2 Machines - 1 cluster - two instances - Application Enabled -> lb Working Fine but unexpected behavior on instance failover
Test Scenario: A load tester is sending http request to lb synchronously at the rate of 5 requests per second.
When the load starts lb creates a session with instance 1, sends requests to it, receives 200OK sends back to cleint.
Now to test failover we deleted complete application from the server.
Result : Failover did not happen for the ongoing session, the client started receiving 404,
When I restarted the test the first http request was still transferred to instance one but all following http requests are transferred to instance2
This is consistent behavior
Questions
1. Why were the requests of ongoing session not handed over to second instance?
2. After restarting test why was even first request sent to the unhealthy instance ..what is healthcheaker for. (Yes it is enabled with url set to context root of the application)
Any kind of input will be highly appreciated
Appserver version: 8.2 (No I cant upgrade to 9.1)
[Message sent by forum member 'anzy' (anzy)]
http://forums.java.net/jive/thread.jspa?messageID=322583