why amazon cloud load balancer was flapping between two instances?

Issue/Symptom :

• Why SMOKETEST LB in amazon aws cloud was flapping between qpass-prod-smktst-201.dub1.qpass.net & qpass-prod-smktst-101.dub1.qpass.net. States changed between "InService" & "OutofService" ?

■ OS Environment :

• Both nodes has RHEL 6, LB is provided by Amazon

■ Investigation :

• LB is mapped to above two nodes. Incomming port is 443, destination port is 80. It is found that applications are listening port 80 on both nodes. Server Health check timeout has been increased in LB, but issue still persisted.

■ Permanent Solution :

For the LB, at Health check section, Ping target will be TCP:80, Timeout set 5 seconds, Interval 30 seconds, Unhealthy Threshold 2
Healthy Threshold  10

■ Root Cause Analysis :

• It was found that Ping target was HTTP:80, Ping path was /ping.html. Though webbased ping returns OK[200 status code] but it does't work properly.