We found out that the WLC treats IP helper/relay addresses quite different than for example a switch does. We tested that with DCHP and two IP helper addresses (one local and one remote) as we’d like to set up a redundant DHCP environment.
With a switch IOS the DHCP broadcast of the client is converted to a unicast and sent out to both relay addresses simultaneously. Thus, the one that replies first (i.e. the one with the scope activated) will actually assign the client’s IP address.
On the WLC it is like this: The broadcast is sent out to the first configured relay address. If the server is available but does not answer to the DHCP request (since the scope is currently not active) the request times out. The WLC will not sent out the uncast request to the second relay address unless the first server is not reachable at all. Hence, if you have a problem with the scope on the first DHCP server you will have to reconfigure the IP helper addresses on the WLC.
Can you confirm this behavior? What is the recommended way to mitigate this issue on the WLC?