This seems to be a weird panic that I think should be avoidable. Currently our service just resets itself, but this should be investigated.