Skip to content

Roadmap: put the Cloudflare bootstrap owner in charge and keep it currentΒ #21

Open
9 of 9 issues completed

Description

@devantler

πŸ€– Generated by the Agentic Engineer

Objective

The Cloudflare resources the platform needs before it can start β€” the zone, bootstrap DNS, backup storage and scoped service tokens β€” are owned by this repository, kept current, and recoverable without a running cluster.

Where it stands (verified 2026-11-02)

  • State foundation delivered: Verify independent remote state and lock recoveryΒ #55 is complete. The reviewed state migration proved native locking, a clean provider plan and recovery on a separate hosted runner. The resulting encrypted recovery package was merged and independently authenticated from main in chore(recovery): retain independently recovered backend stateΒ #60. The temporary migration input was removed.
  • Provider adoption remains: the protected state-bucket apply is delivered, but existing DNS, other storage and service identities have not yet been adopted into the general execution path. State migration did not change DNS or activate the alert journal.
  • Current declared provider: 5.28.1, with automated dependency proposals configured. This is the checked-in version, not a claim about the latest upstream release.

Remaining delivery

  1. Extend the protected execution path to import existing resources and review a no-destroy plan before provider changes, retaining independent state recovery.
  2. Declare and provision the alert journal's dedicated storage and service credential custody. Its conditional writes, concurrency, lost acknowledgements and restart behavior need native provider trials; infrastructure-state tests do not satisfy that acceptance.
  3. Complete reviewed DNS adoption and the Email migration through this repository, then exercise recovery of the resulting configuration and state from a fresh operator environment.

Success signal

The reviewed provider owner manages the required resources, every change has an accepted no-destroy plan, dependent services pass their real user paths, and a clean recovery drill is recorded.

Not in scope

In-cluster consumers of Cloudflare remain in devantler-tech/platform.

Activity

  1. added theissue type on Oct 20, 2026
  2. moved this to πŸ“₯ Backlog in 🌊 Project Boardon Oct 20, 2026
  3. added a commit that references this issue on Nov 3, 2026
  4. devantler commented on Nov 3, 2026

    @devantler
    ContributorAuthor

    πŸ€– Generated by the Agentic Engineer

    Consolidation decision for devantler-tech/monorepo#3274: this existing product roadmap owns the remaining Cloudflare adoption and recovery work. The external repository, ownership decision and portfolio registration already exist; keeping a second creation umbrella adds a second status surface without a separate product outcome. This roadmap remains open, and its state-foundation delivery does not establish general resource adoption.

    The retained acceptance scope from #3274 is:

    • Inventory the live zone, DNS, R2 buckets, lifecycle and object-lock settings, Origin CA resources, service identities, token scopes and platform consumers; identify each owner without publishing credentials.
    • Keep bootstrap state and execution independent of Kubernetes, Flux, Crossplane, OpenBao and cluster-hosted runners. Import existing production resources and accept a no-replacement/no-deletion plan before authoritative adoption.
    • Preserve least-privilege bootstrap/runtime identities, private credential custody, rotation, emergency recovery, drift detection and a fresh independent recovery drill. State-bucket delivery alone does not satisfy these gates for other resources.
    • Retain Isolate wedding database backups with a dedicated R2 identityΒ platform#2867's separate-identity R2 requirements. Keep in-cluster consumers and capability boundaries in Platform; remove manual dashboard steps only after their declarative owner covers them. Any post-bootstrap tenant ownership must avoid circular dependencies.
    • Keep the bootstrap-order ADR/operator runbook current and verify a clean drift plan after migration. The alert journal still needs its dedicated storage/custody and native conditional-write, concurrency, lost-acknowledgement and restart trials; DNS/Email adoption still needs its real acceptance and recovery evidence.

    This is a tracking consolidation only. No provider apply, credential change, inventory-completeness claim or recovery acceptance is implied. This Epic will stand independently as the portfolio's existing Cloudflare roadmap; the historical creation umbrella will close as not planned with its original criteria retained.

  5. devantler commented on Nov 15, 2026

    @devantler
    ContributorAuthor

    DNS is now managed here, and devantler.tech is served by the platform (2026-11-15).

    Done:

    • DNS adoption (step 1, and the DNS half of step 3), in feat(dns): adopt the zone's DNS records through a guarded plan and applyΒ #75:
      • The new main-only DNS plan and guarded apply workflow imported the zone's 14 owned records with no changes (run 43674749476). That covers the site entry points, the company Email records (iCloud MX, SPF, DMARC, DKIM, Apple) and the domain verification records.
      • Records the platform's external-dns owns stay with it, the SES bounce mail records included. No record in the zone is left unmanaged.
    • Safety checks: the workflow refuses deletes, replacements, changing imports and writes to records external-dns owns. It also refuses ownership records that would hand over a record nobody chose: a marker may only claim records the same change forgets, or ones named in the reviewed .github/dns-handover.json. It refuses the typeless _externaldns.devantler.tech outright.
    • Recovery: every apply keeps an encrypted state copy from before and after. A fresh runner re-initialises from R2 and must plan no changes, and the next run is refused until the after copy is retained here.
    • www β†’ platform (feat(dns): hand www to the platform's external-dns without a gapΒ #76, run 43675268687): an ownership record let external-dns replace the www CNAME with the gateway addresses in one Cloudflare transaction, with no gap. external-dns now owns www.
    • Apex β†’ platform (feat(dns): move the apex to the platform and restore external-dns ownership of wwwΒ #78 and fix(dns): leave the restored www ownership records to external-dnsΒ #79, run 43679002478): the apex CNAME was edited in place from devantler.github.io to www.devantler.tech (proxied), so the apex follows www and the gateway. Mail and every other record are unchanged.
    • No manual dashboard step was used.
    • Recovery copies: each generation was authenticated with the independent identity (serials 3 β†’ 4 β†’ 5 β†’ 6, same lineage). chore(recovery): retain the state after the apex applyΒ #80 retains serial 6.

    Remaining on this roadmap:

    1. Other resources: adopt the other R2 buckets, lifecycle rules and service tokens through a guarded path. Only the zone, the state bucket and DNS are managed today.
    2. Alert journal: its storage and custody (unchanged).
    3. Recovery drill: run one from a fresh operator environment, beyond the per-apply fresh-runner recovery.

    Rollback for the apex: an in-place edit back to devantler.github.io with proxying off, while GitHub Pages keeps the custom domain.

    πŸ€– Generated with Claude Code

  6. devantler commented on Nov 15, 2026

    @devantler
    ContributorAuthor

    Zone settings that rewrote the proxied site are now managed here too (2026-11-15).

    πŸ€– Generated with Claude Code

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    Fields

    No fields configured for Epic.

    Projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions