Keepalived安装与配置详解：手把手教学，轻松上手

目录

1 Keepalived简介

1.1 VRRP协议简介

1.2 Keepalived工作原理

2 Keepalived安装

2.1 环境准备

2.2 下载Keepalived

2.3 解压、编译并安装

2.4 复制Keepalived到系统服务目录

2.5 复制配置文件

2.6 设置开机自启动

2.7 验证安装

3 Keepalived配置

3.1 配置示例

3.1.1 主服务器配置

3.1.2 备服务器配置

3.2 配置解析

VRRP脚本检查 

VRRP同步组

VRRP实例

虚拟服务器配置

4 启动Keepalived服务

5 检查VIP绑定情况

6 手动触发故障切换

7 附件

7.1 MySQL状态健康脚本

7.2 Keepalived监控脚本

---

1 Keepalived简介

Keepalived：是一个基于 VRRP（Virtual Router Redundancy Protocol，虚拟路由器冗余协议）的高可用性解决方案，主要用于实现 IP 地址的故障转移和高可用性。它可以在多个服务器之间实现负载均衡和故障切换，确保服务的持续可用性。Keepalived 通常用于 Web 服务器、数据库服务器等关键服务的故障转移。

1.1 VRRP协议简介

VRRP：是一种网络协议，用于在多个路由器之间提供冗余。通过VRRP，多个路由器可以组成一个虚拟路由器组，其中一个路由器作为主路由器（Master），其他路由器作为备份路由器（Backup）。当主路由器发生故障时，备份路由器会自动接管主路由器的职责，确保网络的持续可用性。

1.2 Keepalived工作原理

Keepalived通过VRRP协议实现IP地址的故障转移。它会在多个服务器之间选举出一个主服务器（Master），并将虚拟IP地址（VIP）绑定到主服务器上。当主服务器发生故障时，Keepalived会自动将VIP转移到备份服务器（Backup）上，从而实现服务的高可用性。

2 Keepalived安装

2.1 环境准备

##安装依赖包
yum install gcc make openssl openssl-devel -y

2.2 下载Keepalived

网址：Keepalived for Linux

2.3 解压、编译并安装

tar -zxvf keepalived-2.2.8.tar.gz
cd keepalived-2.2.8
./configure --prefix=/usr/local/keepalived
 make install

2.4 复制Keepalived到系统服务目录

cp /root/tool/keepalived-2.2.8/keepalived/etc/init.d/keepalived /etc/rc.d/init.d

2.5 复制配置文件

mkdir -p /etc/keepalived
cp /usr/local/keepalived/etc/keepalived/keepalived.conf.sample /etc/keepalived/keepalived.conf

2.6 设置开机自启动

chkconfig --add keepalived
chkconfig keepalived on

2.7 验证安装

/usr/local/keepalived/sbin/keepalived --version

[root@node-2 ~]# /usr/local/keepalived/sbin/keepalived --version
Keepalived v2.2.8 (04/04,2023), git commit v2.2.7-154-g292b299e+

Copyright(C) 2001-2023 Alexandre Cassen, <acassen@gmail.com>

Built with kernel headers for Linux 3.10.0
Running on Linux 3.10.0-1160.el7.x86_64 #1 SMP Mon Oct 19 16:18:59 UTC 2020
Distro: CentOS Linux 7 (Core)

configure options: --prefix=/usr/local/keepalived

Config options:  LVS VRRP VRRP_AUTH VRRP_VMAC OLD_CHKSUM_COMPAT INIT=systemd

System options:  VSYSLOG RTA_ENCAP RTA_EXPIRES RTA_PREF FRA_SUPPRESS_PREFIXLEN FRA_TUN_ID RTAX_CC_ALGO RTAX_QUICKACK RTA_VIA IFA_FLAGS NET_LINUX_IF_H_COLLISION LIBIPTC_LINUX_NET_IF_H_COLLISION IFLA_LINK_NETNSID GLOB_BRACE GLOB_ALTDIRFUNC INET6_ADDR_GEN_MODE SO_MARK
[root@node-2 ~]# 

