mysql+keepalived 切换后vip无法被备机接管，反之可以

Question

今天一共搭建了6台3组两两互备的centos 6.5 mysql+keepalived的主主复制+双backup模式的集群。主机107的keepalived.conf如下：
! Configuration File for keepalived

global_defs {
notification_email {
[e-mail]
[e-mail]
[e-mail]
}
notification_email_from [e-mail]
smtp_server 127.0.0.1
smtp_connect_timeout 30
router_id mysql_ha
vrrp_skip_check_adv_addr
vrrp_strict
vrrp_garp_interval 0
vrrp_gna_interval 0
}

vrrp_instance VI_1 {
state BACKUP
interface eth0
virtual_router_id 117
priority 100
advert_int 1
nopreempt
authentication {
auth_type PASS
auth_pass 1111
}
virtual_ipaddress {
10.40.6.117
}
}

virtual_server 10.40.6.117 3306 {
delay_loop 2
#lb_algo wrr
#lb_kind DR
persistence_timeout 60
protocol TCP

real_server 10.40.6.107 3306 {
    weight 3
    notify_down /usr/local/etc/keepalived/mysql.sh
    TCP_CHECK {
        connect_timeout 3
        nb_get_retry 3
        delay_before_retry 3
        connect_port 3306
    }
}

}
备机108的keepalived.conf如下：
! Configuration File for keepalived

global_defs {
notification_email {
[e-mail]
[e-mail]
[e-mail]
}
notification_email_from [e-mail]
smtp_server 127.0.0.1
smtp_connect_timeout 30
router_id mysql_ha
vrrp_skip_check_adv_addr
vrrp_strict
vrrp_garp_interval 0
vrrp_gna_interval 0
}

vrrp_instance VI_1 {
state BACKUP
interface eth0
virtual_router_id 117
priority 90
advert_int 1
authentication {
auth_type PASS
auth_pass 1111
}
virtual_ipaddress {
10.40.6.117
}
}

virtual_server 10.40.6.117 3306 {
delay_loop 2
#lb_algo wrr
#lb_kind DR
persistence_timeout 60
protocol TCP

real_server 10.40.6.108 3306 {
    weight 3
    notify_down /usr/local/etc/keepalived/mysql.sh
    TCP_CHECK {
        connect_timeout 3
        nb_get_retry 3
        delay_before_retry 3
        connect_port 3306
    }
}

}

其中有两台集群很奇怪，real_ip分别为107和108，vip为117，当MySQL服务和keepalived服务都启动完成后，一切正常，107占有117的虚拟ip，此时测试切换，将mysql服务停止，按理说3306端口检查不健康的时候会执行我的mysql.sh脚本，实际上就是pkill keepalived，使备机占有vip，但实际上117的vip没有正常漂移到备机，而一直被主机占有，查看message日志发现一直报错：

Jul 12 16:09:12 hs-10-40-6-107 Keepalived_healthcheckers[9204]: TCP connection to [10.40.6.107]:3306 failed.
Jul 12 16:09:15 hs-10-40-6-107 Keepalived_healthcheckers[9204]: TCP connection to [10.40.6.107]:3306 failed.
Jul 12 16:09:15 hs-10-40-6-107 Keepalived_healthcheckers[9204]: Check on service [10.40.6.107]:3306 failed after 1 retry.
Jul 12 16:09:15 hs-10-40-6-107 Keepalived_healthcheckers[9204]: Removing service [10.40.6.107]:3306 from VS [10.40.6.117]:3306
Jul 12 16:09:15 hs-10-40-6-107 Keepalived_healthcheckers[9204]: IPVS: Service not defined
Jul 12 16:09:15 hs-10-40-6-107 Keepalived_healthcheckers[9204]: SMTP connection ERROR to [127.0.0.1]:25.
Jul 12 16:09:17 hs-10-40-6-107 Keepalived_healthcheckers[9204]: TCP connection to [10.40.6.107]:3306 failed.
Jul 12 16:09:20 hs-10-40-6-107 Keepalived_healthcheckers[9204]: TCP connection to [10.40.6.107]:3306 failed.
Jul 12 16:09:20 hs-10-40-6-107 Keepalived_healthcheckers[9204]: Check on service [10.40.6.107]:3306 failed after 1 retry.
Jul 12 16:09:20 hs-10-40-6-107 Keepalived_healthcheckers[9204]: Removing service [10.40.6.107]:3306 from VS [10.40.6.117]:3306
Jul 12 16:09:20 hs-10-40-6-107 Keepalived_healthcheckers[9204]: IPVS: Service not defined
Jul 12 16:09:20 hs-10-40-6-107 Keepalived_healthcheckers[9204]: SMTP connection ERROR to [127.0.0.1]:25.

