Scenarios — Real Incidents & Interview Discussions
Six incidents told end to end: the symptom, the investigation, the mechanism underneath, and how to discuss each one in an interview.
Scenario: The 2GB Hello-World
A small Node service produced a 2.2GB image and a twelve-minute pipeline. The investigation, the fix, and what interviewers really test here.
Scenario: The Database That Forgot Everything
A volume was declared, attached, and visible in docker inspect. The data still vanished. Why storage being present isn't the same as storage being used.
Scenario: The Deploy That Always Took Ten Seconds
A suspiciously round number is a diagnosis. Why /bin/sh at PID 1 swallows SIGTERM, and how it turns every rolling deploy into dropped requests.
Scenario: The Token That Shipped
A generated Dockerfile built cleanly, passed review, and published a credential to every host that pulled it. The failure with no symptom at all.
Scenario: exec format error
One opaque line killed a service in production while it ran perfectly on every developer laptop. The ARM-versus-x86 mismatch and how manifest lists fix it.
Scenario: The Stack That Only Failed in CI
Passing locally, failing one run in four. Why depends_on doesn't mean what it looks like, and how to tell a race condition from flakiness.