哨兵架构存在的一些缺陷：
1. 瞬断问题，哨兵架构中，当主节点挂断后需要选举，然后切换，中间有几秒到几十秒的服务暂停时间
2. 高并发受限问题，哨兵对外只有一个节点提供写服务，一台就几w的QPS。
3. 单个节点的内存不能设置过大，也就10G左右，否者导致持久化文件过大，影响数据恢复和主从同步效率。
Redis集群架构可以解决上面提到的问题，
redis集群是有多个主从节点小群组成的分布式服务器集群
特点： 复制，高可用，分片特性，没有中心节点，可水平扩容。官网称可以支持上万个节点（推荐不超过1000），配置简单。
瞬断的问题还是没有彻底解决，如果刚好要访问的哪个key所在的小集群挂了，也会等待那个小集群选举并重新起来。但是对于位于其他集群的key服务是可以正常服务。
下面为了简便，搭建的时候是一主一从，不是上面的一主2从的结构。
reids高可用集群的搭建：
扫描二维码关注公众号，回复：
12165343 查看本文章
1. 根据已经安装好redis的机器，通过虚拟机克隆3台机器出来
2. 设置3台机器的ip地址分别为192.168.0.101，192.168.0.102，192.168.0.103
配置网络的教程：https://blog.csdn.net/pengweismile/article/details/112910138
3. 分别在3台机器上做以下配置，
3.1 新建目录，修改配置文件
cd /home/allen/packages/redis-5.0.3/
mkdir redis-cluster
cd redis-cluster
mkdir 8001 8004
cp /home/allen/packages/redis-5.0.3/conf/redis.config 8001
cp /home/allen/packages/redis-5.0.3/conf/redis.config 8004
vi 8001/redis.config3.2 在redis.config文件中配置如下信息，最主要的是下面的加粗字段：
port 8001
protected-mode no
daemonize yes
pidfile "/var/run/redis_8001.pid"
logfile "8001.log"
dir "/home/allen/packages/redis-5.0.3/8001"
masterauth allen
replica-read-only yes
requirepass allen
appendonly yes
appendfilename "appendonly.aof"
cluster-enabled yes
cluster-config-file nodes-8001.conf
cluster-node-timeout 100004.在各个机器的redis-5.0.3目中创建端口号对应得目录，存储log数据等
cd /home/allen/packages/redis-5.0.3
mkdir 8001 80044.1. 分别在192.168.0.102 得/home/allen/packages/redis-5.0.3/redis-cluster下创建目录8002，8005，192.168.0.102 上创建8003，8006， 然后被配置文件redis.config copy进去
cd /home/allen/packages/redis-5.0.3/redis-cluster
mkdir 8001 80045.分别在对应目录下的redis.config文件中的进行全部替换,达到修改每个配置文件得目的
:%s/8001/8004/g
6.在各个机器上分别启动redis服务端
[allen@localhost redis-5.0.3]$ src/redis-server redis-cluster/8001/redis.conf
[allen@localhost redis-5.0.3]$ src/redis-server redis-cluster/8004/redis.conf
[allen@localhost redis-5.0.3]$ src/redis-server redis-cluster/8002/redis.conf
[allen@localhost redis-5.0.3]$ src/redis-server redis-cluster/8005/redis.conf
[allen@localhost redis-5.0.3]$ src/redis-server redis-cluster/8003/redis.conf
[allen@localhost redis-5.0.3]$ src/redis-server redis-cluster/8006/redis.conf 7. 创建集群，贴出启动成功的log
src/redis-cli -a allen --cluster create 192.168.0.101:8001 192.168.0.101:8004 192.168.0.102:8002 192.168.0.102:8005 192.168.0.103:8003 192.168.0.103:8006 --cluster-replicas 1
[allen@localhost redis-5.0.3]$ src/redis-cli -a allen --cluster create 192.168.0.101:8001 192.168.0.101:8004 192.168.0.102:8002 192.168.0.102:8005 192.168.0.103:8003 192.168.0.103:8006 --cluster-replicas 1
Warning: Using a password with '-a' or '-u' option on the command line interface may not be safe.
>>> Performing hash slots allocation on 6 nodes...
Master[0] -> Slots 0 - 5460
Master[1] -> Slots 5461 - 10922
Master[2] -> Slots 10923 - 16383
Adding replica 192.168.0.102:8005 to 192.168.0.101:8001
Adding replica 192.168.0.101:8004 to 192.168.0.102:8002
Adding replica 192.168.0.103:8006 to 192.168.0.103:8003
>>> Trying to optimize slaves allocation for anti-affinity
[OK] Perfect anti-affinity obtained!
