Prod Went Down. Everyone Says Check the Logs. In Which Order?

Prod went down.

Everyone says, “Check the logs.”

But in which order?

Here’s the route I take 👇

1. Did the request even reach my service?

• No → Not my bug yet. Go upstream: gateway, pod health, did the pod restart?
• Yes → Now it’s mine. Open the service logs.

2. Walk the request, don’t read everything

Inside the service, I follow the request instead of reading every log line:

endpoint → downstream service call → DB call → external call → queue publish

Wherever the log trail goes quiet, that’s where to dig deep.

3. Measure the impact before touching anything

• How many requests are failing?
• How many users are affected?
• Which tables or records got corrupted?

4. Replicate it locally and attach a debugger

If I can’t replicate it, I haven’t understood it. A fix written on top of a guess is just a second bug I’ll be debugging later. ✅

Logs never tell you the error. They show you the path to the bug.