Check this thread out, it may apply to your problem.
My past issues like this were "solved" when I switched to a different cluster with a tach. (upgrade I wanted anyways). This newer info. is worth checking out, haven't tried it on the old cluster yet as I just noticed this recently.