M: 7e6aa35995b79bef083cbfe60855d6d4ec3ba999 192.168.0.101:8001
slots:[0-5460] (5461 slots) master
S: 499787d7590f9df6d102339c1341c00a1585726f 192.168.0.101:8004
replicates 0f1f5538d02a78e9e5266afc285a4a524e3daa51
M: 383e0df8c3e16aacf86a17cbb7b2819426f9e454 192.168.0.102:8002
slots:[5461-10922] (5462 slots) master
S: 3689c09d371ea3b0b11b4cc3827412220a8b6a11 192.168.0.102:8005
replicates 7e6aa35995b79bef083cbfe60855d6d4ec3ba999
M: 0f1f5538d02a78e9e5266afc285a4a524e3daa51 192.168.0.103:8003
slots:[10923-16383] (5461 slots) master
S: 9a5e3238d9648eba8084c6f5ad62814c61fb73b3 192.168.0.103:8006
replicates 383e0df8c3e16aacf86a17cbb7b2819426f9e454
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 192.168.0.101:8001)
M: 7e6aa35995b79bef083cbfe60855d6d4ec3ba999 192.168.0.101:8001
slots:[0-5460] (5461 slots) master
1 additional replica(s)
S: 3689c09d371ea3b0b11b4cc3827412220a8b6a11 192.168.0.102:8005
slots: (0 slots) slave
replicates 7e6aa35995b79bef083cbfe60855d6d4ec3ba999
M: 383e0df8c3e16aacf86a17cbb7b2819426f9e454 192.168.0.102:8002
slots:[5461-10922] (5462 slots) master
1 additional replica(s)
S: 9a5e3238d9648eba8084c6f5ad62814c61fb73b3 192.168.0.103:8006
slots: (0 slots) slave
replicates 383e0df8c3e16aacf86a17cbb7b2819426f9e454
M: 0f1f5538d02a78e9e5266afc285a4a524e3daa51 192.168.0.103:8003
slots:[10923-16383] (5461 slots) master
1 additional replica(s)
S: 499787d7590f9df6d102339c1341c00a1585726f 192.168.0.101:8004
slots: (0 slots) slave
replicates 0f1f5538d02a78e9e5266afc285a4a524e3daa51
[OK] All nodes agree about slots configuration.
>>> Check for open slots...
>>> Check slots coverage...
[OK] All 16384 slots covered.
[allen@localhost redis-5.0.3]$ 8. 查看用cluster的形式登录客户端，并查看cluster信息，可以看到cluster_know_nodes:6是6台机器。
[allen@localhost redis-5.0.3]$ src/redis-cli -a allen -c -h 192.168.0.101 -p 8004
Warning: Using a password with '-a' or '-u' option on the command line interface may not be safe.
192.168.0.101:8004> cluster info
cluster_state:ok
cluster_slots_assigned:16384
cluster_slots_ok:16384
cluster_slots_pfail:0
cluster_slots_fail:0
cluster_known_nodes:6
cluster_size:3
cluster_current_epoch:6
cluster_my_epoch:5
cluster_stats_messages_ping_sent:104
cluster_stats_messages_pong_sent:113
cluster_stats_messages_meet_sent:4
cluster_stats_messages_sent:221
cluster_stats_messages_ping_received:112
cluster_stats_messages_pong_received:108
cluster_stats_messages_meet_received:1
cluster_stats_messages_received:221
9. 查看cluster nodes可以看到每台机器的主从结构，这里可以看到这里的主重都是分别在不同的机器的。就是同一台ip的2个实例不会是主从结构。
192.168.0.101， 8001 8004
192.168.0.102， 8002 8005
192.168.0.103, 8003 8006
8002(主)-8006(从)
8001(主)-8005(从)
8003(主)-8004(从)
192.168.0.101:8004> cluster nodes
9a5e3238d9648eba8084c6f5ad62814c61fb73b3 192.168.0.103:8006@18006 slave 383e0df8c3e16aacf86a17cbb7b2819426f9e454 0 1611145153000 6 connected
3689c09d371ea3b0b11b4cc3827412220a8b6a11 192.168.0.102:8005@18005 slave 7e6aa35995b79bef083cbfe60855d6d4ec3ba999 0 1611145154081 4 connected
0f1f5538d02a78e9e5266afc285a4a524e3daa51 192.168.0.103:8003@18003 master - 0 1611145152121 5 connected 10923-16383
383e0df8c3e16aacf86a17cbb7b2819426f9e454 192.168.0.102:8002@18002 master - 0 1611145153172 3 connected 5461-10922
7e6aa35995b79bef083cbfe60855d6d4ec3ba999 192.168.0.101:8001@18001 master - 0 1611145154181 1 connected 0-5460
499787d7590f9df6d102339c1341c00a1585726f 192.168.0.101:8004@18004 myself,slave 0f1f5538d02a78e9e5266afc285a4a524e3daa51 0 1611145151000 2 connected
192.168.0.101:8004>
10. 可以看到，以上的信息都记录在了每天机器的文件中，如下，所以当你期间重启后，不用再去创建集群，只要每台集群的redis服务起来，集群就会自己其起来，根据下面的这些信息。
[allen@localhost redis-5.0.3]$ pwd
/home/allen/packages/redis-5.0.3
[allen@localhost redis-5.0.3]$ cd 8001
[allen@localhost 8001]$ ls
8001.log appendonly.aof nodes-8001.conf
[allen@localhost 8001]$ less nodes-8001.conf
[allen@localhost 8001]$ cat nodes-8001.conf
3689c09d371ea3b0b11b4cc3827412220a8b6a11 192.168.0.102:8005@18005 slave 7e6aa35995b79bef083cbfe60855d6d4ec3ba999 0 1611145053207 4 connected
383e0df8c3e16aacf86a17cbb7b2819426f9e454 192.168.0.102:8002@18002 master - 0 1611145050117 3 connected 5461-10922
9a5e3238d9648eba8084c6f5ad62814c61fb73b3 192.168.0.103:8006@18006 slave 383e0df8c3e16aacf86a17cbb7b2819426f9e454 0 1611145052000 6 connected
0f1f5538d02a78e9e5266afc285a4a524e3daa51 192.168.0.103:8003@18003 master - 0 1611145052000 5 connected 10923-16383
7e6aa35995b79bef083cbfe60855d6d4ec3ba999 192.168.0.101:8001@18001 myself,master - 0 1611145050000 1 connected 0-5460
499787d7590f9df6d102339c1341c00a1585726f 192.168.0.101:8004@18004 slave 0f1f5538d02a78e9e5266afc285a4a524e3daa51 0 1611145052187 5 connected
vars currentEpoch 6 lastVoteEpoch 0
[allen@localhost 8001]$