3 Keepalived配置

        这里我们以MySQL主从复制为例进行配置示例操作，用于在两个服务器之间实现IP地址的故障转移。

3.1 配置示例

3.1.1 主服务器配置

cp /etc/keepalived/keepalived.conf /etc/keepalived/keepalived.conf_bak
>/etc/keepalived/keepalived.conf

vi /etc/keepalived/keepalived.conf

#编辑文件添加如下内容
vrrp_script vs_mysql_dbchk {
    script "/etc/keepalived/chk_mysql.sh"
    interval 10
}

vrrp_sync_group VG_1 {
group {
VM_1
}
}

vrrp_instance VM_1 {
    state MASTER					#在备机上修改为BACKUP
    interface ens33				    #VIP要绑定到enss33上,是具体情况而定，填写具体的主机网卡名称
    virtual_router_id 51
    priority 100					#对应备机的值要小于这个值
    nopreempt
    advert_int 1
    authentication {
        auth_type PASS				#备机上要与之一致
        auth_pass La123456				#备机上要与之一致
    }
    track_script {
        vs_mysql_dbchk
        }
    virtual_ipaddress {
        192.168.10.29/24 dev ens33 label ens33:0	#VIP要绑定到ens33上,是具体情况而定，填写具体的主机网卡名称,修改为对应的VIP
    }
}

virtual_server 192.168.10.29 3306{			#修改为对应的VIP
        delay_loop 6
        lb_algo rr					        #lvs负载均衡算法
        lb_kind DR					        #lvs的转发模式
        nat_mask 255.255.255.0
        persistence_timeout 50
        protocol TCP

        real_server 192.168.10.30 3306 {		#修改为对应的realserever
                weight 2
                TCP_CHECK {
                connect_timeout 3
                nb_get_retry 3
                delay_before_retry 3
                connect_port 3306
                }
        }
        real_server 192.168.10.31 3306 {		#修改为对应的realserver
                weight 2
                TCP_CHECK {
                connect_timeout 3
                nb_get_retry 3
                delay_before_retry 3
                connect_port 3306
                    }
        }
}

3.1.2 备服务器配置

cp /etc/keepalived/keepalived.conf /etc/keepalived/keepalived.conf_bak
>/etc/keepalived/keepalived.conf

vim >/etc/keepalived/keepalived.conf

##编辑问价添加如下内容
vrrp_script vs_mysql_dbchk {
    script "/etc/keepalived/chk_mysql.sh"
    interval 10
}

vrrp_sync_group VG_1 {
    group {
        VM_1
    }
}

vrrp_instance VM_1 {
    state BACKUP                    # 备节点设置为 BACKUP
    interface ens33                 # VIP 绑定的网卡名称（与主节点一致）
    virtual_router_id 51            # VRRP 路由 ID，必须与主节点一致
    priority 90                     # 优先级，备节点的值要小于主节点
    nopreempt                       # 禁止抢占
    advert_int 1                    # VRRP 通告间隔时间（秒）
    authentication {
        auth_type PASS              # 认证类型，必须与主节点一致
        auth_pass La123456          # 认证密码，必须与主节点一致
    }
    track_script {
        vs_mysql_dbchk              # 关联的 VRRP 脚本检查
    }
    virtual_ipaddress {
        192.168.10.29/24 dev ens33 label ens33:0  # VIP 地址及绑定的网卡
    }
}

virtual_server 192.168.10.29 3306 {  # VIP 地址及端口
    delay_loop 6                    # 检查间隔时间（秒）
    lb_algo rr                      # LVS 负载均衡算法（轮询）
    lb_kind DR                      # LVS 转发模式（直接路由）
    nat_mask 255.255.255.0
    persistence_timeout 50          # 持久连接超时时间（秒）
    protocol TCP                    # 协议类型

    real_server 192.168.10.30 3306 {  # 真实服务器 1
        weight 2                    # 权重
        TCP_CHECK {
            connect_timeout 3       # 连接超时时间（秒）
            nb_get_retry 3          # 重试次数
            delay_before_retry 3    # 重试间隔时间（秒）
            connect_port 3306       # 检查的端口
        }
    }
    real_server 192.168.10.31 3306 {  # 真实服务器 2（如果有多个后端服务器）
        weight 2
        TCP_CHECK {
            connect_timeout 3
            nb_get_retry 3
            delay_before_retry 3
            connect_port 3306
        }
    }
}

3.2 配置解析

vrrp_script vs_mysql_dbchk {
    script "/etc/keepalived/chk_mysql.sh"
    interval 10
}

