大家都知道，现在互联网的发展迅速，单台服务器已经处理不了高并发等请求，内存存储有限，并发处理请求能力瓶颈，这是我们要，水平扩展，多台机器来处理高并发请求，实现读写分离，主服务器只来写数据，同步从服务数据，从服务器呢只是用来读数据，那么高可用系统有什么优点呢，主要就是减少停工时间，保持服务高度可用性，24小时不宕机，为客户长期稳定的提供服务。那怎样搭建一个高可用的redis集群呢，redis提供了哨兵模式，监视master服务器的健康状态，如果master服务器宕机了，会自动剔除，并且在从服务器中，通过选举机制，选出一台从服务器，升级为master，哨兵模式呢，有什么缺点呢，不停的对redis服务器做监视，加大网络开销，消耗机器的性能，这里呢，主要为大家推荐redis-cluster来搭建redis集群。
什么是redis-cluster
redis3.0后支持redis-cluster，至少需要3(master)+3(slave)，建立集群，采用无中心结构，这个概念和现在热门的区块链技术很相像，每个节点都可相互访问，，这里涉及到一个投票的概念，那怎样判断一台服务器挂掉了呢，半数以上的master节点访问这台master服务器时，访问不了，说明这台master宕机，集群会自动剔除，那又怎样判断一个redis集群失败了呢，半数以上的master挂掉了的话，无论是否有从服务器，集群进入fail状态。
redis-cluster采用虚拟槽分区，所有的键根据哈希函数映射到0-16383个整数槽内，每个节点负责维护一部分槽以及槽所映射的键值数据。三个主master分别是A,B,C，它们可以是一台机器上三个端口，也可以是三台机器，采用哈希槽(hash slot)的方式分配这16384个slot的话，三个节点分别承担slot的区间是0-5460,5461-10922,10923-16383。
搭建redis-cluster集群
- 下载配置文件
由于docker安装的redis，没有redis.conf配置文件，我们需要手动下载，https://github.com/antirez/redis/blob/unstable/redis.conf
- 修改redis.conf配置文件
我们通过上边的url下载的redis.conf是个模板文件，我们按照下边的内容，修改配置文件，由于，redis-cluster集群至少3主3从的配置，我们这里准备6个redis的配置文件，分别是7000，7001，7002，7003，7004，7005。粗体部分需要按照实际端口号修改
关闭保护模式，用于公网访问
protected-mode no
port 7000
开启集群模式
cluster-enabled yes
cluster-config-file nodes-7000.conf
后台启动
daemonize no
pidfile /var/run/redis_7000.pid
logfile "7000.log"
#bind 127.0.0.1
masterauth 123456
requirepass 123456
启动redis各个节点
# docker run -d -p 7000:7000 --net host --name myredis7000 -v /usr/local/redis/redis_cluster/7000/redis.conf:/etc/redis/redis.conf -v /usr/local/redis/redis_cluster/7000/data:/data redis redis-server /etc/redis/redis.conf --appendonly yes
# docker run -d -p 7001:7001 --net host --name myredis7001 -v /usr/local/redis/redis_cluster/7001/redis.conf:/etc/redis/redis.conf -v /usr/local/redis/redis_cluster/7001/data:/data redis redis-server /etc/redis/redis.conf --appendonly yes
# docker run -d -p 7002:7002 --net host --name myredis7002 -v /usr/local/redis/redis_cluster/7002/redis.conf:/etc/redis/redis.conf -v /usr/local/redis/redis_cluster/7002/data:/data redis redis-server /etc/redis/redis.conf --appendonly yes
# docker run -d -p 7003:7003 --net host --name myredis7003 -v /usr/local/redis/redis_cluster/7003/redis.conf:/etc/redis/redis.conf -v /usr/local/redis/redis_cluster/7003/data:/data redis redis-server /etc/redis/redis.conf --appendonly yes
# docker run -d -p 7004:7004 --net host --name myredis7004 -v /usr/local/redis/redis_cluster/7004/redis.conf:/etc/redis/redis.conf -v /usr/local/redis/redis_cluster/7004/data:/data redis redis-server /etc/redis/redis.conf --appendonly yes
docker run -d -p 7005:7005 --net host --name myredis7005 -v /usr/local/redis/redis_cluster/7005/redis.conf:/etc/redis/redis.conf -v /usr/local/redis/redis_cluster/7005/data:/data redis redis-server /etc/redis/redis.conf --appendonly yes
这里需要注意的是，启动参数带上--net host，我自己环境测试的时候，不加的话，通过create --cluster命令创建集群时，创建失败，一直等待，其他的参数的话，例如-v 主机路径映射到docker的redis中路径，
