本例单机集群部署使用三个节点，分别为node1、node2、node3

一、安装与配置

1 创建数据目录

mkdir -p node1/data  node1/logs
mkdir -p node2/data  node2/logs
mkdir -p mode3/data  node3/logs

2 数据目录授权

chown 1000 node1/ node2/ node3/
chown 1000 node1/data/ node1/logs/
chown 1000 node2/data/ node2/logs/
chown 1000 node3/data/ node3/logs/

3 创建docker-compose.yml文件

version: '3'
services:emqx1:image: emqx:5.3.1container_name: emqx1environment:- "EMQX_NODE_NAME=emqx@node1.emqx.io"- "EMQX_CLUSTER__DISCOVERY_STRATEGY=static"- "EMQX_CLUSTER__STATIC__SEEDS=[emqx@node1.emqx.io,emqx@node2.emqx.io,emqx@node3.emqx.io]"healthcheck:test: ["CMD", "/opt/emqx/bin/emqx ctl", "status"]interval: 5stimeout: 25sretries: 5networks:emqx-bridge:aliases:- node1.emqx.ioports:- 1883:1883- 8083:8083- 8084:8084- 8883:8883- 18083:18083volumes:- /etc/localtime:/etc/localtime:ro- ./node1/logs:/opt/emqx/log- ./node1/data:/opt/emqx/dataemqx2:image: emqx:5.3.1container_name: emqx2environment:- "EMQX_NODE_NAME=emqx@node2.emqx.io"- "EMQX_CLUSTER__DISCOVERY_STRATEGY=static"- "EMQX_CLUSTER__STATIC__SEEDS=[emqx@node1.emqx.io,emqx@node2.emqx.io,emqx@node3.emqx.io]"healthcheck:test: ["CMD", "/opt/emqx/bin/emqx ctl", "status"]interval: 5stimeout: 25sretries: 5networks:emqx-bridge:aliases:- node2.emqx.ioports:- 1873:1883- 8073:8083- 8074:8084- 8873:8883- 18073:18083volumes:- /etc/localtime:/etc/localtime:ro- ./node2/logs:/opt/emqx/log- ./node2/data:/opt/emqx/dataemqx3:image: emqx:5.3.1container_name: emqx3environment:- "EMQX_NODE_NAME=emqx@node3.emqx.io"- "EMQX_CLUSTER__DISCOVERY_STRATEGY=static"- "EMQX_CLUSTER__STATIC__SEEDS=[emqx@node1.emqx.io,emqx@node2.emqx.io,emqx@node3.emqx.io]"healthcheck:test: ["CMD", "/opt/emqx/bin/emqx ctl", "status"]interval: 5stimeout: 25sretries: 5networks:emqx-bridge:aliases:- node3.emqx.ioports:- 1863:1883- 8063:8083- 8064:8084- 8863:8883- 18063:18083volumes:- /etc/localtime:/etc/localtime:ro- ./node3/logs:/opt/emqx/log- ./node3/data:/opt/emqx/datanetworks:emqx-bridge:driver: bridge

4 启动运行EMQX

docker-compose up -d

5 EMQX的Dashboard访问地址 http://${ip}:${port}   参考：http://192.168.1.xxx:18083/

默认用户名： admin

默认密码： public

登录后提示修改密码

6 配置Nginx的负载均衡策略

nginx.conf内容如下：

user  nginx;
worker_processes  auto;error_log  /var/log/nginx/error.log warn;
pid        /var/run/nginx.pid;events {worker_connections  1024;
}stream {upstream emqx_tcp_cluster {server 10.10.1.100:1883 weight=1 max_fails=3 fail_timeout=30s;server 10.10.1.110:1873 weight=1 max_fails=3 fail_timeout=30s;server 10.10.1.120:1863 weight=1 max_fails=3 fail_timeout=30s;}server {listen 1893;proxy_pass emqx_tcp_cluster;proxy_buffer_size 8k;tcp_nodelay on;}
}

7 扩展其他负载均衡策略

随机轮询

随机轮询就是从待选列表中随机分配连接

upstream emqx_tcp_cluster {random;# emqx的三个实例server 10.10.1.100:1883;server 10.10.1.110:1873;server 10.10.1.120:1863;
}

带权轮询

部署的服务器存在性能差异，我们可以通过配置权重 weight 来修改轮询的几率

upstream emqx_tcp_cluster {# emqx的三个实例server 10.10.1.100:1883 weight=1;server 10.10.1.110:1873 weight=2;server 10.10.1.120:1863 weight=3;
}

least_conn 策略

最小连接数策略，优先分配给获得连接数较少的服务器

upstream emqx_tcp_cluster {least_conn;# emqx的三个实例server 10.10.1.100:1883;server 10.10.1.110:1873;server 10.10.1.120:1863;
}

IP Hash 策略

有特殊需求，希望同一个客户端连接至同一个服务器，则可以使用 IP Hash 策略

upstream emqx_tcp_cluster {ip_hash;# emqx的三个实例server 10.10.1.100:1883;server 10.10.1.110:1873;server 10.10.1.120:1863;
}