that is correct... the set is asked to ping the other mbg's ip address with an 800ms timeout and if mbg doesn't get a response back in a reasonable time, we give the set 10 seconds to respond, then the set is "stuck"... like I said, it's a bug that was fixed.. enabling the other option "reboot_stuck_set" does clear the problem, at the cost of forcing the set through a reboot cycle.
note that mbg doesn't ask a set to ping/or redirect if it's "busy"... ie. the set has to not be in a call, or been in a call, had a digit, handset or tone (ringing) operation performed in the past 15 seconds.