Hope you find out the root cause and don't have to start swapping out servers! We don't have hundreds of clusters but 20 or so and only one node on one cluster had the problem