查看redis各节点的启动状况
# ps -ef | grep redis
出现下边的信息时，说明各个节点创建成功。
polkitd 14343 14325 0 10:14 ? 00:00:01 redis-server *:7000 [cluster]
polkitd 14447 14432 0 10:14 ? 00:00:01 redis-server *:7001 [cluster]
polkitd 14570 14552 0 10:14 ? 00:00:00 redis-server *:7002 [cluster]
polkitd 14702 14681 0 10:14 ? 00:00:00 redis-server *:7003 [cluster]
polkitd 14805 14790 0 10:14 ? 00:00:00 redis-server *:7004 [cluster]
polkitd 14905 14889 0 10:14 ? 00:00:00 redis-server *:7005 [cluster]
创建redis集群
# docker exec -it myredis7000 redis-cli -a 123456 --cluster create 192.168.100.103:7000 192.168.100.103:7001 192.168.100.103:7002 192.168.100.103:7003 192.168.100.103:7004 192.168.100.103:7005 --cluster-replicas 1
上边的命令执行后，需要一个确认输入，主要的意思是，为我们创建完3主3从，以及存储数据的虚拟内存槽，可以的话，输入yes，出现下边的信息的话，说明创建成功。
>>> Performing hash slots allocation on 6 nodes...
Master[0] -> Slots 0 - 5460
Master[1] -> Slots 5461 - 10922
Master[2] -> Slots 10923 - 16383
Adding replica 192.168.100.103:7004 to 192.168.100.103:7000
Adding replica 192.168.100.103:7005 to 192.168.100.103:7001
Adding replica 192.168.100.103:7003 to 192.168.100.103:7002
>>> Trying to optimize slaves allocation for anti-affinity
[WARNING] Some slaves are in the same host as their master
M: df053178fc56e3b1f8979e4009ca2d4f33bf5ff5 192.168.100.103:7000
slots:[0-5460] (5461 slots) master
M: f98948fd80d3c2069b695ccc3715662e22c7553c 192.168.100.103:7001
slots:[5461-10922] (5462 slots) master
M: 75663af230928253b8e52c869f877289fbe98834 192.168.100.103:7002
slots:[10923-16383] (5461 slots) master
S: 5385962bc7c50917ab15feeb670a2a2e4776deac 192.168.100.103:7003
replicates f98948fd80d3c2069b695ccc3715662e22c7553c
S: f059bca510ca725bafaf86ed1706ab7d88ff9af6 192.168.100.103:7004
replicates 75663af230928253b8e52c869f877289fbe98834
S: aa7956bf558774a99aa71e4d592043a93743ce67 192.168.100.103:7005
replicates df053178fc56e3b1f8979e4009ca2d4f33bf5ff5
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 192.168.100.103:7000)
M: df053178fc56e3b1f8979e4009ca2d4f33bf5ff5 192.168.100.103:7000
slots:[0-5460] (5461 slots) master
1 additional replica(s)
S: f059bca510ca725bafaf86ed1706ab7d88ff9af6 192.168.100.103:7004
slots: (0 slots) slave
replicates 75663af230928253b8e52c869f877289fbe98834
M: 75663af230928253b8e52c869f877289fbe98834 192.168.100.103:7002
slots:[10923-16383] (5461 slots) master
1 additional replica(s)
M: f98948fd80d3c2069b695ccc3715662e22c7553c 192.168.100.103:7001
slots:[5461-10922] (5462 slots) master
1 additional replica(s)
S: 5385962bc7c50917ab15feeb670a2a2e4776deac 192.168.100.103:7003
slots: (0 slots) sl
进入redis集群确认主从信息以及节点信息
走到这里，说明redis的集群已经创建成功，但是主从是怎样对应的呢，这就需要我们进入redis的集群客户端来确认，查看当前节点信息，是主，还是从，使用 info replication命令
# docker exec -it myredis7000 redis-cli -h 127.0.0.1 -c -p 7000 -a 123456
Warning: Using a password with '-a' or '-u' option on the command line interface may not be safe.
