Skip to content

Proposal: Feature Recovery and Reconciliation Protocol #63

Description

@anoopath

Proposal

Define an opt-in Feature Recovery and Reconciliation Protocol for the Connection
Coordinator. The current protocol requires convergence on missing features but
does not specify the recovery owner, foreground-to-background handoff, or safe
handling of ambiguous operations and immutable feature replacement.

The proposal covers explicit per-channel ownership, reconciliation before retry,
bounded foreground attempts with durable background work, existing requestId
idempotency, authoritative feature failure information, and coordinated cleanup.
It preserves existing three-of-four provisioning eligibility and the full
verification handshake. No production implementation is proposed.

Related public discussion: #51 covers retries, reconciliation, and identity.
This proposal extends that discussion with ownership, lifecycle, compatibility,
and unresolved cross-provider fencing requirements; it does not assume #51 is
accepted.

Decisions Needed

  • Bilateral capability negotiation and generated-client enum compatibility.
  • Distributed generation/ownership fencing and safe handoff before automated
    replacement or cleanup can be enabled.
  • Retry budgets, replay semantics, retention, failure codes, and escalation.
  • Explicit approval of stronger safeguards for the existing seven-day cleanup
    permission and resolution of deferConnection/deferProvisioning naming.

These are review gates, not claims that distributed recovery races are solved.

Activity

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

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions