存入master2'
OK
192.168.239.146:7003> set job teacher  #'自动跳转到master2，再创建一个键值'
-> Redirected to slot [2906] located at 192.168.239.145:7001  #'存入master1'
OK
192.168.239.145:7001> 
[root@cluster1 7001]# src/redis-cli -c  -h 192.168.239.141 -p 7006 #'连接master2的slave'
192.168.239.141:7006> keys *        #'查看键值，发现已同步'
1) "name"
192.168.239.141:7006> get name
-> Redirected to slot [5798] located at 192.168.239.146:7003  #'在master2上找到'
"xiaoming"
192.168.239.146:7003> 
[root@cluster1 7001]# src/redis-cli -c  -h 192.168.239.146 -p 7004 #'连接master1的slave'
192.168.239.146:7004> keys *     #'查看键值，发现已同步'
1) "job"
192.168.239.146:7004> get job
-> Redirected to slot [2906] located at 192.168.239.145:7001  #'在master1上找到'
"teacher"
192.168.239.145:7001>

1.13模拟master服务器宕机情况

[root@cluster1 7001]# src/redis-cli -c  -h 192.168.239.145 -p 7001 shutdown  #'关闭master1'
[root@cluster1 7001]# src/redis-cli -c  -h 192.168.239.146 -p 7004  #'连接master1的slave'
192.168.239.146:7004> keys *    #'依旧可以查看到键值，redis集群正常'
1) "job"
192.168.239.146:7004> get job
"teacher"
192.168.239.146:7004> 
[root@cluster1 7001]# src/redis-cli -c  -h 192.168.239.146 -p 7003  #'连接master2'
192.168.239.146:7003> set age 18   #'创建一个键值存入到master1的slave'
-> Redirected to slot [741] located at 192.168.239.146:7004 
OK
192.168.239.146:7004> keys *  #'查看master1的slave，发现两键值都存在'
1) "age"
2) "job"
192.168.239.146:7004> get age   #'刚创建的键值也可以查看'
"18"
192.168.239.146:7004>

总结：在redis集群中，在3个msater上依次写入数据，会在master对应的slave上复制一份数据，在一台master宕机后，由它的slave顶替进行数据的读写工作。

1.14添加主、从节点

启动事先准备好的cluster4

[root@cluster4 8001]# src/redis-server redis.conf 
[root@cluster4 8001]# ../8002/src/redis-server ../8002/redis.conf 
root@cluster4 8001]# ss -nplt
State      Recv-Q Send-Q             Local Address:Port                            Peer Address:Port              
LISTEN     0      100                    127.0.0.1:25                                         *:*                   users:(("master",pid=1110,fd=13))
LISTEN     0      128                            *:8001                                       *:*                   users:(("redis-server",pid=1447,fd=7))
LISTEN     0      128                            *:8002                                       *:*                   users:(("redis-server",pid=1453,fd=7))
LISTEN     0      128                            *:18001                                      *:*                   users:(("redis-server",pid=1447,fd=8))
LISTEN     0      128                            *:18002                                      *:*                   users:(("redis-server",pid=1453,fd=8))
LISTEN     0      128                            *:22                                         *:*                   users:(("sshd",pid=940,fd=3))
LISTEN     0      100                        [::1]:25                                      [::]:*                   users:(("master",pid=1110,fd=14))
LISTEN     0      128                         [::]:22                                      [::]:*                   users:(("sshd",pid=940,fd=4))
[root@cluster4 8001]#

添加主节点

[root@cluster1 7002]# src/redis-cli -c -h 192.168.239.145 -p 7001 cluster meet 192.168.239.139 8001
OK

查看是否添加

[root@cluster1 7002]# src/redis-cli -c -h 192.168.239.145 -p 7001
192.168.239.145:7001> cluster nodes
409b8508a5ea87fdfe7a29c35ad9ed333798ec15 192.168.239.145:7001@17001 myself,slave 71b2a492596aaf8fefa060f3aa8bdb8dc241e982 0 1707274235000 7 connected
f4ddbf650fa050776daf019471a0e51224b2c503 192.168.239.146:7003@17003 master - 0 1707274236012 12 connected 5461-10922
52f5b9e838339595a2ca66633464a1ca041deee4 192.168.239.141:7006@17006 slave f4ddbf650fa050776daf019471a0e51224b2c503 0 1707274237527 12 connected
5fb3421f52f1db4d5de9092770a83dfafd91111e 192.168.239.145:7002@17002 master - 0 1707274235910 10 connected 10923-16383
734442f680633d87127cf8c59741317da49a6949 192.168.239.139:8001@18001 master - 0 1707274237000 0 connected
71b2a492596aaf8fefa060f3aa8bdb8dc241e982 192.168.239.146:7004@17004 master - 0 1707274236416 7 connected 0-5460
d38b84d902ba175258862bf429af1a133127d495 192.168.239.141:7005@17005 slave 5fb3421f52f1db4d5de9092770a83dfafd91111e 0 1707274237426 10 connected
192.168.239.145:7001>

