Skip to content

Appendix E – Chaos Scenario Definitions

E.1. General Principle

Chaos testing scenarios (sections 5.10–5.11) are no longer a static list in the document. They are stored in machine‑readable format (YAML/JSON) as a signed artifact in IPFS. Each scenario has a weight for computing the ResilienceScore (section 5.10.2) and is linked to specific modules or phases. This appendix contains: - a reference to the current scenarios artifact (CID); - description of the structure and examples; - table of scenarios with their parameters and weights; - instructions for adding new scenarios and validation.

E.2. Current Scenarios Artifact

Field Value
CID (IPFS) QmChaosScenariosV3
BLAKE3 hash d1e2f3a4b5c6d7e8f9a0b1c2d3e4f5a6b7c8d9e0f1a2b3c4d5e6f7a8b9c0
File name chaos_scenarios.yaml
Schema version 2.0
Signature ed25519:4b3c5d6e…

Download and verification:

ipfs get QmChaosScenariosV3 -o chaos_scenarios.yaml
sha256sum chaos_scenarios.yaml
# Expected output: d1e2f3a4b5c6… chaos_scenarios.yaml

E.3. Scenario File Structure

The file chaos_scenarios.yaml has the following structure:

version: "2.0"
description: "Chaos engineering scenarios for Black Swan Phase 1-3"
last_updated: "2026-04-20T12:00:00Z"
default_resilience_weight: 0.1
scenarios:
  - id: "network_delay"
    type: "network"
    description: "Introduce fixed delay and jitter"
    params:
      delay_ms: 500
      jitter_ms: 100
    resilience_weight: 0.15
    applicable_phases: [1, 2, 3]
    expected_behavior: "degraded_performance"
    severity: "medium"
  - id: "packet_loss"
    type: "network"
    description: "Emulate packet loss"
    params:
      loss_percent: 10
    resilience_weight: 0.15
    applicable_phases: [1, 2, 3]
    expected_behavior: "retry_with_backoff"
    severity: "medium"

E.4. Scenario Table

The following is a summary of all scenarios included in the current artifact QmChaosScenariosV3.

ID Type Parameters Weight Phases Severity
network_delay network delay_ms=500, jitter_ms=100 0.15 1–3 medium
packet_loss network loss_percent=10 0.15 1–3 medium
cpu_throttle compute cpu_quota=0.5 0.15 1–3 medium
memory_pressure memory memory_limit_mb=512 0.15 1–3 medium
random_kill fault probability=0.2 0.20 2–3 high
byzantine_behavior byzantine corruption_probability=0.1 0.25 3 high
slow_loris dos connections=50, keepalive_interval=10 0.10 2–3 medium
clock_skew time offset_seconds=3600 0.10 2–3 low
escape_attempt security target_process=systemd, payload=harmless 0.30 0–3 critical

E.4.1. Notes on Scenarios

  • escape_attempt – emulates a fileless injection attempt. Weight 0.30 reflects the critical importance of isolation. If this scenario fails, the entire patch is rejected regardless of other metrics.
  • byzantine_behavior – relevant only for Phase 3 (distributed swarm). Emulates compromise of an edge node.
  • random_kill – applied only in the shadow environment; in production it is replaced with a controlled shutdown.

E.5. Resilience Score Computation

The final ResilienceScore (section 5.10.2) is computed as the weighted geometric mean of the results of individual scenarios:

ResilienceScore = (Π_{s in scenarios} score_s^{w_s})^{1 / Σ w_s}

Where: - score_s – result of scenario s (0..1), - w_s – resilience_weight from the scenario configuration.

Scenarios with severity: critical (escape_attempt) zero out the overall ResilienceScore if score_s = 0.

E.6. Adding New Scenarios

  1. Create a YAML description of the new scenario according to the schema (JSON Schema available at CID QmChaosScenarioSchemaV1).
  2. Test the scenario locally in an isolated shadow environment.
  3. Add the scenario to chaos_scenarios.yaml.
  4. Sign and publish the new version: bash sign_artifact chaos_scenarios.yaml --key /etc/swarm/keys/node_key.pem ipfs add chaos_scenarios.yaml
  5. Update the CID in GlobalState.security_state.chaos_scenarios_cid.

E.7. Integration with the Validation Pipeline

Scenarios are loaded by ShadowBenchmark (Appendix C) when launching shadow testing:

def load_chaos_scenarios():
    cid = GlobalState.security_state.chaos_scenarios_cid
    scenarios_yaml = ipfs.cat(cid)
    return yaml.safe_load(scenarios_yaml)['scenarios']

The results of each scenario are stored in the ShadowBenchmarkArtifact and influence the decision on patch deployment.

E.8. Relationship with Other Sections

  • 5.10 Chaos Engineering in Shadow Environment – logic of fault injection.
  • 5.10.2 Quantitative Resilience Scoring – ResilienceScore formula.
  • 5.11 Extended Chaos Scenarios for Distributed Components – additional scenarios for the swarm.
  • 9.2 Safety & Isolation Boundaries – escape_attempt as an isolation check.

E.9. Change History

Artifact version Date Changes CID
V1 2026-01-15 Basic scenarios (network, cpu, memory) QmChaosScenariosV1
V2 2026-03-10 Added byzantine, slow_loris, clock_skew QmChaosScenariosV2
V3 (current) 2026-04-20 Introduced resilience weights, added escape_attempt QmChaosScenariosV3