127.0.0.1:7000> info replication
# Replication
role:master
connected_slaves:1
slave0:ip=192.168.100.103,port=7005,state=online,offset=826,lag=0
master_replid:93e8fa7aaec91fba6a2fdcf2cfc3cdc201d74452
master_replid2:0000000000000000000000000000000000000000
master_repl_offset:826
master_repl_meaningful_offset:0
second_repl_offset:-1
repl_backlog_active:1
repl_backlog_size:1048576
repl_backlog_first_byte_offset:1
repl_backlog_histlen:826
127.0.0.1:7000>
或者也可以使用 cluster nodes命令，它是以节点id来区分的，
127.0.0.1:7000> cluster nodes
f059bca510ca725bafaf86ed1706ab7d88ff9af6 192.168.100.103:7004@17004 slave 75663af230928253b8e52c869f877289fbe98834 0 1589777473584 5 connected
75663af230928253b8e52c869f877289fbe98834 192.168.100.103:7002@17002 master - 0 1589777474506 3 connected 10923-16383
df053178fc56e3b1f8979e4009ca2d4f33bf5ff5 192.168.100.103:7000@17000 myself,master - 0 1589777473000 1 connected 0-5460
f98948fd80d3c2069b695ccc3715662e22c7553c 192.168.100.103:7001@17001 master - 0 1589777473993 2 connected 5461-10922
5385962bc7c50917ab15feeb670a2a2e4776deac 192.168.100.103:7003@17003 slave f98948fd80d3c2069b695ccc3715662e22c7553c 0 1589777473000 4 connected
aa7956bf558774a99aa71e4d592043a93743ce67 192.168.100.103:7005@17005 slave df053178fc56e3b1f8979e4009ca2d4f33bf5ff5 0 1589777472971 6 connected
127.0.0.1:7000>
集群测试
我的环境配置完成后，master是7000，7001,7002，所对应的从分别是7005,7003,7004，这样，就完成了集群搭建，当我们set数据是，redis会为我们将数据保存到不同端口的虚拟槽中，通过下边的命令可以看出，redis自动帮我们切换到数据存数的虚拟槽中，例如下边的-> Redirected to slot。
127.0.0.1:7000> set aa 123
OK
127.0.0.1:7000> keys *
1) "aa"
127.0.0.1:7000> get aa
"123"
127.0.0.1:7000> set bb 123
-> Redirected to slot [8620] located at 192.168.100.103:7001
OK
192.168.100.103:7001> keys *
1) "bb"
192.168.100.103:7001> get aa
-> Redirected to slot [1180] located at 192.168.100.103:7000
"123"
192.168.100.103:7000> keys *
1) "aa"
192.168.100.103:7000> get aa
"123"
192.168.100.103:7000> get bb
-> Redirected to slot [8620] located at 192.168.100.103:7001
"123"
192.168.100.103:7001> set cc 111
-> Redirected to slot [700] located at 192.168.100.103:7000
OK
192.168.100.103:7000> keys *
1) "aa"
2) "cc"
192.168.100.103:7000> get cc
"111"
192.168.100.103:7000> set dd 123
-> Redirected to slot [11212] located at 192.168.100.103:7002
OK
192.168.100.103:7002> get aa
-> Redirected to slot [1180] located at 192.168.100.103:7000
"123"
192.168.100.103:7000> get bb
-> Redirected to slot [8620] located at 192.168.100.103:7001
"123"
192.168.100.103:7001> get cc
-> Redirected to slot [700] located at 192.168.100.103:7000
"111"
192.168.100.103:7000> get dd
-> Redirected to slot [11212] located at 192.168.100.103:7002
"123"
192.168.100.103:7002> get dd
"123"
192.168.100.103:7002> get ee
-> Redirected to slot [2268] located at 192.168.100.103:7000
(nil)
192.168.100.103:7000> get aa
"123"
192.168.100.103:7000> get bb
-> Redirected to slot [8620] located at 192.168.100.103:7001
"123"
192.168.100.103:7001> keys *
1) "bb"
192.168.100.103:7001> get aa
-> Redirected to slot [1180] located at 192.168.100.103:7000
"123"
192.168.100.103:7000> keys *
1) "aa"
2) "cc"
192.168.100.103:7000> set e 1
-> Redirected to slot [15363] located at 192.168.100.103:7002
OK
192.168.100.103:7002> keys *
1) "dd"
2) "e"
高可用的redis集群的配置就OK了， 怎么样，小伙伴们搭建的还顺利吗，中间过程肯定会有和想象中不一样的地方，我也是一边baidu，一边试，这样会加深对docker，redis，linux命令的理解以及熟练度，大家动手试试，有什么不一样的地方，或者困惑的地方，欢迎来一起交流。
扫一扫
671
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
