Roll back failed edge-update boots without retry loops
A failing candidate is selected repeatedly on each restart. Persist failed-candidate disposition and restore the known-good slot.
- Focused work estimate
- 3h + prerequisites
- Priority in the scenario
- High
- Engineering practice
- Recovery · Failure handling
Estimated field mix
- Embedded and edge60%
- Site reliability40%
Field percentages are editorial estimates of the ticket's engineering focus. They total 100%; they are not measured time, proficiency scores, or ownership evidence.
Review it, then add it to your workspace.
The board opens an editable draft; nothing is saved until you confirm it. Sign-in and workspace permissions apply, and Demo boards remain ephemeral.
Project context
Fictional information kiosks receive small generated image blobs through a local update adapter. Simulate slots and boot outcomes in software; do not flash hardware or execute image contents.
Setup prerequisites
- Generate inert byte images with manifests and development-only signing fixtures.
- Implement simulated boot success failure and power-cut points.
Preceding work
Complete these dependencies, or supply their agreed outputs before taking this ticket.
- CUPDATE-101 · Parse edge-update manifests with explicit supported versions
- CUPDATE-102 · Verify edge-update image size and digest before staging
- CUPDATE-103 · Validate edge-update manifests using local trust fixtures
- CUPDATE-104 · Write edge-update images only into the inactive simulated slot
- CUPDATE-105 · Record edge-update activation intent before changing the boot selector
- CUPDATE-106 · Mark an edge-update candidate healthy only after simulated boot confirmation
Acceptance criteria
- Failed candidate does not auto-reactivate
- Prior slot is selected
- No valid fallback produces explicit recovery-required state
Implementation constraints
- Do not label fallback existence without validating its metadata.
Verification to include
- Fail candidate boot
- Invalidate both slot manifests
Deliverables
- Rollback coordinator and failure cases
Rollout and recovery
Stop automatic activation when no validated fallback exists.
Value of the work
For the engineer: Practice update state machines and recoverable activation.
For the team: Inspect whether an engineer can prevent partial rollout from stranding devices under declared conditions.
Evidence boundaries
Outcome Evidence: Tests, patches, and runbooks are requested deliverables. They become Outcome Evidence only through a qualified Mission and immutable Evidence IDs.
Ownership Evidence: Independent adaptation must be observed under a declared verification policy and cite immutable Evidence IDs. Completing a planning ticket establishes no Ownership Evidence.