Radius load balancing not balancing

Latest: Unable to replicate problem with eapol_test on test system - load balanced perfectly even under quite heavy load. When I tried datagram LB and parstitence on avp it generally didn't load balance or broke altogether. So, yesterday during a maintenance period, I activated datagram LB and avp persistence profile (using attribute 1) and so far all seems good. The load is roughly equally balanced between the two radius servers (< 5% difference on bits) and has been for the last 24 hours. I agree with you however that we may still have a problem - probably with our wireless controllers as I don't think they should be sending varying numbers of requests. I suspect it may be load - we have considerably increased our number of access points over the past 6 months - or the number of separate devices in the chain (user device->wap->wireless controller->F5->radius server->AD controller). So, I think we need to look at that in more detail But it looks like the initial problem is resolved. Thanks very much for your advice and support