Redis cluster 簡介
1.採用去中心化的思想，沒有中心節點的說法，它使用hash slot方式將16348個hash slot覆蓋到所有節點上，對於存儲的每個key值，使用CRC16（KEY）&16348=slot得到他對應的hash slot，並在訪問key的時候就去找他的hash slot在哪一個節點上，然後由當前訪問節點從實際被分配了這個hash slot的節點去取數據，節點之間使用輕量協議通信 減少帶寬佔用 性能很高，自動實現負載均衡與高可用，自動實現failover並且支持動態擴展，官方已經玩到可以1000個節點 實現的複雜度低。
2.其內部中也需要配置主從，並且內部也是採用哨兵模式，如果有半數節點發現某個異常節點，共同決定更改異常節點的狀態，如果改節點是主節點，則對應的從節點自動頂替為主節點，當原先的主節點上線後，則會變為從節點。如果集群中的master沒有slave節點，則master掛掉後整個集群就會進入fail狀態，因為集群的slot映射不完整。如果集群超過半數以上的master掛掉，無論是否有slave，集群都會進入fail狀態。
3.根據官方推薦 集群部署至少要3臺以上的master節點。
搭建過程
Ip 端口
192.168.83.128 6000
192.168.83.128 6001
192.168.83.128 6002
192.168.83.128 6003
192.168.83.128 6004
192.168.83.128 6005
1. 先安裝redis
cd /usr/local
wget http://download.redis.io/releases/redis-5.0.5.tar.gz
tar xzf redis-5.0.5.tar.gz
cd redis-5.0.5/
make
2.創建目錄
mkdir rediscluster
cd rediscluster/
[root@localhost rediscluster]# mkdir redis6000
[root@localhost rediscluster]# mkdir redis6001
[root@localhost rediscluster]# mkdir redis6002
[root@localhost rediscluster]# mkdir redis6003
[root@localhost rediscluster]# mkdir redis6004
[root@localhost rediscluster]# mkdir redis6005
在每個目錄下面創建 data和log
3.啟動參數
存放目錄 /usr/local/rediscluster/redis6000/redis6000.conf,由於是6個節點，每個節點的目錄和端口是不一樣的，請參照這個進行修改
protected-mode no
port 6005
pidfile "/var/run/redis_6005.pid"
save 9 1
save 300 10
save 60 10000
daemonize yes
timeout 1800
databases 16
loglevel notice
logfile "/usr/local/rediscluster/redis6005/log/redis.log"
rdbcompression yes
dbfilename "dump.rdb"
dir "/usr/local/rediscluster/redis6005/data"
maxclients 50000
maxmemory 1063256064
stop-writes-on-bgsave-error yes
appendonly yes
appendfilename "appendonly.aof"
appendfsync everysec
auto-aof-rewrite-percentage 100
auto-aof-rewrite-min-size 128mb
activerehashing yes
rdbchecksum yes
#vw-enabled yes
aof-use-rdb-preamble yes
aof-load-truncated yes
no-appendfsync-on-rewrite yes
lua-time-limit 5000
slowlog-log-slower-than 10000
slowlog-max-len 1000
latency-monitor-threshold 0
notify-keyspace-events ""
hash-max-ziplist-entries 512
hash-max-ziplist-value 64
list-max-ziplist-entries 512
list-max-ziplist-value 64
set-max-intset-entries 512
zset-max-ziplist-entries 128
zset-max-ziplist-value 64
hz 10
aof-rewrite-incremental-fsync yes
#repliaction
#masterauth "123456"
#requirepass "123456"
#slave-serve-stale-data yes
#slave-read-only yes
#repl-diskless-sync no
#repl-diskless-sync-delay 5
#repl-disable-tcp-nodelay no
#slave-priority 100
#cluster
cluster-enabled yes
cluster-config-file nodes_6005.conf
cluster-node-timeout 15000
4.啟動redis
[root@localhost redis-5.0.5]# src/redis-server /usr/local/rediscluster/redis6000/redis6000.conf
[root@localhost redis-5.0.5]# src/redis-server /usr/local/rediscluster/redis6001/redis6001.conf
[root@localhost redis-5.0.5]# src/redis-server /usr/local/rediscluster/redis6002/redis6002.conf
[root@localhost redis-5.0.5]# src/redis-server /usr/local/rediscluster/redis6003/redis6003.conf
[root@localhost redis-5.0.5]# src/redis-server /usr/local/rediscluster/redis6004/redis6004.conf
[root@localhost redis-5.0.5]# src/redis-server /usr/local/rediscluster/redis6005/redis6005.conf
[root@localhost redis-5.0.5]# ps -ef |grep redis
root 57466 1 0 10:45 ? 00:00:00 src/redis-server *:6000 [cluster]
root 57473 1 0 10:46 ? 00:00:00 src/redis-server *:6001 [cluster]
root 57487 1 0 10:46 ? 00:00:00 src/redis-server *:6002 [cluster]
root 57492 1 0 10:46 ? 00:00:00 src/redis-server *:6003 [cluster]
root 57498 1 0 10:46 ? 00:00:00 src/redis-server *:6004 [cluster]
root 57504 1 2 10:46 ? 00:00:00 src/redis-server *:6005 [cluster]
root 57510 52442 0 10:46 pts/0 00:00:00 grep --color=auto redis
5.啟動後我們就可以創建集群
注意：
在redis5.0後 創建集群統一使用redis-cli，之前的版本使用redis-trib.rb，但是需要安裝ruby軟件相對複雜,相比之前的版本5.0不需要安裝額外的軟件，方便。具體的可以參照redis官方網站查看
https://redis.io/topics/cluster-tutorial
創建集群命令：其中 cluster-replicas 1 代表 一個master後有幾個slave，1代表為1個slave節點
src/redis-cli --cluster create 192.168.83.128:6000 192.168.83.128:6001 192.168.83.128:6002 192.168.83.128:6003 192.168.83.128:6004 192.168.83.128:6005 --cluster-replicas 1
>>> Performing hash slots allocation on 6 nodes...
