Redis6.0集群环境搭建

原创

用户2141488

修改于 2021-02-10 12:10:27

2.6K0

修改于 2021-02-10 12:10:27

Redis集群(Redis Cluster) 是Redis提供的分布式数据库方案，通过 分片(sharding) 来进行数据共享，并提供复制和故障转移功能。相比于主从复制、哨兵模式，Redis集群实现了较为完善的高可用方案，解决了存储能力受到单机限制，写操作无法负载均衡的问题。

本文是Redis集群学习的实践总结，详细介绍逐步搭建Redis集群环境的过程，并完成集群伸缩的实践。

测试环境

centos8.3
redis 6.0.6

redis安装

cat redis_install.sh

#!/bin/bash

yum update -y
cd /usr/local/src/
wget http://download.redis.io/releases/redis-6.0.6.tar.gz
tar zxvf redis-6.0.6.tar.gz
cd redis-6.0.6/
make
make install
ln -s /usr/local/src/redis-6.0.6/src/redis-server /usr/bin/redis-server
ln -s /usr/local/src/redis-6.0.6/src/redis-cli /usr/bin/redis-cli

集群搭建

因是测试，这里集群环境的所有节点全部位于同一个服务器上，共6个节点以端口号区分，3个主节点+3个从节点（分别是6379、6479、6579、6679、6779、6879）。

注意：从Redis 5.0以后的版本，集群管理软件 redis-trib.rb 被集成到 redis-cli 客户端工具中。

集群的搭建可以分为四步：

启动节点：将节点以集群方式启动（此时节点是独立的）
节点握手：将独立的节点连成网络
槽（solts）分配：将16384个槽位分配给主节点，以达到分片保存数据库键值对的效果。
主从复制：为从节点指定主节点。

每个节点初始状态仍为 master服务器，唯一不同的是：使用 Cluster 模式启动。需要对配置文件进行修改，以端口号为6379的节点为例，主要修改如下几项：

port 6379 # 启动端口
cluster-enabled yes # 开启集群
pidfile /var/run/redis_6379.pid # pid存储目录
cluster-config-file "node-6379.conf" # 集群节点配置文件,这个文件是不能手动编辑的.确保每一个集群节点的配置文件不通
logfile "redis-server-6379.log" # 日志存储目录
dbfilename "dump-6379.rdb" # 数据存储
daemonize yes # 守护进程
appendonly yes # 持久化

为了方便，可以使用`sed`进行批量替换，下面的批量部署脚本可以参考下

#!/bin/bash


# 安装redis6.0
if [[ `redis-server -v` == "Redis server"* ]]; then
    echo "redis已安装"
else
    yum update -y
    cd /usr/local/src/
    wget http://download.redis.io/releases/redis-6.0.6.tar.gz
    tar zxvf redis-6.0.6.tar.gz
    cd redis-6.0.6/
    make && make install
    ln -s /usr/local/src/redis-6.0.6/src/redis-server /usr/bin/redis-server
    ln -s /usr/local/src/redis-6.0.6/src/redis-cli /usr/bin/redis-cli
fi



# 定义安装目录
install_pash="/usr/local/src/redis-6.0.6"

mkdir /var/log/redis 	# 创建日志目录
mkdir ${install_pash}/dump.rdb 	# 创建数据存储目录
mkdir ${install_pash}/nodes.conf 	# 创建节点配置文件目录
cp ${install_pash}/redis.conf ${install_pash}/redis.conf.bak 	# 备份原始配置文件



# 设置端口号
ports_arr=(6379 6479 6579 6679 6779 6879)

for ports in ${ports_arr[@]}
do
    cp ${install_pash}/redis.conf ${install_pash}/redis-$ports.conf
    # 定义启动端口
    sed -i 's#port 6379#port '${ports}'#g' ${install_pash}/redis-$ports.conf
    # 守护进程（后台运行）
    sed -i 's/daemonize no/daemonize yes/g' ${install_pash}/redis-$ports.conf
    # 持久化
    sed -i 's/appendonly no/appendonly yes/g' ${install_pash}/redis-$ports.conf
    # 保存数据的AOF文件名
    sed -i 's#appendfilename "appendonly.aof"#appendfilename "appendonly_'${ports}'.aof"#g' ${install_pash}/redis-$ports.conf
    # 开启集群
    sed -i 's/# cluster-enabled yes/cluster-enabled yes/g' ${install_pash}/redis-$ports.conf
    # 日志文件
    sed -i 's#logfile ""#logfile "/var/log/redis/'${ports}'.log"#g' ${install_pash}/redis-$ports.conf
    # pid文件
    sed -i 's#pidfile /var/run/redis_6379.pid#pidfile /var/run/redis_'${ports}'.pid#g' ${install_pash}/redis-$ports.conf
    # 数据存储文件所在目录
    sed -i 's@dir ./@dir "'${install_pash}'/dump.rdb/"@g' ${install_pash}/redis-$ports.conf
    # 数据存储文件
    sed -i 's@dbfilename dump.rdb@dbfilename "dump-'${ports}'.rdb"@g' ${install_pash}/redis-$ports.conf
    # 集群节点配置文件,这个文件是不能手动编辑的.确保每一个集群节点的配置文件不同
    sed -i 's@# cluster-config-file nodes-6379.conf@cluster-config-file '${install_pash}'/nodes.conf/nodes-'${ports}'.conf@g' ${install_pash}/redis-$ports.conf

    # 启动redis服务
    redis-server ${install_pash}/redis-$ports.conf

done

sleep 3

# 查看redis服务是否启动
ps -aux | grep redis

其中 cluster-config-file 参数指定了集群配置文件的位置，每个节点在运行过程中，会维护一份集群配置文件；每当集群信息发生变化时（如增减节点），集群内所有节点会将最新信息更新到该配置文件；当节点重启后，会重新读取该配置文件，获取集群信息，可以方便的重新加入到集群中。也就是说，当Redis节点以集群模式启动时，会首先寻找是否有集群配置文件，如果有则使用文件中的配置启动，如果没有，则初始化配置并将配置保存到文件中。集群配置文件由Redis节点维护，不需要人工修改。

1.1 启动节点

为6个节点修改好相应的配置文件后，即可利用 redis-server redis-xxxx.conf 工具启动6个服务器（xxxx表示端口号，对应相应的配置文件）。利用ps命令查看进程：

~]# ps -axu | grep redis
root       59134  0.0  0.2  61788  9636 ?        Ssl  21:47   0:00 redis-server 127.0.0.1:6379 [cluster]
root       59136  0.0  0.2  61788  9684 ?        Ssl  21:47   0:00 redis-server 127.0.0.1:6479 [cluster]
root       59142  0.0  0.2  61788  9704 ?        Ssl  21:47   0:00 redis-server 127.0.0.1:6579 [cluster]
root       59148  0.0  0.2  61788  9724 ?        Ssl  21:47   0:00 redis-server 127.0.0.1:6679 [cluster]
root       59154  0.0  0.2  61788  9628 ?        Ssl  21:47   0:00 redis-server 127.0.0.1:6779 [cluster]
root       59160  0.0  0.2  61788  9704 ?        Ssl  21:47   0:00 redis-server 127.0.0.1:6879 [cluster]
root       59182  0.0  0.0  12112  1068 pts/0    S+   21:47   0:00 grep --color=auto redis