VRRP脚本检查 

• 作用：定义一个VRRP脚本检查，用于监控MySQL服务的状态
• script：指定脚本路径/etc/keepalived/chk_mysql.sh，该脚本用于检查MySQL是否正常运行
• interval：每10秒执行一次检查

vrrp_sync_group VG_1 {
    group {
        VM_1
    }
}

VRRP同步组

• 作用：将多个 VRRP 实例组合在一起，确保它们的状态同步
• group：包含一个 VRRP 实例VM_1

vrrp_instance VM_1 {
    state MASTER                  # 在备机上修改为 BACKUP
    interface ens33               # VIP 绑定的网卡名称
    virtual_router_id 51          # VRRP 路由 ID，主备机必须一致
    priority 100                  # 优先级，备机的值要小于这个值
    nopreempt                     # 禁止抢占
    advert_int 1                  # VRRP 通告间隔时间（秒）
    authentication {
        auth_type PASS            # 认证类型，备机必须一致
        auth_pass La123456        # 认证密码，备机必须一致
    }
    track_script {
        vs_mysql_dbchk            # 关联的 VRRP 脚本检查
    }
    virtual_ipaddress {
        192.168.10.29/24 dev ens33 label ens33:0  # VIP 地址及绑定的网卡
    }
}

VRRP实例

• state：当前节点的角色，MASTER表示主节点，备节点需要改为BACKUP
• interface：VIP绑定的网卡名称（如ens33）。
• virtual_router_id：VRRP路由 ID，主备机必须一致
• priority：优先级，主节点值应大于备节点
• nopreempt：禁止抢占模式，确保主节点故障恢复后不会抢占VIP

• virtual_ipaddress：VIP地址及绑定的网卡

virtual_server 192.168.10.29 3306 {  # VIP 地址及端口
    delay_loop 6                    # 检查间隔时间（秒）
    lb_algo rr                      # LVS 负载均衡算法（轮询）
    lb_kind DR                      # LVS 转发模式（直接路由）
    nat_mask 255.255.255.0
    persistence_timeout 50          # 持久连接超时时间（秒）
    protocol TCP                    # 协议类型

    real_server 192.168.10.30 3306 {  # 真实服务器 1
        weight 2                    # 权重
        TCP_CHECK {
            connect_timeout 3       # 连接超时时间（秒）
            nb_get_retry 3          # 重试次数
            delay_before_retry 3    # 重试间隔时间（秒）
            connect_port 3306       # 检查的端口
        }
    }
    real_server 192.168.10.30 3306 {  # 真实服务器 2
        weight 2
        TCP_CHECK {
            connect_timeout 3
            nb_get_retry 3
            delay_before_retry 3
            connect_port 3306
        }
    }
}

虚拟服务器配置

• virtual_server：定义虚拟服务器（VIP）及其端口
• lb_algo：LVS负载均衡算法，rr 表示轮询
• lb_kind：LVS转发模式，DR表示直接路由
• real_server：定义真实服务器（后端 MySQL 服务器）及其健康检查配置

4 启动Keepalived服务

systemctl start keepalived

[root@node-1 keepalived-2.2.8]# systemctl status keepalived
● keepalived.service - LVS and VRRP High Availability Monitor
   Loaded: loaded (/usr/lib/systemd/system/keepalived.service; enabled; vendor preset: disabled)
   Active: active (running) since Sun 2025-03-02 21:09:04 CST; 2s ago
     Docs: man:keepalived(8)
           man:keepalived.conf(5)
           man:genhash(1)
           https://keepalived.org
  Process: 41974 ExecStart=/usr/local/keepalived/sbin/keepalived $KEEPALIVED_OPTIONS (code=exited, status=0/SUCCESS)
 Main PID: 41975 (keepalived)
    Tasks: 3
   Memory: 928.0K
   CGroup: /system.slice/keepalived.service
           ├─41975 /usr/local/keepalived/sbin/keepalived -D
           ├─41976 /usr/local/keepalived/sbin/keepalived -D
           └─41977 /usr/local/keepalived/sbin/keepalived -D

Mar 02 21:09:04 node-1 Keepalived_vrrp[41977]: Assigned address 192.168.10.30 for interface ens33
Mar 02 21:09:04 node-1 Keepalived_healthcheckers[41976]: Gained quorum 1+0=1 <= 4 for VS [192.168.10.29]:tcp:3306
Mar 02 21:09:04 node-1 Keepalived_healthcheckers[41976]: Activating healthchecker for service [192.168.10.30]:tcp:3306 for VS [192.168.10.29]:tcp:3306
Mar 02 21:09:04 node-1 Keepalived_healthcheckers[41976]: Activating healthchecker for service [192.168.10.31]:tcp:3306 for VS [192.168.10.29]:tcp:3306
Mar 02 21:09:04 node-1 Keepalived_vrrp[41977]: Registering gratuitous ARP shared channel
Mar 02 21:09:04 node-1 Keepalived_vrrp[41977]: (VM_1) removing VIPs.
Mar 02 21:09:04 node-1 Keepalived[41975]: Startup complete
Mar 02 21:09:04 node-1 Keepalived_vrrp[41977]: (VM_1) Entering BACKUP STATE (init)
Mar 02 21:09:04 node-1 Keepalived_vrrp[41977]: VRRP sockpool: [ifindex(  2), family(IPv4), proto(112), fd(12,13) multicast, address(224.0.0.18)]
Mar 02 21:09:06 node-1 Keepalived_healthcheckers[41976]: TCP connection to [192.168.10.30]:tcp:3306 success.
[root@node-1 keepalived-2.2.8]# 

