Files
git.stella-ops.org/docs/airgap/staleness-and-time.md
2025-12-24 16:26:06 +02:00

3.2 KiB
Raw Blame History

Air-Gapped Time Anchors & Staleness Budgets

Audience: AirGap Time/Controller/Policy guilds, DevOps

Purpose: Document how air-gapped installations maintain trusted time anchors, compute staleness windows, and expose drift telemetry.

1. Overview

Air-gapped clusters cannot contact external NTP servers. StellaOps distributes signed time anchor tokens alongside mirror bundles so services can reason about freshness and seal state without external clocks.

Key goals:

  • Provide deterministic time anchors signed by the mirror authority.
  • Track drift and staleness budgets for scanner reports, advisories, and runtime evidence.
  • Surface warnings to operators (UI/CLI/Notifier) before anchors expire.

2. Components

Component Responsibility
AirGap Controller Stores the active time_anchor token and enforces sealed/unsealed transitions.
AirGap Time service Parses anchor bundles, validates signatures, records monotonic offsets, and exposes drift metrics.
Scheduler & Policy Engine Query the time service to gate scheduled runs and evidence evaluation.
UI / Notifier Display remaining budget and raise alerts when thresholds are crossed.

3. Time Anchor Tokens

  • Distributed as part of mirror/offline bundles (airgap/time-anchor.json).
  • Signed with mirror key; includes issuance time, validity window, and monotonic counter.
  • Validation steps:
    1. Verify detached signature.
    2. Compare bundle counter to previously applied anchors.
    3. Persist anchor with checksum for audit.

4. Staleness Budgets

Each tenant/configuration defines budgets:

  • Advisory freshness maximum age of advisory/VEX data before rescans are required.
  • Scanner evidence acceptable drift between last scan and current anchor.
  • Runtime posture tolerated drift before Notifier raises incidents.

AirGap Time calculates drift = now(monotonic) - anchor.issued_at and exposes:

  • /api/v1/time/status current anchor metadata, drift, remaining budget.
  • /api/v1/time/metrics Prometheus counters (airgap_anchor_drift_seconds, airgap_anchor_expiry_seconds).

5. Operator Workflow

  1. Import new mirror bundle (includes time anchor).
  2. AirGap Time validates and stores the anchor; Controller records audit entry.
  3. Services subscribe to change events and recompute drift.
  4. UI displays badge (green/amber/red) based on thresholds.
  5. Notifier sends alerts when drift exceeds warning or expiry limits.

6. Implementation Notes

  • Use IAirGapTimeStore for persistence; default implementation relies on PostgreSQL with tenant partitioning.
  • Ensure deterministic JSON serialization (UTC ISO-8601 timestamps, sorted keys).
  • Test vectors located under src/AirGap/StellaOps.AirGap.Time/fixtures/.
  • For offline testing, simulate monotonic clock via ITestClock to avoid system clock drift in CI.
  • Staleness calculations use StalenessCalculator + StalenessBudget/StalenessEvaluation (see src/AirGap/StellaOps.AirGap.Time/Services and .Models); warning/breach thresholds must be non-negative and warning ≤ breach.

7. References

  • docs/airgap/airgap-mode.md
  • src/AirGap/StellaOps.AirGap.Time
  • src/AirGap/StellaOps.AirGap.Controller
  • src/AirGap/StellaOps.AirGap.Policy