Skip to content

Give the DR rebuild a signed-artifact publication path #2839

Description

@devantler

🤖 Generated by the Agentic Engineer

Problem

The DR rebuild (dr-rebuild.yaml) publishes to the same mutable tag as a normal prod deploy — ghcr.io/devantler-tech/platform/manifests:latest — but it never signs what it pushes, and its rebuild job holds no id-token: write, so keyless signing is not even possible.

That makes it a hard blocker on the parent: root-source signature verification cannot be switched on while one writer to the tag publishes unsigned bytes under an identity the allow-list does not name. The first recovery after enabling verification would produce an artifact Flux refuses — during an incident, when manual repair is least available.

Live evidence (2026-07-28)

The posture is currently inverted — every tenant source is verified, the infrastructure root is not:

Source spec.verify
flux-system/flux-system (root) null
ascoachingogvaner, aws, doggy-countdown, github-config, wedding-app cosign

dr-rebuild.yaml job rebuild declares only contents: read + packages: write, and the file contains no cosign invocation at all.

Scope

Give DR a signed-artifact path, leaving verification OFF:

  • id-token: write on the rebuild job.
  • Install cosign and sign the digest DR already resolves (platform_manifest), not the mutable tag.
  • Name the DR workflow identity in ksail.prod.yaml matchOIDCIdentity, pinned to refs/heads/main.
  • A CI-enforced contract so none of the four can silently regress.

Why this is safe to land first

Signing an artifact that nothing yet verifies is inert — it changes no reconcile behaviour. It is a strict prerequisite for the parent, and it is the only part that can ship without first making publication atomic.

Acceptance criteria

  • The DR rebuild signs the artifact it publishes, using the digest it resolved.
  • The DR signing identity is in the cosign allow-list, pinned to refs/heads/main.
  • CI fails if the permission, the sign step, the digest pinning, or the allow-list entry is removed.
  • No change to production reconcile behaviour (verification remains off).

Part of #2627

Activity

  1. coderabbitai commented on Jul 28, 2026

    @coderabbitai
    Contributor
    🔗 Related PRs

    #2319 - feat(security): cosign-verify the infrastructure Flux OCI artifact [merged]
    #2404 - fix: tenant artifact verification rejects the dual signing identities from #2399 [merged]
    #2457 - fix(talos): accept the actions publish-app identity at node image verification [merged]
    #2625 - fix(deploy): synchronize production GHCR pull auth [merged]
    #2633 - fix(deploy): verify incoming image on Talos nodes [merged]


    📝 Issue Planner

    Check the box below or use the @coderabbitai plan command to generate an implementation plan and prompts that you can use with your favorite coding assistant.

    • Create Plan

    🧪 Issue enrichment is currently in open beta.

    You can configure auto-planning by selecting labels in the issue_enrichment configuration.

    To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

    issue_enrichment:
      auto_enrich:
        enabled: false

    💬 Have feedback or questions? Drop into our discord!

  2. moved this to 🏃🏻‍♂️ In Progress in 🌊 Project Boardon Jul 28, 2026
  3. moved this from 🏃🏻‍♂️ In Progress to 👀 In Review in 🌊 Project Boardon Jul 28, 2026
  4. moved this from 👀 In Review to 🚀 Ready to Merge in 🌊 Project Boardon Jul 28, 2026
  5. moved this from 🚀 Ready to Merge to ✅ Done in 🌊 Project Boardon Jul 28, 2026
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

    Fields

    No fields configured for Security.

    Projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions