Upgrade and Migration
Use this page when you need the stack-level upgrade and migration model: sequencing, rollback posture, and the shared verification steps that sit above any single product.
When to use this page
- You are planning a stack-wide or multi-product change.
- You need the release sequencing model before following a product-specific upgrade page.
- You need rollback and post-check posture in one place.
Prerequisites
- You have read Compatibility and relevant release notes.
- You know whether the change is stack-wide or primarily product-specific.
Upgrade model
Safe upgrades should be:
- compatibility-aware
- release-note-informed
- verified before and after change
- explicit about rollback boundaries
Architecture / Flow
Step-by-step procedure
1. Start with compatibility and release notes
Use:
2. Define the sequence clearly
Important questions:
- which product or docs layer moves first
- where shared-truth publication changes
- what the rollback boundary is if verification fails
2a. Keep the OpenTofu migration fail-closed
Current Helpifyr OpenTofu migration truth is still owner-driven and staged:
JaddaHelpifyr/jhf-deploymentowns the migration program and deployment-side runtime truthJaddaHelpifyr/helpifyr-fabricowns only the shared alignment boundary, not runtime execution- this public docs page is guidance, not migration completion evidence
Until the owner repos publish verified implementation evidence, treat the migration posture as planned transition work rather than a completed platform default.
3. Keep Platform Truth in the loop
Use shared-truth and readiness surfaces as the contract boundary during rollout:
GET /api/v1/platform/version-truth
GET /api/v1/contracts/drift
GET /api/v1/observability/readiness
GET /api/v1/recovery/readiness
4. Verify after change, not only before it
Post-upgrade checks should include:
- shared-truth readback
- readiness readback
- one representative user or operator path
Verification
This page is being applied correctly when:
- upgrade sequencing is explicit
- rollback posture is explicit
- post-change verification is part of the plan rather than an afterthought
Common failure modes
Treating an upgrade as only a version bump
Problem:
- shared-truth and operational impact are under-scoped.
Better path:
- include compatibility, sequencing, rollback, and verification together
Planning rollback without defining the boundary
Problem:
- the team knows it should rollback but not what state can be safely reversed.
Better path:
- state the rollback boundary before execution