Master[0] -> Slots 0 - 5460
Master[1] -> Slots 5461 - 10922
Master[2] -> Slots 10923 - 16383
Adding replica 192.168.83.128:6004 to 192.168.83.128:6000
Adding replica 192.168.83.128:6005 to 192.168.83.128:6001
Adding replica 192.168.83.128:6003 to 192.168.83.128:6002
>>> Trying to optimize slaves allocation for anti-affinity
[WARNING] Some slaves are in the same host as their master
M: 6104062a1f3e7f183a01064ca225a43d903c6fa9 192.168.83.128:6000
slots:[0-5460] (5461 slots) master
M: 236ef722de05fb03af442eb528b7aa12da482f35 192.168.83.128:6001
slots:[5461-10922] (5462 slots) master
M: 94f272ebb67027861a5e788b86388b979e16a4c3 192.168.83.128:6002
slots:[10923-16383] (5461 slots) master
S: 72de2e9a41f4b4d5f155efac41aebbeb90b5af7b 192.168.83.128:6003
replicates 94f272ebb67027861a5e788b86388b979e16a4c3
S: 17900d952666c054aeb169c2880d49662b222a58 192.168.83.128:6004
replicates 6104062a1f3e7f183a01064ca225a43d903c6fa9
S: 16d1233717f7f2c26f398e155328b324a4d26449 192.168.83.128:6005
replicates 236ef722de05fb03af442eb528b7aa12da482f35
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 192.168.83.128:6000)
M: 6104062a1f3e7f183a01064ca225a43d903c6fa9 192.168.83.128:6000
slots:[0-5460] (5461 slots) master
1 additional replica(s)
S: 17900d952666c054aeb169c2880d49662b222a58 192.168.83.128:6004
slots: (0 slots) slave
replicates 6104062a1f3e7f183a01064ca225a43d903c6fa9
M: 236ef722de05fb03af442eb528b7aa12da482f35 192.168.83.128:6001
slots:[5461-10922] (5462 slots) master
1 additional replica(s)
M: 94f272ebb67027861a5e788b86388b979e16a4c3 192.168.83.128:6002
slots:[10923-16383] (5461 slots) master
1 additional replica(s)
S: 72de2e9a41f4b4d5f155efac41aebbeb90b5af7b 192.168.83.128:6003
slots: (0 slots) slave
replicates 94f272ebb67027861a5e788b86388b979e16a4c3
S: 16d1233717f7f2c26f398e155328b324a4d26449 192.168.83.128:6005
slots: (0 slots) slave
replicates 236ef722de05fb03af442eb528b7aa12da482f35
[OK] All nodes agree about slots configuration.
>>> Check for open slots...
>>> Check slots coverage...
[OK] All 16384 slots covered.