1.2 节点握手

1.1中将每个节点启动后，节点间是相互独立的，他们都处于一个只包含自己的集群当中，以端口号6379的服务器为例，利用 CLUSTER NODES 查看当前集群包含的节点。

[root@VM-0-10-centos ~]# redis-cli -p 6379 -c cluster nodes
162be0c0e035e9dfb52765e41a33069cd021f8b0 :6379@16379 myself,master - 0 0 0 connected

我们需要将各个独立的节点连接起来，构成一个包含多个节点的集群，使用 CLUSTER MEET 命令。

[root@VM-0-15-centos ~]# redis-cli -p 6379 -c cluster meet 127.0.0.1 6479
OK
[root@VM-0-15-centos ~]# redis-cli -p 6379 -c cluster meet 127.0.0.1 6579
OK
[root@VM-0-15-centos ~]# redis-cli -p 6379 -c cluster meet 127.0.0.1 6679
OK
[root@VM-0-15-centos ~]# redis-cli -p 6379 -c cluster meet 127.0.0.1 6779
OK
[root@VM-0-15-centos ~]# redis-cli -p 6379 -c cluster meet 127.0.0.1 6879
OK

再次查看此时集群中包含的节点情况：

[root@VM-0-15-centos ~]# redis-cli -p 6379 -c cluster nodes
693de0512d73a16c7525e3d85e79f6da4fcf67b2 127.0.0.1:6779@16779 master - 0 1611474390792 4 connected
d3a3d410da3db56d33dd34df8eb8f6fd3cdbced6 127.0.0.1:6579@16579 master - 0 1611474389000 0 connected
c4dbef9824daa77c553e716a968b1628d78f5377 127.0.0.1:6679@16679 master - 0 1611474389789 3 connected
162be0c0e035e9dfb52765e41a33069cd021f8b0 127.0.0.1:6379@16379 myself,master - 0 1611474387000 1 connected
92f8bc8d863e73b94b04d3f612eebf6c8b9e4834 127.0.0.1:6479@16479 master - 0 1611474389000 2 connected
1b403710ba735ad76533caa3d8f5627cf45bbda3 127.0.0.1:6879@16879 master - 0 1611474388786 5 connected

可以发现此时6个节点均作为主节点加入到集群中， CLUSTER NODES 返回的结果各项含义如下：

<id> <ip:port@cport> <flags> <master> <ping-sent> <pong-recv> <config-epoch> <link-state> <slot> <slot> ... <slot>

节点id: 由40个16进制字符串组成，节点id只在集群初始化时创建一次，然后保存到集群配置文件（即前文提到的cluster-config-file）中，以后节点重新启动时会直接在集群配置文件中读取。
port@cport: 前者为普通端口，用于为客户端提供服务；后者为集群端口，分配方法为：普通端口+10000，只用于节点间的通讯。

其余各项的详细解释可以参考文档cluster nodes(集群节点)

1.3 槽（solts）分配（分片）

Redis集群通过分片(sharding)的方式保存数据库的键值对，整个数据库被分为16384个槽(slot)，数据库每个键都属于这16384个槽的一个，集群中的每个节点都可以处理0个或者最多16384个slot。

槽是数据管理和迁移的基本单位。当数据库中的16384个槽都分配了节点时，集群处于上线状态（ok）；如果有任意一个槽没有分配节点，则集群处于下线状态（fail）。

注意，只有主节点有处理槽的能力，如果将槽指派步骤放在主从复制之后，并且将槽位分配给从节点，那么集群将无法正常工作（处于下线状态）。

命令格式：redis-cli -h <节点ip> -p <节点端口> cluster addslots <指派的solts数量>

~]# redis-cli -h 127.0.0.1 -p 6379 cluster addslots {0..5000}
OK
~]# redis-cli -h 127.0.0.1 -p 6579 cluster addslots {5001..10000}
OK
~]# redis-cli -h 127.0.0.1 -p 6779 cluster addslots {10001..16383}
OK

solts指派后集群中节点情况如下:

[root@VM-0-15-centos ~]# redis-cli -p 6379 -c cluster nodes
693de0512d73a16c7525e3d85e79f6da4fcf67b2 127.0.0.1:6779@16779 master - 0 1611474535000 4 connected 10001-16383
d3a3d410da3db56d33dd34df8eb8f6fd3cdbced6 127.0.0.1:6579@16579 master - 0 1611474536216 0 connected 5001-10000
c4dbef9824daa77c553e716a968b1628d78f5377 127.0.0.1:6679@16679 master - 0 1611474533000 3 connected
162be0c0e035e9dfb52765e41a33069cd021f8b0 127.0.0.1:6379@16379 myself,master - 0 1611474532000 1 connected 0-5000
92f8bc8d863e73b94b04d3f612eebf6c8b9e4834 127.0.0.1:6479@16479 master - 0 1611474535000 2 connected
1b403710ba735ad76533caa3d8f5627cf45bbda3 127.0.0.1:6879@16879 master - 0 1611474535214 5 connected

查看集群info

[root@VM-0-15-centos ~]# redis-cli -p 6379 -c cluster info
cluster_state:ok            # 集群处于上线状态
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:5
cluster_my_epoch:1
cluster_stats_messages_ping_sent:295
cluster_stats_messages_pong_sent:327
cluster_stats_messages_meet_sent:5
cluster_stats_messages_sent:627
cluster_stats_messages_ping_received:327
cluster_stats_messages_pong_received:300
cluster_stats_messages_received:627

1.4 主从复制

上述步骤后，集群节点均作为主节点存在，仍不能实现Redis的高可用，配置主从复制之后，才算真正实现了集群的高可用功能。

CLUSTER REPLICATE <node_id> 用来让集群中接收命令的节点成为 node_id 所指定节点的从节点，并开始对主节点进行复制。
命令格式：redis-cli -p <从节点端口> cluster replicate <主节点node_id>

[root@VM-0-15-centos ~]# redis-cli -p 6479 cluster replicate 162be0c0e035e9dfb52765e41a33069cd021f8b0
OK
[root@VM-0-15-centos ~]# redis-cli -p 6679 cluster replicate d3a3d410da3db56d33dd34df8eb8f6fd3cdbced6
OK
[root@VM-0-15-centos ~]# redis-cli -p 6879 cluster replicate 693de0512d73a16c7525e3d85e79f6da4fcf67b2
OK