添加从节点

[root@cluster1 7002]# src/redis-cli -c -h 192.168.239.145 -p 7001 cluster meet 192.168.239.139 8002
OK
#replicate后面这里指定的id为主的ip
[root@cluster1 7002]# src/redis-cli  -h 192.168.239.139 -p 8002 cluster replicate 734442f680633d87127cf8c59741317da49a6949
OK

查看是否添加

[root@cluster2 7003]# src/redis-cli -c -h 192.168.239.145 -p 7001
192.168.239.145:7001> cluster nodes
409b8508a5ea87fdfe7a29c35ad9ed333798ec15 192.168.239.145:7001@17001 myself,master - 0 1707275189000 18 connected 0-5460
f4ddbf650fa050776daf019471a0e51224b2c503 192.168.239.146:7003@17003 slave 52f5b9e838339595a2ca66633464a1ca041deee4 0 1707275189335 22 connected
52f5b9e838339595a2ca66633464a1ca041deee4 192.168.239.141:7006@17006 master - 0 1707275188339 22 connected 5461-10922
5fb3421f52f1db4d5de9092770a83dfafd91111e 192.168.239.145:7002@17002 slave d38b84d902ba175258862bf429af1a133127d495 0 1707275189536 23 connected
734442f680633d87127cf8c59741317da49a6949 192.168.239.139:8001@18001 master - 0 1707275189000 13 connected
71b2a492596aaf8fefa060f3aa8bdb8dc241e982 192.168.239.146:7004@17004 slave 409b8508a5ea87fdfe7a29c35ad9ed333798ec15 0 1707275187314 18 connected
d38b84d902ba175258862bf429af1a133127d495 192.168.239.141:7005@17005 master - 0 1707275189133 23 connected 10923-16383
ff9a1e6774361382b1b209317942ddc754cf9b62 192.168.239.139:8002@18002 slave 734442f680633d87127cf8c59741317da49a6949 0 1707275189000 13 connected
192.168.239.145:7001>

1.15自动分配槽位

[root@cluster1 7002]# src/redis-cli --cluster rebalance --cluster-threshold 1 --cluster-use-empty-masters 192.168.239.145:7001

[root@cluster2 7003]# src/redis-cli -c -h 192.168.239.145 -p 7001
192.168.239.145:7001> cluster nodes
409b8508a5ea87fdfe7a29c35ad9ed333798ec15 192.168.239.145:7001@17001 myself,slave 71b2a492596aaf8fefa060f3aa8bdb8dc241e982 0 1707275487000 25 connected
f4ddbf650fa050776daf019471a0e51224b2c503 192.168.239.146:7003@17003 slave 52f5b9e838339595a2ca66633464a1ca041deee4 0 1707275488559 22 connected
52f5b9e838339595a2ca66633464a1ca041deee4 192.168.239.141:7006@17006 master - 0 1707275488000 22 connected 6827-10922
5fb3421f52f1db4d5de9092770a83dfafd91111e 192.168.239.145:7002@17002 slave d38b84d902ba175258862bf429af1a133127d495 0 1707275487000 23 connected
734442f680633d87127cf8c59741317da49a6949 192.168.239.139:8001@18001 master - 0 1707275488960 24 connected 0-1364 5461-6826 10923-12287
71b2a492596aaf8fefa060f3aa8bdb8dc241e982 192.168.239.146:7004@17004 master - 0 1707275487000 25 connected 1365-5460
d38b84d902ba175258862bf429af1a133127d495 192.168.239.141:7005@17005 master - 0 1707275488000 23 connected 12288-16383
ff9a1e6774361382b1b209317942ddc754cf9b62 192.168.239.139:8002@18002 slave 734442f680633d87127cf8c59741317da49a6949 0 1707275487546 24 connected
192.168.239.145:7001>

1.16回收节点

回收节点需先回收槽位

回收槽位

[root@cluster1 7002]# src/redis-cli -h 192.168.239.145 -p 7001 cluster reset

语法:[root@cluster1 7002]# src/redis-cli -h 回收节点IP -p 端口 cluster reset

下线主机

语法：src/redis-cli --cluster del-node 登录集群ip:端口 被下线主机id

[root@cluster1 7002]# src/redis-cli --cluster del-node 192.168.239.145:7002 409b8508a5ea87fdfe7a29c35ad9ed333798ec15
>>> Removing node 409b8508a5ea87fdfe7a29c35ad9ed333798ec15 from cluster 192.168.239.145:7002
>>> Sending CLUSTER FORGET messages to the cluster...
>>> Sending CLUSTER RESET SOFT to the deleted node.