Keepalived master: 10.192.52.60
Keepalived backup: 10.192.52.61
For webserver1 HTTP service: 10.192.52.62 (in/var/VWW/HTML/index. The HTML to write node3)
Webserver2 HTTP service: 10.192.52.63 (in/var/VWW/HTML/index. The HTML to write node4)
In for webserver1, VIP 10.192.52.70 webserver2 configuration
The client (the curl command) : 10.192.44.150
The structure of
Configuration file is (adopt the way of polling)
Keepalived master
Keepalived backup
two arp file on the webserver configuration
related to the log file :
Keepalive master and backup normal boot
ipvsadm - l - n information
Problems
To access through the curl 10.192.52.70 the VIP will also start very happy polling
The ipvsadm - l - n InActConn data on the increase, but ActConn 0
After a period of time, not in polling, only to 10.192.52.63 (node4) request,
then use tcpdump host 10.192.44.150 command (see the client sends packets to) in 10.192.52.60 (keepalived master node) 10.192.52.62 (node) for webserver1 10.192.52.63 (webserver2 node) to check the packet information found by the client (10.192.44.150) sending a request, without the LVS, directly to the 10.192.52.63 (webserver2 node), I think this is the reason of abnormal polling, so the question for why will appear this kind of problem? And this kind of situation will appear alternately, polling is normal after a while, a moment is unusual,