Spent 11 hours last week chasing a webhook retry loop that turned out to be one wrong status code
Our billing sync started double-charging a small handful of accounts, maybe 6 out of 400, and I figured it was a race condition in our job queue. I went down that path for most of Tuesday and Wednesday, added logging everywhere, even rebuilt part of the queue config. Turned out a downstream API was returning 200 on a partial failure, so our retry logic kept firing because it never saw an error. One line fix. The logging I added is still running in prod and honestly I might keep it. Has anyone else gotten burned by a vendor sending a 200 when they really meant 500?