Shoutout to the 2am pager at my last on-call job, one bad deploy taught me more than a year of runbooks
I was on call in Austin when a config push at 2am took down billing for 340 accounts, and after 40 minutes of panic we rolled back and found the fix was one env var someone forgot to set, so now I want to know who else has a war story about a rollout that broke prod and what actually stopped the bleeding.