5 检查VIP绑定情况

ip addr show ens33

[root@node-2 ~]# ip addr show ens33
2: ens33: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
    link/ether 00:0c:29:8f:5d:8e brd ff:ff:ff:ff:ff:ff
    inet 192.168.10.31/24 brd 192.168.10.255 scope global noprefixroute ens33
       valid_lft forever preferred_lft forever
    inet 192.168.10.29/24 scope global secondary ens33:0
       valid_lft forever preferred_lft forever
[root@node-2 ~]# 

6 手动触发故障切换

        通过手动停止主节点的Keepalived服务，测试VIP是否漂移到备节点。

##主节点上运行停止命令
[root@node-2 ~]# systemctl stop keepalived
[root@node-2 ~]# 

##备节点上检查VIP是否绑定
[root@node-1 keepalived-2.2.8]# ip addr show ens33
2: ens33: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
    link/ether 00:0c:29:e5:d1:9d brd ff:ff:ff:ff:ff:ff
    inet 192.168.10.30/24 brd 192.168.10.255 scope global noprefixroute ens33
       valid_lft forever preferred_lft forever
    inet 192.168.10.29/24 scope global secondary ens33:0
       valid_lft forever preferred_lft forever
[root@node-1 keepalived-2.2.8]# 

7 附件

7.1 MySQL状态健康脚本

#!/bin/bash
#每隔10秒（keepalived配置文件里面设置）检查一次mysql进程和状态是否正常，不正常则关闭keepalived服务以停止对外提供服务


user=root
pass=La12345678
sockfile=/tmp/mysql.sock
logfile=/var/log/keepalived/mysql_check.log

count=1
while true
do
    mysql -u${user} -p${pass} -S ${sockfile} -e "show status;" > /dev/null 2>&1
    i=$?
    ps aux | grep mysqld | grep -v grep > /dev/null 2>&1
    j=$?

    if [ $i = 0 ] && [ $j = 0 ]
    then
        exit 0
    else
        if [ $count -gt 5 ]
        then
            echo "$(date): MySQL check failed, stopping Keepalived" >> $logfile
            systemctl stop keepalived
            exit 1
        fi
        let count++
        sleep 10
        continue
    fi
done

7.2 Keepalived监控脚本

#!/bin/bash
#监控keepalived，keepalived服务正常则退出不做任何操作，服务异常，则每5s检查一次mysql进程和状态是否正常，检查10次，Mysql进程和状态都正常则重新启动keepalived服务

user=root
pass=La12345678
sockfile=/tmp/mysql.sock
logfile=/var/log/keepalived/monitor.log

count=1
while true
do
    # 检查 Keepalived 进程
    h=$(ps aux | grep keepalived | grep -v grep | wc -l)

    # 如果 Keepalived 进程存在，退出脚本
    if [ $h -gt 0 ]
    then
        echo "$(date): Keepalived is running, exiting" >> $logfile
        exit 0
    fi

    # 检查 MySQL 服务状态
    mysql -u${user} -p${pass} -S ${sockfile} -e "show status;" > /dev/null 2>&1
    i=$?
    ps aux | grep mysqld | grep -v grep > /dev/null 2>&1
    j=$?

    # 如果 MySQL 服务正常，重启 Keepalived
    if [ $i = 0 ] && [ $j = 0 ]
    then
        echo "$(date): MySQL is healthy, restarting Keepalived" >> $logfile
        systemctl start keepalived
        exit 0
    else
        # 如果重试次数超过 10 次，退出脚本
        if [ $count -gt 10 ]
        then
            echo "$(date): MySQL check failed after 10 retries, exiting" >> $logfile
            exit 1
        fi
        # 等待 5 秒后重试
        sleep 5
        let count++
        continue
    fi
done