如何用docker部署redis cluster的方法

前言

由于本人是个docker控，不喜欢安装各种环境，而且安装redis-trib也有点繁琐，索性用docker来做redis cluster。

本文用的是伪集群，真正的集群放到不同的机器即可。端口是7001-7006。

工作目录： /data/redis

创建文件夹

首先创建一堆对应端口的文件夹，下面是脚本

create.sh

for i in `seq 7001 7006`
do
mkdir -p ${i}/data
done

添加执行权限并执行

chmod 777 create.sh
./create.sh

部署redis

本人不喜欢docker run，所以用了docker-compose。官方文档 https://docs.docker.com/compose/overview/

创建docker-compose.yml

用publicisworldwide/redis-cluster镜像的原因是人家已经把配置文件写好了，配置文件没有挂载是懒，别学我

docker-compose.yml

version: '3.4'

x-image:
&default-image
publicisworldwide/redis-cluster
x-restart:
&default-restart
always
x-netmode:
&default-netmode
host

services:
redis1:
image: *default-image
network_mode: *default-netmode
restart: *default-restart
volumes:
- /data/redis/7001/data:/data
environment:
- REDIS_PORT=7001

redis2:
image: *default-image
network_mode: *default-netmode
restart: *default-restart
volumes:
- /data/redis/7002/data:/data
environment:
- REDIS_PORT=7002

redis3:
image: *default-image
network_mode: *default-netmode
restart: *default-restart
volumes:
- /data/redis/7003/data:/data
environment:
- REDIS_PORT=7003

redis4:
image: *default-image
network_mode: *default-netmode
restart: *default-restart
volumes:
- /data/redis/7004/data:/data
environment:
- REDIS_PORT=7004

redis5:
image: *default-image
network_mode: *default-netmode
restart: *default-restart
volumes:
- /data/redis/7005/data:/data
environment:
- REDIS_PORT=7005

redis6:
image: *default-image
network_mode: *default-netmode
restart: *default-restart
volumes:
- /data/redis/7006/data:/data
environment:
- REDIS_PORT=7006

启动所有redis

docker-compose up -d

如果报错，那就是版本问题。

查看docker版本

docker -v

在文档中可查看docker版本支持的docker-compose.yml版本，为了方便大家查看，我复制出来了。不过一般来说，docker升级比较快，功能迭代也很快，最好还是用最新版本。

部署cluster

运行以下命令(inem0o/redis-trib没有pull会自动pull)

注意：加上-it，不然后续的确认没法继续

docker run --rm -it inem0o/redis-trib create --replicas 1 192.168.30.70:7001 192.168.30.70:7002 192.168.30.70:7003 192.168.30.70:7004 192.168.30.70:7005 192.168.30.70:7006

会出现

>>> Creating cluster
>>> Performing hash slots allocation on 6 nodes...
Using 3 masters:
192.168.30.70:7001
192.168.30.70:7002
192.168.30.70:7003
Adding replica 192.168.30.70:7004 to 192.168.30.70:7001
Adding replica 192.168.30.70:7005 to 192.168.30.70:7002
Adding replica 192.168.30.70:7006 to 192.168.30.70:7003
M: 5a7bd7698b1fe55beb44faac051d66c8a03fd1b1 192.168.30.70:7001
 slots:0-5460 (5461 slots) master
M: bb8fda08e1dcd39e937443f81b5458e80f52d804 192.168.30.70:7002
 slots:5461-10922 (5462 slots) master
M: d907530ee9f6356e0e61a6c7f4d0cc1b22da1189 192.168.30.70:7003
 slots:10923-16383 (5461 slots) master
S: 52eee69afa751d71c84d5436d14b0e16a37536fa 192.168.30.70:7004
 replicates 5a7bd7698b1fe55beb44faac051d66c8a03fd1b1
S: 701ed2fbb3df9fc63b083818620f5c020d05e323 192.168.30.70:7005
 replicates bb8fda08e1dcd39e937443f81b5458e80f52d804
S: a3548a9dffa225f05786ea2289db65f5f1c623be 192.168.30.70:7006
 replicates d907530ee9f6356e0e61a6c7f4d0cc1b22da1189
Can I set the above configuration? (type 'yes' to accept):

输入yes

Waiting for the cluster to join.....
>>> Performing Cluster Check (using node 192.168.30.70:7001)
M: 5a7bd7698b1fe55beb44faac051d66c8a03fd1b1 192.168.30.70:7001
 slots:0-5460 (5461 slots) master
 1 additional replica(s)
M: d907530ee9f6356e0e61a6c7f4d0cc1b22da1189 192.168.30.70:7003@17003
 slots:10923-16383 (5461 slots) master
 1 additional replica(s)
S: a3548a9dffa225f05786ea2289db65f5f1c623be 192.168.30.70:7006@17006
 slots: (0 slots) slave
 replicates d907530ee9f6356e0e61a6c7f4d0cc1b22da1189
S: 701ed2fbb3df9fc63b083818620f5c020d05e323 192.168.30.70:7005@17005
 slots: (0 slots) slave
 replicates bb8fda08e1dcd39e937443f81b5458e80f52d804
S: 52eee69afa751d71c84d5436d14b0e16a37536fa 192.168.30.70:7004@17004
 slots: (0 slots) slave
 replicates 5a7bd7698b1fe55beb44faac051d66c8a03fd1b1
M: bb8fda08e1dcd39e937443f81b5458e80f52d804 192.168.30.70:7002@17002
 slots:5461-10922 (5462 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.

完成。

遇到的问题

创建集群的时候会遇到Waiting for the cluster to join....................一直没有成功

答：网络模式改成host，docker run 的方式加上--net host，docker-compose方式加上network_mode: host。

我原本是端口映射 7001:7000+17001:17000，但是不行，不知道什么原因。

来源：http://www.cnblogs.com/chenchuxin/p/8404699.html