Runtime health, ledger, queue and recovery

What it covers

• Runtime health combines units, workers, integrity, pending operations, queues, locks, incidents and readiness findings.
• The operation ledger retains request, actor, target, plan, phase and result; the queue serialises conflicting applies.
• Domain locks include software, apply and firewall transactions; stale locks are released only through recovery.
• Watchdog/supervisor limit restart loops; repeated failure becomes a finding.
• The recovery coordinator proposes resume, rollback, mark-for-review or controlled lock release.

How to use it

1. Export diagnostic state.
2. Check ledger/events/lock.
3. Run the coordinator plan.
4. Verify health/audit.

Important rules

• A green service is insufficient with an incomplete journal.
• Recovery does not delete history.
• Repeated restart stops for human review.