FWIW: The "unstable nodes" weren't the reason for the unstakes... the 0 communication about any potential problems caused the unstakes (in my case anyway). When the system(s) stopped running, I hung around for some time, hoping that there would be ANY sort of communication... but alas.