Move_base is stuck while status is ACTIVE

edit

Hello,

I am driving around my robot and testing the move_base node. It moves around fairly well as long as I order goal positions that the robot can reach. The problem is:

when I order a goal position that the robot can not reach (due to a small door for example) the status of that goal remains ACTIVE forever but the robot does not move. If I order a new goal, the status of the new goal is PENDING but it never becomes ACTIVE because the first order is still ACTIVE (and will be like that forever). If I keep sending new goal positions, the old ones which are PENDING are discarded, and the new ones are introduced as PENDING as well.

How should I proceed in order to regain control over my robot? Thanks for your tips and comments!

EDIT-1 (answer to 1st comment): the way I communicate with the move_base node is by pusblishing and subscribing to its topics:
/move_base/status provides me with the status of every goal
/move_base_simple/goal accepts goal positions
In order to learn how the navigation stack works I followed this tutorial (it was months ago, now parameters are tunned to fit my needs): http://www.ist.tugraz.at/robotics/bin/view/Main/Poineer_pathplanning

EDIT-2: I checked the log of move_base node and this is what happens:
( WAR] (1358262472.775372377, 81184.500000000]: Clearing costmap to unstuck robot.
( WARN] (1358262472.879543309, 81184.600000000]: Rotate recovery behavior started.
(but not recovery behaviour is started)

Moreover, If I request to cencel all the current goals of move_base (publishing a cancel with all goal_ids), then the status of the stuck goal changes from ACTIVE to PREEMPTING and eventually to PREEMPTED. Any ideas?