CentOS 配置 keepalived 实现服务 HA

0. 背景

局域网内两台主机

192.168.1.16 主节点
192.168.1.43 备节点

分别在 9439 端口提供相同的服务，现在使用 keepalived 配置 HA。

1. 安装keepalived

分别在两台主机安装 keepalived及相关依赖包

yum install -y net-tools psmisc
yum install -y net-snmp
yum install -y ipvsadm  keepalived

2. 配置 keepalived

（1）配置主节点192.168.1.16
新建文件配置文件/etc/keepalived/keepalived.conf

! Configuration File for keepalived
global_defs {

}

vrrp_script check_docker_losa {
    # 自身状态检测
    script "sh /etc/keepalived/chk_losa.sh"
    interval 3
    weight 5
}

vrrp_instance VI_1 {
    # 初始化状态
    state MASTER
    # 虚拟 ip 绑定的网卡
    interface ens192
    # 此 ID 要与 Backup 配置一致
    virtual_router_id 80
    # 默认启动优先级，要比 Backup 大点，但要控制量，保证自身状态监测生效
    priority 100
    advert_int 1

    virtual_ipaddress {
        # 虚拟 ip 地址
        192.168.1.2
    }

    track_script {
        check_docker_losa
    }
}

另外在编写一个状态检测脚本/etc/keepalived/chk_losa.sh，检测 docker 中的服务是否正常

#!/bin/bash

RMQ_CONTAINER=$(docker ps -a|awk '/losa-front-nginx/ {print $1}')
if [[ -n "$RMQ_CONTAINER" ]]; then
   exit 0
fi
exit 1

（2）配置备节点192.168.1.43
新建文件配置文件/etc/keepalived/keepalived.conf

! Configuration File for keepalived
global_defs {

}

vrrp_script check_docker_losa {
    # 自身状态检测
    script "sh /etc/keepalived/chk_losa.sh"
    interval 3
    weight 5
}

vrrp_instance VI_1 {
    # 初始化状态
    state BACKUP
    # 虚拟 ip 绑定的网卡
    interface ens192
    # 此 ID 要与 MASTER 配置一致
    virtual_router_id 80
    # 默认启动优先级
    priority 90
    advert_int 1

    virtual_ipaddress {
        # 虚拟 ip 地址
        192.168.1.2
    }

    track_script {
        check_docker_losa
    }
}

状态检测脚本类似（1）中主节点配置

3. 启动 keepalived

（1）在主节点上(192.168.1.16)

# 启动 keepalived 服务
systemctl start keepalived

# 查看启动日志
tail -f /var/log/messages
Oct 29 05:23:18 localhost Keepalived_vrrp[129640]: VRRP_Instance(VI_1) Transition to MASTER STATE
Oct 29 05:23:19 localhost Keepalived_vrrp[129640]: VRRP_Instance(VI_1) Changing effective priority from 100 to 105
Oct 29 05:23:19 localhost Keepalived_vrrp[129640]: VRRP_Instance(VI_1) Entering MASTER STATE
Oct 29 05:23:19 localhost Keepalived_vrrp[129640]: VRRP_Instance(VI_1) setting protocol VIPs.
Oct 29 05:23:19 localhost Keepalived_vrrp[129640]: Sending gratuitous ARP on ens192 for 192.168.1.2
...

此时可通过192.168.1.2访问主节点上的服务

（2）在备节点上(192.168.1.43)

# 启动 keepalived 服务
systemctl start keepalived

# 查看启动日志
tail -f /var/log/messages
Oct 29 17:32:09 localhost Keepalived_vrrp[15599]: VRRP_Instance(VI_1) Entering BACKUP STATE
...

4. 验证双活效果

（1）停止主节点上的服务，或者使用systemctl stop keepalived停止主节点上的 keepalived 服务

（2）这时应观察到备节点192.168.1.43上的日志：

Oct 29 17:33:05 localhost Keepalived_vrrp[15599]: VRRP_Instance(VI_1) Transition to MASTER STATE
Oct 29 17:33:06 localhost Keepalived_vrrp[15599]: VRRP_Instance(VI_1) Entering MASTER STATE
Oct 29 17:33:06 localhost Keepalived_vrrp[15599]: VRRP_Instance(VI_1) setting protocol VIPs.
Oct 29 17:33:06 localhost Keepalived_vrrp[15599]: Sending gratuitous ARP on enp3s0 for 192.168.1.2
...

说明IP192.168.1.2已经成功漂移到了备节点上，验证访问192.168.1.2依然能够正常提供服务，此时是备节点提供服务。

（3）恢复主节点上的服务，或者使用systemctl start keepalived恢复 keepalived 服务

这时观察到备节点上的日志：

Oct 29 17:38:42 localhost Keepalived_vrrp[15599]: VRRP_Instance(VI_1) Entering BACKUP STATE
Oct 29 17:38:42 localhost Keepalived_vrrp[15599]: VRRP_Instance(VI_1) removing protocol VIPs.
...

说明192.168.1.2又漂移回了主节点上，主节点继续提供服务。