环境准备

测试一下keepalived的双主模式，所谓双主模式就是两个keepavlied节点各持有一个/组虚IP，默认情况下，二者互为主备，同时对外提供服务，任何一个节点宕机，虚IP自动漂移到另外一台服务器上，从而实现双活高可用。
这里准备两台CentOS服务器，IP规划及服务器拓扑如下：

部署安装

上传安装包，并在两台服务器上安装 keepalived 和nginx,具体安装过程参见：
nginx安装
keepalived安装配置

drwxrwxr-x. 11 1000 1000    4096 Feb  2 22:29 keepalived-2.2.8
-rw-r--r--.  1 root root 1202602 Nov 29 14:15 keepalived-2.2.8.tar.gz
lrwxrwxrwx.  1 root root      12 Feb  2 22:19 nginx -> nginx-1.24.0
drwxr-xr-x.  9 1001 1001     186 Feb  2 22:20 nginx-1.24.0
-rw-r--r--.  1 root root 1112471 Nov 22 09:28 nginx-1.24.0.tar.gz

完成安装后,可以通过
systemctl start/stop/restart keepalived 来启动停止重启 keepalived

[root@localhost apps]# systemctl status keepalived
● keepalived.service - LVS and VRRP High Availability MonitorLoaded: loaded (/usr/lib/systemd/system/keepalived.service; enabled; vendor preset: disabled)Active: active (running) since Tue 2024-02-20 23:53:53 +08; 57min agoDocs: man:keepalived(8)man:keepalived.conf(5)man:genhash(1)https://keepalived.orgProcess: 6926 ExecStart=/usr/local/keeplived/sbin/keepalived $KEEPALIVED_OPTIONS (code=exited, status=0/SUCCESS)Main PID: 6927 (keepalived)CGroup: /system.slice/keepalived.service├─6927 /usr/local/keeplived/sbin/keepalived -D└─6928 /usr/local/keeplived/sbin/keepalived -D

keepavlived配置

keepalived 配置文件路径如下
/usr/local/keeplived/etc/keepalived
修改keepalived 配置节点1配置如下：

-rw-r--r--. 1 root root 1039 Feb 20 23:53 keepalived.conf
-rw-r--r--. 1 root root 3550 Feb  2 22:29 keepalived.conf.sample
-rwxr-xr-x. 1 root root   32 Feb 20 22:44 nginx_check.sh
drwxr-xr-x. 2 root root 4096 Feb  2 22:29 samples

global_defs {router_id KEEPALIVED_NODE_11  # 每个keepalived节点要唯一
}
# 这里配置一个nginx 状态检测脚本
vrrp_script chk_nginx {script "/usr/local/keeplived/etc/keepalived/nginx_check.sh"interval 2weight -20fall 2rise 1
}vrrp_instance VI_1{state MASTERinterface ens33virtual_router_id 51priority 100advert_int 1authentication {auth_type PASSauth_pass 1111  # 每个虚IP的pass 最好配置单独的密码，这个是通信权限鉴定的手段}track_script {chk_nginx}unicast_src_ip 192.168.126.11   #配置单播原地址unicast_peer {192.168.126.12}virtual_ipaddress {192.168.126.21/24}
}
vrrp_instance VI_2{state BACKUPinterface ens33   virtual_router_id 52priority 80advert_int 1authentication {auth_type PASSauth_pass 11111}track_script {chk_nginx}unicast_src_ip 192.168.126.11   #配置单播原地址  unicast_peer {192.168.126.12}virtual_ipaddress {192.168.126.22/24}
}

keepalived节点2配置如下

global_defs {router_id KEEPALIVED_NODE_12
}vrrp_script chk_nginx {script "/usr/local/keeplived/etc/keepalived/nginx_check.sh" interval 2 weight -20   #上述脚本返回非0时 自动降低优先级 20fall 2 rise 1 
}vrrp_instance VI_1{state BACKUP interface ens33 virtual_router_id 51   # 每个虚IP的id必须唯一 priority 80 advert_int 1 authentication {         auth_type PASSauth_pass 1111  # 每个虚IP的pass 最好配置单独的密码，这个是通信权限鉴定的手段}track_script {chk_nginx}unicast_src_ip 192.168.126.12   #配置单播原地址unicast_peer {192.168.126.11}virtual_ipaddress {192.168.126.21/24 }
}
vrrp_instance VI_2{state MASTERinterface ens33virtual_router_id 52  # 每个虚IP的id必须唯一 如这里不能和 VI1中的 virtual_router_id  重复priority 100 advert_int 1 authentication {         auth_type PASSauth_pass 11111}track_script {chk_nginx}unicast_src_ip 192.168.126.12   #配置单播原地址unicast_peer {192.168.126.11}virtual_ipaddress {192.168.126.22/24}
}

上述配置中用到了一个nginx_check.sh脚本编写脚本内容如下

[root@localhost keepalived]# vi nginx_check.sh 
nginxpid=`ps -C nginx --no-header | wc -l`
#!/bin/bash
DIR=`ps -C nginx --no-header | wc -l`
if [ 0 -eq $DIR ]# 如果nginx进程消失  则返回2
then
exit 2  
fi

启动测试

根据上述配置，启动keepalived 和 nginx
可以看到和预期相符 .11服务器持有 192.168.126.21 虚ip .12服务器持有 192.168.126.22 虚ip


修改nginx index.html 文件内容 ，增加本机IP用于区分开两台服务器

分别通过虚IP访问web服务 http://192.168.126.21 和 http://192.168.126.22 分别访问到了 .11 和.12服务器

模拟Nginx宕机

模拟宕机，停止192.168.126.11上的nginx服务

等待一定时间后可以看到keepalived日志中，VI_1 和VI_2 的优先级都降低

之后又等了很长时间 没有再看到其他日志输出。怀疑可能和他的 机制有关，执行一次失败后就不在执行了。执行失败后优先级-20 刚好和 备节点优先级相同 所以，没有进行主备切换。我们修改配置文件，脚本执行返回非0结果后优先级-30，保存并重启服务

启动nginx服务重新测试

[root@localhost keepalived]# systemctl restart keepalived
[root@localhost keepalived]# /usr/local/nginx/sbin/nginx 

停止nginx服务，通过keepalived日志看到 优先级降低并切换到了 BACKUP 状态,虚ip漂移到了.12服务器上。
keepalived 日志:/var/log/messages

通过虚IP 192.168.126.21 访问web服务直接访问到了.12 节点

重新启动

重新启动.11 节点上的 nginx，可以看到优先级又回来了，VI_1状态又变回了MASTER

同时.12节点变为了BACKUP状态.

问题分析

在测试过程中出现了两个nginx节点都是主的情况，检查发现是防火墙开启了 导致状态监听报文无法发送和接收。双方接收不到其他节点的状态报文，所以就认为自己是MASTER。