[root@VM-0-15-centos ~]# redis-cli -p 6379 -c cluster nodes
693de0512d73a16c7525e3d85e79f6da4fcf67b2 127.0.0.1:6779@16779 master - 0 1611474855000 4 connected 10001-16383
d3a3d410da3db56d33dd34df8eb8f6fd3cdbced6 127.0.0.1:6579@16579 master - 0 1611474857101 0 connected 5001-10000
c4dbef9824daa77c553e716a968b1628d78f5377 127.0.0.1:6679@16679 slave d3a3d410da3db56d33dd34df8eb8f6fd3cdbced6 0 1611474856097 0 connected
162be0c0e035e9dfb52765e41a33069cd021f8b0 127.0.0.1:6379@16379 myself,master - 0 1611474856000 1 connected 0-5000
92f8bc8d863e73b94b04d3f612eebf6c8b9e4834 127.0.0.1:6479@16479 slave 162be0c0e035e9dfb52765e41a33069cd021f8b0 0 1611474855000 1 connected
1b403710ba735ad76533caa3d8f5627cf45bbda3 127.0.0.1:6879@16879 slave 693de0512d73a16c7525e3d85e79f6da4fcf67b2 0 1611474856000 4 connected

1.5 在集群中执行命令

集群此时处于上线状态，可以通过客户端向集群中的节点发送命令。接收命令的节点会计算出命令要处理的键属于哪个solts，并检查这个solts是否指派给自己。

如果键所在的slot刚好指派给了当前节点，会直接执行这个命令。
否则，节点向客户端返回 MOVED 错误，指引客户端转向 redirect 至正确的节点，并再次发送此前的命令。

此处，我们利用 CLUSTER KEYSLOT 查看到键 foo 所在槽号为12182（被分配在6579节点），当对此键操作时，会被重定向到相应的节点。

127.0.0.1:6379> set ddk ddv
-> Redirected to slot [7560] located at 127.0.0.1:6579
OK

值得注意的是，当我们将命令通过客户端发送给一个从节点时，命令会被重定向至对应的主节点。

127.0.0.1:6479> set aak aav
-> Redirected to slot [2445] located at 127.0.0.1:6379
OK

1.6 集群故障转移

集群中主节点下线时，复制此主节点的所有的从节点将会选出一个节点作为新的主节点，并完成故障转移。和主从复制的配置相似。当原先的从节点再次上线，需要手动将宕机启动后的节点加入集群

下面模拟6379节点宕机的情况（将其shutdown），可以观察到其从节点6679将作为新的主节点继续工作。

# 6379节点宕机前

127.0.0.1:6379> cluster nodes
693de0512d73a16c7525e3d85e79f6da4fcf67b2 127.0.0.1:6779@16779 master - 0 1611475290285 4 connected 10001-16383
d3a3d410da3db56d33dd34df8eb8f6fd3cdbced6 127.0.0.1:6579@16579 master - 0 1611475289000 0 connected 5001-10000
c4dbef9824daa77c553e716a968b1628d78f5377 127.0.0.1:6679@16679 slave d3a3d410da3db56d33dd34df8eb8f6fd3cdbced6 0 1611475287000 0 connected
162be0c0e035e9dfb52765e41a33069cd021f8b0 127.0.0.1:6379@16379 myself,master - 0 1611475288000 1 connected 0-5000
92f8bc8d863e73b94b04d3f612eebf6c8b9e4834 127.0.0.1:6479@16479 slave 162be0c0e035e9dfb52765e41a33069cd021f8b0 0 1611475287278 1 connected
1b403710ba735ad76533caa3d8f5627cf45bbda3 127.0.0.1:6879@16879 slave 693de0512d73a16c7525e3d85e79f6da4fcf67b2 0 1611475289282 4 connected

上面可以看出6379节点的id：162be0c0e035e9dfb52765e41a33069cd021f8b0
对应从节点6679的节点id：92f8bc8d863e73b94b04d3f612eebf6c8b9e4834

kill掉6379节点的服务" kill -9 PID"

[root@VM-0-15-centos src]# ps -ef | grep redis
root       58341       1  0 15:40 ?        00:00:01 redis-server 127.0.0.1:6379 [cluster]
root       58343       1  0 15:40 ?        00:00:01 redis-server 127.0.0.1:6479 [cluster]
root       58345       1  0 15:40 ?        00:00:01 redis-server 127.0.0.1:6579 [cluster]
root       58347       1  0 15:40 ?        00:00:01 redis-server 127.0.0.1:6679 [cluster]
root       58353       1  0 15:40 ?        00:00:01 redis-server 127.0.0.1:6779 [cluster]
root       58355       1  0 15:40 ?        00:00:01 redis-server 127.0.0.1:6879 [cluster]
[root@VM-0-15-centos src]# kill -9 58341
[root@VM-0-15-centos src]# ps -ef | grep redis
root       58343       1  0 15:40 ?        00:00:01 redis-server 127.0.0.1:6479 [cluster]
root       58345       1  0 15:40 ?        00:00:01 redis-server 127.0.0.1:6579 [cluster]
root       58347       1  0 15:40 ?        00:00:01 redis-server 127.0.0.1:6679 [cluster]
root       58353       1  0 15:40 ?        00:00:01 redis-server 127.0.0.1:6779 [cluster]
root       58355       1  0 15:40 ?        00:00:01 redis-server 127.0.0.1:6879 [cluster]

# 6379节点宕机后（可以看到6379节点的状态是 disconnected(断开)）

127.0.0.1:6479> cluster nodes
162be0c0e035e9dfb52765e41a33069cd021f8b0 127.0.0.1:6379@16379 master,fail - 1611475424650 1611475420000 1 disconnected
693de0512d73a16c7525e3d85e79f6da4fcf67b2 127.0.0.1:6779@16779 master - 0 1611475548000 4 connected 10001-16383
1b403710ba735ad76533caa3d8f5627cf45bbda3 127.0.0.1:6879@16879 slave 693de0512d73a16c7525e3d85e79f6da4fcf67b2 0 1611475549000 4 connected
92f8bc8d863e73b94b04d3f612eebf6c8b9e4834 127.0.0.1:6479@16479 myself,master - 0 1611475549000 6 connected 0-5000
c4dbef9824daa77c553e716a968b1628d78f5377 127.0.0.1:6679@16679 slave d3a3d410da3db56d33dd34df8eb8f6fd3cdbced6 0 1611475550098 0 connected
d3a3d410da3db56d33dd34df8eb8f6fd3cdbced6 127.0.0.1:6579@16579 master - 0 1611475549095 0 connected 5001-10000

恢复6379节点服务，再看下集群节点状态

127.0.0.1:6479> cluster nodes
162be0c0e035e9dfb52765e41a33069cd021f8b0 :0@0 master,fail,noaddr - 1611475424650 1611475420000 1 disconnected
693de0512d73a16c7525e3d85e79f6da4fcf67b2 127.0.0.1:6779@16779 master - 0 1611475808942 4 connected 10001-16383
1b403710ba735ad76533caa3d8f5627cf45bbda3 127.0.0.1:6879@16879 slave 693de0512d73a16c7525e3d85e79f6da4fcf67b2 0 1611475809944 4 connected
92f8bc8d863e73b94b04d3f612eebf6c8b9e4834 127.0.0.1:6479@16479 myself,master - 0 1611475807000 6 connected 0-5000
c4dbef9824daa77c553e716a968b1628d78f5377 127.0.0.1:6679@16679 slave d3a3d410da3db56d33dd34df8eb8f6fd3cdbced6 0 1611475807939 0 connected
d3a3d410da3db56d33dd34df8eb8f6fd3cdbced6 127.0.0.1:6579@16579 master - 0 1611475808000 0 connected 5001-10000

