if you've set up 3300 resiliency/clustering correctly, then it should just work.
but make sure you have the ip trunking networking forms on the 3300 set up so they know each other's ip addresses, otherwise they cannot tell the set (and MBG) where to go in case of a failure.
for a set on the lan, they can "cheat" a little bit even if the 3300's are not set up correctly. They can also get resiliency addresses from DHCP and even if the 3300 isn't programmed correctly, they can find the secondary.
A way to test if this kind of misconfiguration is in place, use the '7' key to program a LAN set as a "teleworker", but point it directly at the primary 3300. Then it will not use the DHCP resiliency info. Perform a primary 3300 failure and see if it fails over to the secondary. If it doesn't, then look to the 3300 cluster programming (sorry I can't help you much there).
(you can use the '7' again to remove the teleworker programming on the set afterwards).
there's also a way to see in tug.log if the 3300 is sending the correct 'redirect' information.