然后重新恢复所有服务，测试从108切换到107，一切正常，108的mysql 停止之后，执行notify_down脚本，杀掉keepalived进程，从而使之前108所占用的vip 117备107所抢占，108的操作系统日志如下：
ul 12 14:18:40 hs-10-40-6-108 Keepalived_healthcheckers[6258]: TCP connection to [10.40.6.108]:3306 failed.
Jul 12 14:18:43 hs-10-40-6-108 Keepalived_healthcheckers[6258]: TCP connection to [10.40.6.108]:3306 failed.
Jul 12 14:18:43 hs-10-40-6-108 Keepalived_healthcheckers[6258]: Check on service [10.40.6.108]:3306 failed after 1 retry.
Jul 12 14:18:43 hs-10-40-6-108 Keepalived_healthcheckers[6258]: Removing service [10.40.6.108]:3306 from VS [10.40.6.117]:3306
Jul 12 14:18:43 hs-10-40-6-108 Keepalived_healthcheckers[6258]: IPVS: No such destination
Jul 12 14:18:43 hs-10-40-6-108 Keepalived_healthcheckers[6258]: Executing [/usr/local/etc/keepalived/mysql.sh] for service [10.40.6.108]:3306 in VS [10.40.6.117]:3306
Jul 12 14:18:43 hs-10-40-6-108 Keepalived_healthcheckers[6258]: Lost quorum 1-0=1 > 0 for VS [10.40.6.117]:3306
Jul 12 14:18:43 hs-10-40-6-108 Keepalived_healthcheckers[6258]: SMTP connection ERROR to [127.0.0.1]:25.
Jul 12 14:18:43 hs-10-40-6-108 Keepalived_vrrp[6259]: VRRP_Instance(VI_1) sent 0 priority
Jul 12 14:18:43 hs-10-40-6-108 Keepalived[6257]: Stopping
Jul 12 14:18:43 hs-10-40-6-108 Keepalived_vrrp[6259]: VRRP_Instance(VI_1) removing protocol VIPs.
Jul 12 14:18:43 hs-10-40-6-108 Keepalived_healthcheckers[6258]: Netlink reflector reports IP 10.40.6.117 removed
Jul 12 14:18:43 hs-10-40-6-108 Keepalived_healthcheckers[6258]: IPVS: No such file or directory
Jul 12 14:18:43 hs-10-40-6-108 Keepalived_healthcheckers[6258]: Stopped

今天一共装了6台机器，只有这一组主切备的时候有问题，notify_down 脚本一直不会执行，并且报错，不知道哪位大牛知道原因？

Answer 1

keepalived 切换后vip无法被备机接管，反之可以 1C
今天一共搭建了6台3组两两互备的centos 6.5 mysql+keepalived的主主复制+双backup模式的集群。主机107的keepalived.conf如下：
! Configuration File for keepalived
global_defs {
notification_email {
[e-mail]
[e-mail]
[e-mail]
}
notification_email_from [e-mail]
smtp_server 127.0.0.1
smtp_connect_timeout 30
router_id mysql_ha
vrrp_skip_check_adv_addr
vrrp_strict