Why Next💻 Dev20 min
We Got an Alert That the Database Was Down. The Database Was Never Down.
Four CRITICAL alerts piled up in a single day. A 504, a Prisma P2028, an admin 500, and 'DATABASE service down.' I opened the RDS metrics first: 21 straight hours with CPU peaking at 19.7%, perfectly healthy. What was slow was not the database but a round trip that crossed the Atlantic twenty-five times, and the 'down' alert was something the health check had manufactured all by itself.
- #Backend
- #Incident Response
- #Prisma
- #PostgreSQL
- #Observability