The Order I Use to Narrow Deployment Bugs

Deployment problems feel large when viewed loosely, but most narrow down to startup failure, missing configuration, auth failure, or routing.

Fixed order

1. Tail logs and check whether the process started.

2. Call the health endpoint directly.

3. Reproduce an authenticated request with the smallest curl command.

4. Compare local and deployed configuration.

Writing this order down prevents repeating the same reasoning during the next incident.