Clock sync issues also sucked, we had a recurring issue which would cause the cluster to become unstable due to clock syncs and support would throw it in the to hard basket.