看到6379节点还是 disconnected 状态，可以看到6379节点服务是成功启动的

[root@VM-0-15-centos src]# redis-cli -p 6379 -c cluster nodes
b36cfffaff5d9bdd4cf06f3cb482d0acda381c19 127.0.0.1:6379@16379 myself,master - 0 0 0 connected

注意：观察到重新启动后的 node_id 和之前的不一样

6379宕机前的节点id：162be0c0e035e9dfb52765e41a33069cd021f8b0

6379宕机后的节点id：b36cfffaff5d9bdd4cf06f3cb482d0acda381c19

由此断定可能是重新启动节点后 node_id 发生改变导致集群无法识别，现在需要咱们手动将集群内失效的6379节点移除，加入新的6379节点

1.6.1 从集群移除节点

使用 forget 命令，剔除集群失效的6379节点（也就是：162be0c0e035e9dfb52765e41a33069cd021f8b0），因为是5个节点，所以需要每个节点挨个执行cluster forget命令

[root@VM-0-15-centos src]# redis-cli -c -p 6479 cluster forget 162be0c0e035e9dfb52765e41a33069cd021f8b0
OK
[root@VM-0-15-centos src]# redis-cli -c -p 6579 cluster forget 162be0c0e035e9dfb52765e41a33069cd021f8b0
OK
[root@VM-0-15-centos src]# redis-cli -c -p 6679 cluster forget 162be0c0e035e9dfb52765e41a33069cd021f8b0
OK
[root@VM-0-15-centos src]# redis-cli -c -p 6779 cluster forget 162be0c0e035e9dfb52765e41a33069cd021f8b0
OK
[root@VM-0-15-centos src]# redis-cli -c -p 6879 cluster forget 162be0c0e035e9dfb52765e41a33069cd021f8b0
OK

这里估计有小伙伴就发问了，为什么不用 cluster del-node 命令删除节点，用 forget 命令还要每个节点里都执行一次，用del-node命令可以删除节点，但是 del-node 命令的格式是：cluster del-node 《host:port》《node_id 》；这时候集群内的失效节点没有对应的host：port，所以得用 forget 命令；

可以看到失效的节点已从集群内移除

[root@VM-0-15-centos src]# redis-cli -p 6479 -c cluster nodes
693de0512d73a16c7525e3d85e79f6da4fcf67b2 127.0.0.1:6779@16779 master - 0 1611476542374 4 connected 10001-16383
1b403710ba735ad76533caa3d8f5627cf45bbda3 127.0.0.1:6879@16879 slave 693de0512d73a16c7525e3d85e79f6da4fcf67b2 0 1611476542000 4 connected
92f8bc8d863e73b94b04d3f612eebf6c8b9e4834 127.0.0.1:6479@16479 myself,master - 0 1611476543000 6 connected 0-5000
c4dbef9824daa77c553e716a968b1628d78f5377 127.0.0.1:6679@16679 slave d3a3d410da3db56d33dd34df8eb8f6fd3cdbced6 0 1611476543376 0 connected
d3a3d410da3db56d33dd34df8eb8f6fd3cdbced6 127.0.0.1:6579@16579 master - 0 1611476542000 0 connected 5001-10000

1.6.2 添加节点

命令格式：redis-cli -c -p 《集群任意节点port》 --cluster add-node 《需要加入集群的ip:port》《集群任意节点ip:port》 --cluster-slave --cluster-master-id 《master node_id》

下面命令意思是：将6379节点作为6479节点的从节点加入集群

[root@VM-0-15-centos ~]# redis-cli -c -p 6479 --cluster add-node 127.0.0.1:6379 127.0.0.1:6479 --cluster-slave --cluster-master-id 92f8bc8d863e73b94b04d3f612eebf6c8b9e4834
>>> Adding node 127.0.0.1:6379 to cluster 127.0.0.1:6479
>>> Performing Cluster Check (using node 127.0.0.1:6479)
M: 92f8bc8d863e73b94b04d3f612eebf6c8b9e4834 127.0.0.1:6479
   slots:[0-5000] (5001 slots) master
M: 693de0512d73a16c7525e3d85e79f6da4fcf67b2 127.0.0.1:6779
   slots:[10001-16383] (6383 slots) master
   1 additional replica(s)
S: 1b403710ba735ad76533caa3d8f5627cf45bbda3 127.0.0.1:6879
   slots: (0 slots) slave
   replicates 693de0512d73a16c7525e3d85e79f6da4fcf67b2
S: c4dbef9824daa77c553e716a968b1628d78f5377 127.0.0.1:6679
   slots: (0 slots) slave
   replicates d3a3d410da3db56d33dd34df8eb8f6fd3cdbced6
M: d3a3d410da3db56d33dd34df8eb8f6fd3cdbced6 127.0.0.1:6579
   slots:[5001-10000] (5000 slots) master
   1 additional replica(s)
[OK] All nodes agree about slots configuration.
>>> Check for open slots...
>>> Check slots coverage...
[OK] All 16384 slots covered.
>>> Send CLUSTER MEET to node 127.0.0.1:6379 to make it join the cluster.
Waiting for the cluster to join

>>> Configure node as replica of 127.0.0.1:6479.
[OK] New node added correctly.

可以看到6379节点已成功加入集群并作为6479的从节点

[root@VM-0-15-centos src]# redis-cli -p 6479 -c cluster nodes
693de0512d73a16c7525e3d85e79f6da4fcf67b2 127.0.0.1:6779@16779 master - 0 1611477298000 4 connected 10001-16383
b36cfffaff5d9bdd4cf06f3cb482d0acda381c19 127.0.0.1:6379@16379 slave 92f8bc8d863e73b94b04d3f612eebf6c8b9e4834 0 1611477299456 6 connected
1b403710ba735ad76533caa3d8f5627cf45bbda3 127.0.0.1:6879@16879 slave 693de0512d73a16c7525e3d85e79f6da4fcf67b2 0 1611477299000 4 connected
92f8bc8d863e73b94b04d3f612eebf6c8b9e4834 127.0.0.1:6479@16479 myself,master - 0 1611477301000 6 connected 0-5000
c4dbef9824daa77c553e716a968b1628d78f5377 127.0.0.1:6679@16679 slave d3a3d410da3db56d33dd34df8eb8f6fd3cdbced6 0 1611477301461 0 connected
d3a3d410da3db56d33dd34df8eb8f6fd3cdbced6 127.0.0.1:6579@16579 master - 0 1611477300459 0 connected 5001-10000

