I can partly reproduce this problem on my machine - the test passes or fails randomly.
Increasing the TIME_OUT like Stephane suggested helps, but doesn't really solve the problem that we don't have control over the time a spawned process takes to be started and terminated. Stefan
