概述

高可用架构是 Keepalived + HAproxy,用 HAproxy 来做 RabbitMQ 负载均衡和高可用,用 Keepalived 来保证 HAproxy 的高可用。

之前安装过RabbitMQ。

也安装过单节点的HAProxy,IP地址为:192.168.0.104。

现在,需要把单节点的HAProxy克隆一份,形成两个HAProxy,新克隆出来的HAProxy节点的IP地址为:192.168.0.105。

安装KeepAlived

在两个节点使用yum安装keepalived:

yum install -y keepalived

在两个节点的/etc/keepalived/文件夹下编写检测haproxy的shell脚本:

#!/bin/bash

case "$1" in
    master)
        notify master
        service haproxy start
        exit 0
    ;;
    backup)
        notify backup
        service haproxy stop
        exit 0
    ;;
    fault)
        notify fault
        service haproxy stop
        exit 0
    ;;
    *)
        echo 'Usage: `basename $0` {master|backup|fault}'
        exit 1
    ;;
esac

编辑keepalived.cfg文件:

vi /etc/keepalived/keepalived.conf

修改192.168.0.104节点的配置文件:

vrrp_script chk_haproxy {
    script "service haproxy status" # 服务探测,返回0说明服务是正常的
    interval 1                      # 每隔1秒探测一次
    weight -2                       # 不正常时,权重-1,即haproxy上线,权重加2;下线,权重减2
}

vrrp_instance haproxy {
    state MASTER            # 主机为MASTER,备机为BACKUP
    interface ens33         # 监测网络端口,用ipconfig查看
    virtual_router_id 108   # 主备机必须相同
    priority 100            # 主备机取不同的优先级,主机要大。
    advert_int 1            # VRRP Multicast广播周期秒数
    authentication {
        auth_type PASS      # VRRP认证方式
        auth_pass 1234      # VRRP口令 主备机密码必须相同
    }

    track_script {          # 调用haproxy进程检测脚本,备节点不配置
        chk_haproxy
    }
    track_interface {
        bond0
    }
    virtual_ipaddress {     # VIP 漂移地址 即集群IP地址
        192.168.0.100/25 dev ens33
    }
}

修改192.168.0.105节点的配置文件:

vrrp_instance haproxy {
    state BACKUP
    interface ens33
    virtual_router_id 108
    priority 99
    advert_int 1
    authentication {
        auth_type PASS
        auth_pass 1234
    }
    track_interface {
        ens33
    }
    virtual_ipaddress {
        192.168.0.100
    }
    notify_master "/etc/keepalived/notify.sh master" # 当前节点成为master时,通知脚本执行任务,一般用于启动某服务
    notify_backup "/etc/keepalived/notify.sh backup" # 当前节点成为backup时,通知脚本执行任务,一般用于关闭某服务
}

在两个节点分别启动keepalived:

systemctl stop NetworkManager
systemctl start keepalived
systemctl enable keepalived

通过虚拟IP访问控制台

十一、使用KeepAlived为RabbitMQ的高可用的负载均衡-LMLPHP

十一、使用KeepAlived为RabbitMQ的高可用的负载均衡-LMLPHP

通过程序测试

修改RabbitMQ的连接地址为虚拟IP:

spring:
  #消息队列配置
  rabbitmq:
    #addresses: 192.168.0.101:5672,192.168.0.102:5672,192.168.0.103:5672
    host: 192.168.0.100 # 虚拟IP
    port: 5672
    username: guest
    password: guest
    virtual-host: /
    publisher-returns: true
    publisher-confirms: true
    connection-timeout: 5000ms
    listener:
      simple:
        acknowledge-mode: manual
        prefetch: 1000

执行发送消息和接收消息,结果为:

2019-05-05 23:04:15.508  INFO 77132 --- [.168.0.100:5672] cc.xuepeng.config.RabbitMQConfig         : ConfirmCallback -> 消息发布到交换器成功,id:CorrelationData [id=581660c3-e3f9-40a4-8905-50dd3681fd50]
2019-05-05 23:04:15.518  INFO 68844 --- [cTaskExecutor-1] cc.yesway.service.DirectQueueListener    : 消费消息成功: ack
05-06 08:35