前文提到 cluster-config-file 会记录下集群节点的状态，打开其中一个节点6379的配置文件 nodes-6379.conf ，可以看到 CLUSTER NODES 所示信息均被保存在配置文件中：

[root@VM-0-15-centos src]# cat nodes-6379.conf 
693de0512d73a16c7525e3d85e79f6da4fcf67b2 127.0.0.1:6779@16779 master - 0 1611478049263 4 connected 10001-16383
c4dbef9824daa77c553e716a968b1628d78f5377 127.0.0.1:6679@16679 slave d3a3d410da3db56d33dd34df8eb8f6fd3cdbced6 0 1611478047000 0 connected
d3a3d410da3db56d33dd34df8eb8f6fd3cdbced6 127.0.0.1:6579@16579 master - 0 1611478048259 0 connected 5001-10000
1b403710ba735ad76533caa3d8f5627cf45bbda3 127.0.0.1:6879@16879 slave 693de0512d73a16c7525e3d85e79f6da4fcf67b2 0 1611478050268 4 connected
92f8bc8d863e73b94b04d3f612eebf6c8b9e4834 :0@0 master,fail,noaddr - 1611478026188 1611478021158 6 disconnected
b36cfffaff5d9bdd4cf06f3cb482d0acda381c19 127.0.0.1:6379@16379 myself,master - 0 1611478048000 8 connected 0-5000
vars currentEpoch 8 lastVoteEpoch 0

2、集群伸缩实践

集群伸缩的关键在于对集群的进行重新分片，实现槽位在节点间的迁移。本节将以在集群中添加节点和删除节点为例，对槽迁移进行实践。

2.1 添加节点

2.1.1 启动节点

按照 1.1 中介绍的步骤，启动7079和7179节点。

2.1.2 节点握手(加入集群)

借助 redis-cli --cluster add-node 命令分别添加节点7079（主）和7179（从）。

# 添加 7079（主）节点
[root@VM-0-6-centos ~]# redis-cli --cluster add-node 127.0.0.1:7079 127.0.0.1:6379
>>> Adding node 127.0.0.1:7079 to cluster 127.0.0.1:6379
>>> Performing Cluster Check (using node 127.0.0.1:6379)
M: 4830b60b806cc944550e564ef807192f00d03620 127.0.0.1:6379
   slots:[0-5000] (5001 slots) master
   1 additional replica(s)
S: aa97f087a96af31ad95768c693a3152019ee4557 127.0.0.1:6879
   slots: (0 slots) slave
   replicates 1d24fcbec775f2c7aff2bc09c29879963700e9af
M: 1d24fcbec775f2c7aff2bc09c29879963700e9af 127.0.0.1:6779
   slots:[10001-16383] (6383 slots) master
   1 additional replica(s)
M: 171c8b624e045dbc8adf517568e602948a0f2f61 127.0.0.1:6579
   slots:[5001-10000] (5000 slots) master
   1 additional replica(s)
S: a3d763ab71d8aa0fc10edc0ac9f4520b82bc6a3d 127.0.0.1:6679
   slots: (0 slots) slave
   replicates 171c8b624e045dbc8adf517568e602948a0f2f61
S: 02777dacff2e2c2c7dabe750ec80baf3333499a7 127.0.0.1:6479
   slots: (0 slots) slave
   replicates 4830b60b806cc944550e564ef807192f00d03620
[OK] All nodes agree about slots configuration.
>>> Check for open slots...
>>> Check slots coverage...
[OK] All 16384 slots covered.
>>> Send CLUSTER MEET to node 127.0.0.1:7079 to make it join the cluster.
[OK] New node added correctly.

可以看到7079节点已添加到集群，node_id:7a06d29c98d6eaaeb39129c79259267adceb65bd

127.0.0.1:6379> CLUSTER NODES
171c8b624e045dbc8adf517568e602948a0f2f61 127.0.0.1:6579@16579 myself,master - 0 1612870545000 2 connected 5001-10000
aa97f087a96af31ad95768c693a3152019ee4557 127.0.0.1:6879@16879 slave 1d24fcbec775f2c7aff2bc09c29879963700e9af 0 1612870544495 4 connected
a3d763ab71d8aa0fc10edc0ac9f4520b82bc6a3d 127.0.0.1:6679@16679 slave 171c8b624e045dbc8adf517568e602948a0f2f61 0 1612870545496 2 connected
7a06d29c98d6eaaeb39129c79259267adceb65bd 127.0.0.1:7079@17079 master - 0 1612870544000 0 connected
1d24fcbec775f2c7aff2bc09c29879963700e9af 127.0.0.1:6779@16779 master - 0 1612870540000 4 connected 10001-16383
4830b60b806cc944550e564ef807192f00d03620 127.0.0.1:6379@16379 master - 0 1612870546498 3 connected 0-5000
02777dacff2e2c2c7dabe750ec80baf3333499a7 127.0.0.1:6479@16479 slave 4830b60b806cc944550e564ef807192f00d03620 0 1612870542000 3 connected

# 添加 7179（从）节点，并指定7079节点为其主节点

[root@VM-0-6-centos ~]# redis-cli --cluster add-node 127.0.0.1:7179 127.0.0.1:6379 --cluster-slave --cluster-master-id 7a06d29c98d6eaaeb39129c79259267adceb65bd
>>> Adding node 127.0.0.1:7179 to cluster 127.0.0.1:6379
>>> Performing Cluster Check (using node 127.0.0.1:6379)
M: 4830b60b806cc944550e564ef807192f00d03620 127.0.0.1:6379
   slots:[0-5000] (5001 slots) master
   1 additional replica(s)
M: 7a06d29c98d6eaaeb39129c79259267adceb65bd 127.0.0.1:7079
   slots: (0 slots) master
S: aa97f087a96af31ad95768c693a3152019ee4557 127.0.0.1:6879
   slots: (0 slots) slave
   replicates 1d24fcbec775f2c7aff2bc09c29879963700e9af
M: 1d24fcbec775f2c7aff2bc09c29879963700e9af 127.0.0.1:6779
   slots:[10001-16383] (6383 slots) master
   1 additional replica(s)
M: 171c8b624e045dbc8adf517568e602948a0f2f61 127.0.0.1:6579
   slots:[5001-10000] (5000 slots) master
   1 additional replica(s)
S: a3d763ab71d8aa0fc10edc0ac9f4520b82bc6a3d 127.0.0.1:6679
   slots: (0 slots) slave
   replicates 171c8b624e045dbc8adf517568e602948a0f2f61
S: 02777dacff2e2c2c7dabe750ec80baf3333499a7 127.0.0.1:6479
   slots: (0 slots) slave
   replicates 4830b60b806cc944550e564ef807192f00d03620
[OK] All nodes agree about slots configuration.
>>> Check for open slots...
>>> Check slots coverage...
[OK] All 16384 slots covered.
>>> Send CLUSTER MEET to node 127.0.0.1:7179 to make it join the cluster.
Waiting for the cluster to join

>>> Configure node as replica of 127.0.0.1:7079.
[OK] New node added correctly.

