provides a very simple load balancing.
It works very well for light workloads (1 req per second) but for a sustained load (300 req/s, generated by httperf) after a few seconds, the output is locked on one of the target address, the other one never being selected.
Worse: after that, even after a delete of the rule, followed by a redefinition of the rule the bad behaviour is not corrected. A reboot is needed to get back the feature...
I have not tryed to debug this, would you have some hints ? Bye.