Keepalived VIP切换的两种情况:
1.keepalived 进程被停止。
2.keepalived中track_script检测失败。
haproxy 检测:
vrrp_script chk_haproxy {
script "killall -0 haproxy" # cheaper than pidof
interval 2 # check every 2 seconds
weight -2
}
切换相关日志:tail -f /var/log/messages
Keepalived_vrrp[32408]: VRRP_Script(chk_haproxy) failed
Keepalived_vrrp[32408]: VRRP_Instance(VI_1) Received higher prio advert
Keepalived_vrrp[32408]: VRRP_Instance(VI_1) Entering BACKUP STATE
Keepalived_vrrp[32408]: VRRP_Instance(VI_1) removing protocol VIPs.
Keepalived_vrrp[32408]: VRRP_Script(chk_haproxy) succeeded
Keepalived_vrrp[32408]: VRRP_Instance(VI_1) forcing a new MASTER election
Keepalived_vrrp[32408]: VRRP_Instance(VI_1) forcing a new MASTER election
Keepalived_vrrp[32408]: VRRP_Instance(VI_1) Transition to MASTER STATE
Keepalived_vrrp[32408]: VRRP_Instance(VI_1) Entering MASTER STATE
Keepalived_vrrp[32408]: VRRP_Instance(VI_1) setting protocol VIPs.
Keepalived_vrrp[32408]: VRRP_Instance(VI_1) Sending gratuitous ARPs on eth1 for 10.1.1.56
Keepalived_vrrp[32408]: VRRP_Instance(VI_1) Sending gratuitous ARPs on eth1 for 10.1.1.56
切换失败的情况:
keepalived 进程被停止时VIP可以正常切换,keepalived 进程运行状态下,haproxy相关服务停止后,VIP未如期进行切换。
分析:
停止haproxy服务,手动执行killall -0 haproxy && echo "success" || echo "fail",返回结果是success。
停止haproxy服务后,ps aux |grep haproxy发现还有一个正常启动的haproxy,killall -9 haproxy后发现 tail -f /var/log/messages 出现VIP正常切换的日志。
Linux下HAProxy+Keepalived双机高可用方案
Haproxy+Keepalived搭建Weblogic高可用负载均衡集群
CentOS 6.3下Haproxy+Keepalived+Apache配置笔记