可以看到7179节点已添加到集群,并是7079的从节点

127.0.0.1:6379> CLUSTER NODES
171c8b624e045dbc8adf517568e602948a0f2f61 127.0.0.1:6579@16579 myself,master - 0 1612870706000 2 connected 5001-10000
aa97f087a96af31ad95768c693a3152019ee4557 127.0.0.1:6879@16879 slave 1d24fcbec775f2c7aff2bc09c29879963700e9af 0 1612870708770 4 connected
a3d763ab71d8aa0fc10edc0ac9f4520b82bc6a3d 127.0.0.1:6679@16679 slave 171c8b624e045dbc8adf517568e602948a0f2f61 0 1612870707768 2 connected
43b20254222d025f7ec7413e7b9559c9222afd88 127.0.0.1:7179@17179 slave 7a06d29c98d6eaaeb39129c79259267adceb65bd 0 1612870707000 0 connected
7a06d29c98d6eaaeb39129c79259267adceb65bd 127.0.0.1:7079@17079 master - 0 1612870707000 0 connected
1d24fcbec775f2c7aff2bc09c29879963700e9af 127.0.0.1:6779@16779 master - 0 1612870704000 4 connected 10001-16383
4830b60b806cc944550e564ef807192f00d03620 127.0.0.1:6379@16379 master - 0 1612870706767 3 connected 0-5000
02777dacff2e2c2c7dabe750ec80baf3333499a7 127.0.0.1:6479@16479 slave 4830b60b806cc944550e564ef807192f00d03620 0 1612870705765 3 connected

2.2 重新分片

借助 redis-cli --cluster reshard 命令对集群重新分片，使得各节点槽位均衡（从节点6779中迁移一些slot到节点7079中）。需要指定：

移动的槽位数：2776
接收槽位的目标节点ID：节点7079的node_id：7a06d29c98d6eaaeb39129c79259267adceb65bd
移出槽位的源节点ID：节点6779的node_id：1d24fcbec775f2c7aff2bc09c29879963700e9af

命令格式：

redis-cli --cluster reshard 《 ip:port(集群任意节点) 》 --cluster-from 《移出槽位的源节点ID，多个用逗号分隔》 --cluster-to 《接收槽位的目标节点ID》 --cluster-slots 《移动的槽位数》 --cluster-yes

[root@VM-0-6-centos ~]# redis-cli --cluster reshard 127.0.0.1:6379 --cluster-from 1d24fcbec775f2c7aff2bc09c29879963700e9af --cluster-to 7a06d29c98d6eaaeb39129c79259267adceb65bd --cluster-slots 2776 --cluster-yes 
>>> Performing Cluster Check (using node 127.0.0.1:6379)
M: 4830b60b806cc944550e564ef807192f00d03620 127.0.0.1:6379
   slots:[0-5000] (5001 slots) master
   1 additional replica(s)
M: 7a06d29c98d6eaaeb39129c79259267adceb65bd 127.0.0.1:7079
   slots:[10001-12766] (2766 slots) master
   1 additional replica(s)
S: aa97f087a96af31ad95768c693a3152019ee4557 127.0.0.1:6879
   slots: (0 slots) slave
   replicates 1d24fcbec775f2c7aff2bc09c29879963700e9af
M: 1d24fcbec775f2c7aff2bc09c29879963700e9af 127.0.0.1:6779
   slots:[12767-16383] (3617 slots) master
   1 additional replica(s)
M: 171c8b624e045dbc8adf517568e602948a0f2f61 127.0.0.1:6579
   slots:[5001-10000] (5000 slots) master
   1 additional replica(s)
S: a3d763ab71d8aa0fc10edc0ac9f4520b82bc6a3d 127.0.0.1:6679
   slots: (0 slots) slave
   replicates 171c8b624e045dbc8adf517568e602948a0f2f61
S: 43b20254222d025f7ec7413e7b9559c9222afd88 127.0.0.1:7179
   slots: (0 slots) slave
   replicates 7a06d29c98d6eaaeb39129c79259267adceb65bd
S: 02777dacff2e2c2c7dabe750ec80baf3333499a7 127.0.0.1:6479
   slots: (0 slots) slave
   replicates 4830b60b806cc944550e564ef807192f00d03620
[OK] All nodes agree about slots configuration.
>>> Check for open slots...
>>> Check slots coverage...
[OK] All 16384 slots covered.

Ready to move 10 slots.
  Source nodes:
    M: 1d24fcbec775f2c7aff2bc09c29879963700e9af 127.0.0.1:6779
       slots:[12767-16383] (3617 slots) master
       1 additional replica(s)
  Destination node:
    M: 7a06d29c98d6eaaeb39129c79259267adceb65bd 127.0.0.1:7079
       slots:[10001-12766] (2766 slots) master
       1 additional replica(s)
  Resharding plan:
    Moving slot 12767 from 1d24fcbec775f2c7aff2bc09c29879963700e9af
    ......
    ......
    Moving slot 12776 from 1d24fcbec775f2c7aff2bc09c29879963700e9af
Moving slot 12767 from 127.0.0.1:6779 to 127.0.0.1:7079: 
    ......
    ......
Moving slot 12776 from 127.0.0.1:6779 to 127.0.0.1:7079:

查看7079节点的solt数量

127.0.0.1:6579> CLUSTER NODES
171c8b624e045dbc8adf517568e602948a0f2f61 127.0.0.1:6579@16579 myself,master - 0 1612871458000 2 connected 5001-10000
aa97f087a96af31ad95768c693a3152019ee4557 127.0.0.1:6879@16879 slave 1d24fcbec775f2c7aff2bc09c29879963700e9af 0 1612871459000 4 connected
a3d763ab71d8aa0fc10edc0ac9f4520b82bc6a3d 127.0.0.1:6679@16679 slave 171c8b624e045dbc8adf517568e602948a0f2f61 0 1612871460976 2 connected
43b20254222d025f7ec7413e7b9559c9222afd88 127.0.0.1:7179@17179 slave 7a06d29c98d6eaaeb39129c79259267adceb65bd 0 1612871455969 7 connected
7a06d29c98d6eaaeb39129c79259267adceb65bd 127.0.0.1:7079@17079 master - 0 1612871460000 7 connected 10001-12776
1d24fcbec775f2c7aff2bc09c29879963700e9af 127.0.0.1:6779@16779 master - 0 1612871458000 4 connected 12777-16383
4830b60b806cc944550e564ef807192f00d03620 127.0.0.1:6379@16379 master - 0 1612871458973 3 connected 0-5000
02777dacff2e2c2c7dabe750ec80baf3333499a7 127.0.0.1:6479@16479 slave 4830b60b806cc944550e564ef807192f00d03620 0 1612871457972 3 connected

2.3 移除节点

主节点要先把solts迁移走（执行 2.2 的步骤将主节点的solt移走）再删除，从节点可以直接删除，利用 redis-cli --cluster del-node 命令依次删除从节点7079和主节点7179

