Exercise 1: Why Verification Is Different From a Per-Step Expected Result — Possible Solution ==================================================================== WHAT A PER-STEP EXPECTED RESULT CONFIRMS ------------------------------ Per Chapter 3's own material, a per-step expected result confirms that one specific action, in isolation, produced the outcome it was supposed to - "Run X. Expected: Y." It answers "did this individual step work," nothing more. WHAT VERIFICATION CONFIRMS INSTEAD ------------------------------ Per this chapter, "verification is a separate, final check confirming the whole procedure achieved its actual goal - every step can complete without error while the underlying problem remains unresolved." Verification answers a different question entirely: "did the overall reason we ran this runbook in the first place actually get achieved." A CONCRETE EXAMPLE OF ALL PER-STEP CHECKS PASSING WHILE VERIFICATION FAILS ------------------------------ Using this chapter's own worked example: every individual step could succeed exactly as expected - the service status shows "active (running)" after the restart, with no error output anywhere. But if the actual underlying cause of the stuck worker wasn't the process itself needing a restart - say, a downstream dependency the worker relies on is what's actually broken - the queue depth might never actually drop, even though every individual step technically succeeded. The runbook's own verification step (confirming queue depth drops below 100 within 5 minutes) is what catches this; none of the per-step checks would have. WHY THIS DISTINCTION MATTERS PRACTICALLY ------------------------------ Without a separate verification step, a reader following the runbook could reasonably conclude the problem is fixed simply because nothing errored - and walk away from a still-broken system. Verification exists specifically to catch exactly this gap between "the steps ran successfully" and "the actual problem is resolved." WHY THIS WORKS AS AN ANSWER ------------------------------ It restates what a per-step expected result confirms versus what verification confirms, and gives a concrete example - a restart that succeeds technically but doesn't address the real underlying cause - showing how every per-step check can pass while the true goal still isn't achieved.