6.我們驗證一下集群是否創建成功
登錄redis客戶端 src/redis-cli -c -p 6000， -c 參數代表連接到集群中
我們先看一下主從配對情況，根據node ID匹配可以得出配比。大家可以看看是否有對應的ID，這樣我們就區分了主從節點的關係了。
[root@localhost redis-5.0.5]# src/redis-cli -c -p 6000
127.0.0.1:6000> cluster nodes
17900d952666c054aeb169c2880d49662b222a58 192.168.83.128:6004@16004 slave 6104062a1f3e7f183a01064ca225a43d903c6fa9 0 1578021017661 5 connected
236ef722de05fb03af442eb528b7aa12da482f35 192.168.83.128:6001@16001 master - 0 1578021016656 2 connected 5461-10922
94f272ebb67027861a5e788b86388b979e16a4c3 192.168.83.128:6002@16002 master - 0 1578021015000 3 connected 10923-16383
6104062a1f3e7f183a01064ca225a43d903c6fa9 192.168.83.128:6000@16000 myself,master - 0 1578021015000 1 connected 0-5460
72de2e9a41f4b4d5f155efac41aebbeb90b5af7b 192.168.83.128:6003@16003 slave 94f272ebb67027861a5e788b86388b979e16a4c3 0 1578021016000 4 connected
16d1233717f7f2c26f398e155328b324a4d26449 192.168.83.128:6005@16005 slave 236ef722de05fb03af442eb528b7aa12da482f35 0 1578021016000 6 connected
查看群集信息
[root@localhost redis-5.0.5]# src/redis-cli -c -p 6000
127.0.0.1:6000> 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:7
cluster_my_epoch:1
cluster_stats_messages_ping_sent:2368
cluster_stats_messages_pong_sent:1921
cluster_stats_messages_fail_sent:4
cluster_stats_messages_auth-ack_sent:1
cluster_stats_messages_update_sent:1
cluster_stats_messages_sent:4295
cluster_stats_messages_ping_received:1916
cluster_stats_messages_pong_received:2020
cluster_stats_messages_meet_received:5
cluster_stats_messages_fail_received:1
cluster_stats_messages_auth-req_received:1
cluster_stats_messages_received:3943
驗證數據匹配是否採用哈希槽的方式。我在6000客戶端設置的數據，被分配到6004上了。
127.0.0.1:6000> set aa "1111"
OK
[root@localhost redis-5.0.5]# src/redis-cli -c -p 6004
127.0.0.1:6004> get aa
-> Redirected to slot [1180] located at 192.168.83.128:6000
"1111"
7.驗證故障轉移
我們從上面可以看出6002為master，6003為其從節點。我們按照如下流程進行操作
先停掉6002，發現6003轉為了master節點，然後恢復6002上線，6002變為slave節點。
[root@localhost redis-5.0.5]# ps -ef |grep redis
root 57919 1 0 11:05 ? 00:00:02 src/redis-server *:6000 [cluster]
root 57925 1 0 11:05 ? 00:00:02 src/redis-server *:6001 [cluster]
root 57931 1 0 11:05 ? 00:00:02 src/redis-server *:6002 [cluster]
root 57937 1 0 11:05 ? 00:00:02 src/redis-server *:6003 [cluster]
root 57943 1 0 11:05 ? 00:00:02 src/redis-server *:6004 [cluster]
root 57949 1 0 11:06 ? 00:00:02 src/redis-server *:6005 [cluster]
root 58505 52442 0 11:36 pts/0 00:00:00 grep --color=auto redis
[root@localhost redis-5.0.5]# kill -9 57931
[root@localhost redis-5.0.5]# ps -ef |grep redis
root 57919 1 0 11:05 ? 00:00:02 src/redis-server *:6000 [cluster]
root 57925 1 0 11:05 ? 00:00:02 src/redis-server *:6001 [cluster]
root 57937 1 0 11:05 ? 00:00:02 src/redis-server *:6003 [cluster]
root 57943 1 0 11:05 ? 00:00:02 src/redis-server *:6004 [cluster]
root 57949 1 0 11:06 ? 00:00:02 src/redis-server *:6005 [cluster]
root 58509 52442 0 11:37 pts/0 00:00:00 grep --color=auto redis
[root@localhost redis-5.0.5]# src/redis-server /usr/local/rediscluster/redis6002/redis6002.conf
[root@localhost redis-5.0.5]# src/redis-cli -c -p 6000
127.0.0.1:6000> cluster nodes
17900d952666c054aeb169c2880d49662b222a58 192.168.83.128:6004@16004 slave 6104062a1f3e7f183a01064ca225a43d903c6fa9 0 1578022689000 5 connected
236ef722de05fb03af442eb528b7aa12da482f35 192.168.83.128:6001@16001 master - 0 1578022689187 2 connected 5461-10922
94f272ebb67027861a5e788b86388b979e16a4c3 192.168.83.128:6002@16002 slave 72de2e9a41f4b4d5f155efac41aebbeb90b5af7b 0 1578022690194 7 connected
6104062a1f3e7f183a01064ca225a43d903c6fa9 192.168.83.128:6000@16000 myself,master - 0 1578022689000 1 connected 0-5460
72de2e9a41f4b4d5f155efac41aebbeb90b5af7b 192.168.83.128:6003@16003 master - 0 1578022688176 7 connected 10923-16383
16d1233717f7f2c26f398e155328b324a4d26449 192.168.83.128:6005@16005 slave 236ef722de05fb03af442eb528b7aa12da482f35 0 1578022687000 6 connected
8.集群節點擴容
在rediscluster,新增2個redis節點配置6006和6007兩個端口。將6006作為新增的master節點，6007作為從節點，然後啟動節點：
添加主節點：
./redis-cli --cluster add-node 192.168.83.128:6006 192.168.83.128:6000
添加從節點：
./redis-cli --cluster add-node 192.168.83.128:6007 192.168.83.128:6000 --cluster-slave --cluster-master-id 22e8a8e97d6f7cc7d627e577a986384d4d181a4f(此ID為6006端口的ID)
添加成功後，為新節點分配數據：
./redis-cli --cluster reshard 192.168.83.128:6006
這樣就新增完畢了，可以通過cluster nodes命令查看一下新增後的slot分佈
9.集群節點縮減
縮減節點時先縮減從節點:
./redis-cli --cluster del-node 192.168.83.128:6007
然後進行主節點slot轉移：
./redis-cli --cluster reshard 192.168.83.128:6006
最後在縮減主節點
./redis-cli --cluster del-node 192.168.83.128:6006
閱讀更多 碩果累累1316 的文章
