Makes sense. I recollect some of the postmortem reports I read where live config changes triggered cascading outages.