Close the deployment record with observable outcomes
What This Concept Is
A release record joins authorization, artifact identity, environment, migration, observation and final status. It closes when the service is verified or explicitly marked unresolved.
Why It Matters Here
A green pipeline can leave an inaccessible service, leaked preview environment or unknown cleanup state.
Concrete Example
The build passed, but a smoke request returns forbidden because deployment identity accidentally became runtime identity. Record the failure and distinguish permissions needed to deploy from permissions needed to serve. Repair the narrow boundary and rerun allowed/denied checks. After teardown, inspect retained resources rather than relying on command exit alone.
Common Confusion / Misconception
Deployment success is not user success. An absent listing caused by denied inventory access does not prove teardown. A local reproduction must not be described as a cloud deployment.
How To Use It
Use the original trusted pipeline and cost/teardown inventory. Record scoped identity, artifact digest, user-path checks, rollback decision, retained items and owner. Preserve local/cloud lane distinctions and original mandatory deployment requirements.
Check Yourself
Who can promote? Which identity serves traffic? What proves resource absence? Which claim depends on a real cloud run?
Mini Drill or Application
Produce one failed-release record and its repaired successor. Explain what evidence changed the release status, rather than merely restating commands.
Read This Only If Stuck
Building Secure and Reliable Systems provides selective operational reasoning. Use the module sources below for current tool behavior; do not read a whole book before attempting the exercise.