命令格式

redis-cli --cluster del-node 《 ip:port(集群任意节点) 》《要移除的节点id》

# 将7079主节点的2776个solts分给6779主节点
[root@VM-0-6-centos ~]# redis-cli --cluster reshard 127.0.0.1:6379 --cluster-from 7a06d29c98d6eaaeb39129c79259267adceb65bd --cluster-to 1d24fcbec775f2c7aff2bc09c29879963700e9af --cluster-slots 2776 --cluster-yes
......
......
# 可以看到7079主节点已没有solts
127.0.0.1:6379> CLUSTER NODES
171c8b624e045dbc8adf517568e602948a0f2f61 127.0.0.1:6579@16579 myself,master - 0 1612871704000 2 connected 5001-10000
aa97f087a96af31ad95768c693a3152019ee4557 127.0.0.1:6879@16879 slave 1d24fcbec775f2c7aff2bc09c29879963700e9af 0 1612871704385 8 connected
a3d763ab71d8aa0fc10edc0ac9f4520b82bc6a3d 127.0.0.1:6679@16679 slave 171c8b624e045dbc8adf517568e602948a0f2f61 0 1612871702000 2 connected
43b20254222d025f7ec7413e7b9559c9222afd88 127.0.0.1:7179@17179 slave 1d24fcbec775f2c7aff2bc09c29879963700e9af 0 1612871705388 8 connected
7a06d29c98d6eaaeb39129c79259267adceb65bd 127.0.0.1:7079@17079 master - 0 1612871702000 7 connected
1d24fcbec775f2c7aff2bc09c29879963700e9af 127.0.0.1:6779@16779 master - 0 1612871705000 8 connected 10001-16383
4830b60b806cc944550e564ef807192f00d03620 127.0.0.1:6379@16379 master - 0 1612871703382 3 connected 0-5000
02777dacff2e2c2c7dabe750ec80baf3333499a7 127.0.0.1:6479@16479 slave 4830b60b806cc944550e564ef807192f00d03620 0 1612871706394 3 connected

# 删除7079和7179节点
[root@VM-0-6-centos ~]# redis-cli --cluster del-node 127.0.0.1:6379 7a06d29c98d6eaaeb39129c79259267adceb65bd
>>> Removing node 7a06d29c98d6eaaeb39129c79259267adceb65bd from cluster 127.0.0.1:6379
>>> Sending CLUSTER FORGET messages to the cluster...
>>> Sending CLUSTER RESET SOFT to the deleted node.
[root@VM-0-6-centos ~]# redis-cli --cluster del-node 127.0.0.1:6379 43b20254222d025f7ec7413e7b9559c9222afd88
>>> Removing node 43b20254222d025f7ec7413e7b9559c9222afd88 from cluster 127.0.0.1:6379
>>> Sending CLUSTER FORGET messages to the cluster...
>>> Sending CLUSTER RESET SOFT to the deleted node.

# 查看当前节点，可以看到7079和7179节点已从节点移除
127.0.0.1:6379> CLUSTER NODES
171c8b624e045dbc8adf517568e602948a0f2f61 127.0.0.1:6579@16579 myself,master - 0 1612872200000 2 connected 5001-10000
aa97f087a96af31ad95768c693a3152019ee4557 127.0.0.1:6879@16879 slave 1d24fcbec775f2c7aff2bc09c29879963700e9af 0 1612872198000 8 connected
a3d763ab71d8aa0fc10edc0ac9f4520b82bc6a3d 127.0.0.1:6679@16679 slave 171c8b624e045dbc8adf517568e602948a0f2f61 0 1612872198165 2 connected
1d24fcbec775f2c7aff2bc09c29879963700e9af 127.0.0.1:6779@16779 master - 0 1612872201000 8 connected 10001-16383
4830b60b806cc944550e564ef807192f00d03620 127.0.0.1:6379@16379 master - 0 1612872201171 3 connected 0-5000
02777dacff2e2c2c7dabe750ec80baf3333499a7 127.0.0.1:6479@16479 slave 4830b60b806cc944550e564ef807192f00d03620 0 1612872199166 3 connected

2.4 设置主从关系

一个主节点可以有多个从节点，下面加入7279节点作为6379主节点的第二个从节点（一般第二个从节点用作数据备份）

命令格式

redis-cli -p 《要添加的从节点端口》 cluster replicate 《对应主节点的node_id》

# 添加7179节点到集群
[root@VM-0-6-centos ~]# redis-cli --cluster add-node 127.0.0.1:7279 127.0.0.1:6379

# 设置主从关系
[root@VM-0-6-centos ~]# redis-cli -p 7279 cluster replicate 4830b60b806cc944550e564ef807192f00d03620
OK

# 查看当前集群主从关系，可以看到6379节点有7179和7279两个从节点
127.0.0.1:6579> CLUSTER NODES
171c8b624e045dbc8adf517568e602948a0f2f61 127.0.0.1:6579@16579 myself,master - 0 1612874189000 2 connected 6725-10000
aa97f087a96af31ad95768c693a3152019ee4557 127.0.0.1:6879@16879 slave 1d24fcbec775f2c7aff2bc09c29879963700e9af 0 1612874189000 10 connected
4830b60b806cc944550e564ef807192f00d03620 127.0.0.1:6379@16379 master - 0 1612874189098 3 connected 3363-5000
02777dacff2e2c2c7dabe750ec80baf3333499a7 127.0.0.1:6479@16479 slave 7a06d29c98d6eaaeb39129c79259267adceb65bd 0 1612874192103 9 connected
52eb187093695b5939630813e46e59eb48ec7489 127.0.0.1:7279@17279 slave 4830b60b806cc944550e564ef807192f00d03620 0 1612874190000 3 connected
a3d763ab71d8aa0fc10edc0ac9f4520b82bc6a3d 127.0.0.1:6679@16679 slave 171c8b624e045dbc8adf517568e602948a0f2f61 0 1612874188097 2 connected
1d24fcbec775f2c7aff2bc09c29879963700e9af 127.0.0.1:6779@16779 master - 0 1612874190100 10 connected 5905-6724 12288-16383
7a06d29c98d6eaaeb39129c79259267adceb65bd 127.0.0.1:7079@17079 master - 0 1612874191101 9 connected 0-3362 5001-5904 10001-12287
43b20254222d025f7ec7413e7b9559c9222afd88 127.0.0.1:7179@17179 slave 4830b60b806cc944550e564ef807192f00d03620 0 1612874188000 3 connected

2.5 均衡各主节点的solts

可以看到集群里各主节点的solts不均衡，这里新加一个主节点分摊其它3个主节点的solts，使其每个节点的solts都一样

[root@VM-0-6-centos ~]# redis-cli --cluster check 127.0.0.1:6379
127.0.0.1:6379 (4830b60b...) -> 0 keys | 5001 slots | 1 slaves.
127.0.0.1:6779 (1d24fcbe...) -> 0 keys | 6383 slots | 1 slaves.
127.0.0.1:6579 (171c8b62...) -> 1 keys | 5000 slots | 1 slaves.

