registrystack / registrystack/registry-stack

Support authenticated replacement-instance bootstrap

Open
#601 1 comment 0 reactions 0 assignees View on GitHub
area:platform needs-jeremi-decision
Dominant language
Rust
Stars
2
Forks
0
Avg merge
2h 57m
Merged PRs (30d)
128

Description

## Outcome

An operator can restore or replace a single-node Registry Stack instance from approved backup history without weakening anti-rollback protection, reusing an old instance identity incorrectly, or requiring an ad hoc signing ceremony during an outage.

This complements #203 and does not expand the 1.0 adopter-runtime PR. High-availability and multi-replica recovery remain in #507.

## Design checkpoint

Define the smallest authenticated proof that lets a fresh instance establish:

- the intended product lane and replacement instance identity;
- the accepted trust anchor and authenticated anchor history;
- the latest approved bundle sequence and hash;
- the matching database, audit, and correctness-state snapshot;
- the operator authorization for replacement rather than rollback or cloning.

## Security invariants

- A stale, partial, cross-lane, or cross-instance backup cannot become current.
- Replacement cannot use AcceptUnsigned, AcceptRollback, or another silent break-glass path.
- Trust anchors are pinned independently of the closure they authorize.
- Audit evidence records the old identity, new identity, approved history, and recovery decision without secret values.
- Product-owned exact verification completes before traffic admission.

## Definition of Done

- [ ] A versioned replacement-bootstrap artifact and ownership model are documented.
- [ ] Relay and Notary authenticate and consume it through closed product-owned actions.
- [ ] Backup inventory and #203 recovery procedures include every required authority and durable artifact.
- [ ] Tests cover valid replacement plus stale sequence, wrong lane, wrong instance, anchor swap, partial snapshot, replay, and mixed-generation rejection.
- [ ] A clean single-node restore exercise reaches exact verification before traffic.

## Non-goals

- Automatic disaster-recovery orchestration.
- Multi-replica fencing or failover.
- Bypassing normal signed approval and anchor-rotation procedures.

Contributor guide

Open the contributing guide

Research direction

Begin with the #203 recovery procedures and the Relay and Notary paths that must authenticate and consume the replacement-bootstrap artifact. Trace the required backup authorities and durable artifacts, then use the listed rejection cases and clean single-node restore exercise to verify that exact verification completes before traffic admission.

Written by the indexing model from the issue text.

Assessment

Tech stack
rust
Domain
authentication, authorization, backend, databases, security
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Active
Clarity
Mostly clear
Newbie friendliness
32/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.