Something got unhappy and died on my K8s cluster that runs this service. I was dreading figuring it out. Luckily, I discovered that past me  from a few years ago left some notes about what to do when this happens. YAY, PAST ME! Thanks for that.