添加7079主节点，使用--cluster rebalance命令进行平衡集群节点slot数量

命令格式

redis-cli --cluster rebalance 《 ip:port(7079节点) 》 --cluster-use-empty-masters

# 添加7079节点到集群
[root@VM-0-6-centos ~]# redis-cli --cluster add-node 127.0.0.1:7079 127.0.0.1:6379

# 让7079节点均衡其他节点的solts
[root@VM-0-6-centos ~]# redis-cli --cluster rebalance 127.0.0.1:7079 --cluster-use-empty-masters
>>> Performing Cluster Check (using node 127.0.0.1:7079)
[OK] All nodes agree about slots configuration.
>>> Check for open slots...
>>> Check slots coverage...
[OK] All 16384 slots covered.
>>> Rebalancing across 4 nodes. Total weight = 4.00
Moving 2287 slots from 127.0.0.1:6779 to 127.0.0.1:7079
###############......###############

# 查看当前集群每个主节点的solts，可以看到每个主节点solts都是一样的
[root@VM-0-6-centos ~]# redis-cli --cluster check 127.0.0.1:6379
127.0.0.1:6379 (4830b60b...) -> 0 keys | 4096 slots | 1 slaves.
127.0.0.1:7079 (7a06d29c...) -> 0 keys | 4096 slots | 0 slaves.
127.0.0.1:6779 (1d24fcbe...) -> 0 keys | 4096 slots | 1 slaves.
127.0.0.1:6579 (171c8b62...) -> 1 keys | 4096 slots | 1 slaves.

2.6 给各主节点设置权重

各节点的机器配置不同，因此需要对配置低的节点权重设置的小一些，减少机器负载，使用--cluster rebalance --cluster-weight命令给各主节点设置权重；

命令格式

redis-cli --cluster rebalance 《ip:port(集群内任意节点都可以)》 --cluster-weight 《node_id=N node_id=N node_id=N 》--cluster-simulate(模拟rebalance操作，不会真正执行迁移操作：可选)

[root@VM-0-6-centos ~]# redis-cli --cluster rebalance 127.0.0.1:6379 --cluster-weight 4830b60b806cc944550e564ef807192f00d03620=1 171c8b624e045dbc8adf517568e602948a0f2f61=2 1d24fcbec775f2c7aff2bc09c29879963700e9af=3 7a06d29c98d6eaaeb39129c79259267adceb65bd=4
>>> Performing Cluster Check (using node 127.0.0.1:6379)
[OK] All nodes agree about slots configuration.
>>> Check for open slots...
>>> Check slots coverage...
[OK] All 16384 slots covered.
>>> Rebalancing across 4 nodes. Total weight = 10.00
Moving 2458 slots from 127.0.0.1:6379 to 127.0.0.1:7079
###############......###############

# 检查当前集群solts的分布情况，可以看到设置的权重越大，分配的solts越多；
[root@VM-0-6-centos ~]# redis-cli --cluster check 127.0.0.1:6379
127.0.0.1:6379 (4830b60b...) -> 0 keys | 1638 slots | 1 slaves.
127.0.0.1:7079 (7a06d29c...) -> 0 keys | 6554 slots | 0 slaves.
127.0.0.1:6779 (1d24fcbe...) -> 0 keys | 4916 slots | 1 slaves.
127.0.0.1:6579 (171c8b62...) -> 1 keys | 3276 slots | 1 slaves.
[OK] 1 keys in 4 masters.

3、集群常用命令

[root@VM-0-3-centos ~]# redis-cli --cluster help
Cluster Manager Commands:
  create         host1:port1 ... hostN:portN			# 创建集群
                 --cluster-replicas <arg>				# 从节点个数
  check          host:port								# 检查集群
                 --cluster-search-multiple-owners		# 检查是否有槽同时被分配给了多个节点
  info           host:port								# 查看集群状态
  fix            host:port								# 修复集群
                 --cluster-search-multiple-owners		# 修复槽的重复分配问题
                 --cluster-fix-with-unreachable-masters
  reshard        host:port								# 指定集群的任意一节点进行迁移slot，重新分slots
                 --cluster-from <arg>					# 需要从哪些源节点上迁移slot，可从多个源节点完成迁移，以逗号隔开，传递的是节点的node                                                           id，还可以直接传递--from all，这样源节点就是集群的所有节点，不传递该参数的话，则会在                                                           迁移过程中提示用户输入
                 --cluster-to <arg>						# slot需要迁移的目的节点的node id，目的节点只能填写一个，不传递该参数的话，则会在迁移过                                                         程中提示用户输入
                 --cluster-slots <arg>					# 需要迁移的slot数量，不传递该参数的话，则会在迁移过程中提示用户输入。
                 --cluster-yes							# 指定迁移时的确认输入
                 --cluster-timeout <arg>				# 设置migrate命令的超时时间
                 --cluster-pipeline <arg>				# 设置migrate命令的超时时间
                 --cluster-replace						# 是否直接replace到目标节点
  rebalance      host:port								# 指定集群的任意一节点进行平衡集群节点slot数量 
                 --cluster-weight <node1=w1...nodeN=wN>	# 指定集群节点的权重
                 --cluster-use-empty-masters			# 设置可以让没有分配slot的主节点参与，默认不允许
                 --cluster-timeout <arg>				# 设置migrate命令的超时时间
                 --cluster-simulate						# 模拟rebalance操作，不会真正执行迁移操作
                 --cluster-pipeline <arg>				# 定义cluster getkeysinslot命令一次取出的key数量，默认值为10
                 --cluster-threshold <arg>				# 迁移的slot阈值超过threshold，执行rebalance操作
                 --cluster-replace						# 是否直接replace到目标节点
  add-node       new_host:new_port existing_host:existing_port	# 添加节点，把新节点加入到指定的集群，默认添加主节点
                 --cluster-slave						# 新节点作为从节点，默认随机一个主节点
                 --cluster-master-id <arg>				# 给新节点指定主节点
  del-node       host:port node_id						# 删除给定的一个节点，成功后关闭该节点服务
  call           host:port command arg arg .. arg		# 在集群的所有节点执行相关命令
  set-timeout    host:port milliseconds					# 设置cluster-node-timeout
  import         host:port								# 将外部redis数据导入集群
                 --cluster-from <arg>					# 将指定实例的数据导入到集群
                 --cluster-copy							# migrate时指定copy
                 --cluster-replace						# migrate时指定replace
  backup         host:port backup_directory
  help           

For check, fix, reshard, del-node, set-timeout you can specify the host and port of any working node in the cluster.	#对于check、fix、reshard、del node、set timeout，您可以指定集群中任何工作节点的主机和端口。

原创声明：本文系作者授权腾讯云开发者社区发表，未经许可，不得转载。

如有侵权，请联系 cloudcommunity@tencent.com 删除。

云数据库 Redis

原创声明：本文系作者授权腾讯云开发者社区发表，未经许可，不得转载。

如有侵权，请联系 cloudcommunity@tencent.com 删除。

云数据库 Redis