At my first DevOps job we had defined a function which deleted old partitions from time to time. If you invoked it manually because of some automation failure w

At my first DevOps job we had defined a function which deleted old partitions from time to time. If you invoked it manually because of some automation failure with the wrong arguments, it had a habit of nuking production data, which, needless to say, happened at least once. Naturally, the function was called footgun. We can pretend that the proliferation of managed databases, newfangled NoSQL datastores, and other abstractions preclude the need for accumulated empirical observation across a range of tech stacks and time, but sometimes there's really no substitute for greybeard wisdom.