Harness Certified Chaos Engineering flashcards
Harness Certified Chaos Engineering · Build recall with a focused review session.
32 cards. Read the term and the definition here. The deck at the top of this page is optional.
- Steady-state hypothesis
- A testable statement that defines normal system behavior using measurable metrics and acceptable ranges, used to determine whether a fault caused unacceptable deviation.
- Blast radius
- The scope of potential impact from a chaos experiment, including affected services, users, and infrastructure; controlled via target selectors and duration limits.
- Chaos experiment loop
- The iterative process of defining steady state, injecting a fault, observing deviations, learning from results, and improving system resilience or the hypothesis.
- Game day
- A planned exercise where teams inject faults and practice incident response, validating both system resilience and human/process readiness under controlled conditions.
- Fault injection
- The deliberate introduction of failures such as latency, packet loss, resource stress, or pod kills into a system to observe its behavior and validate resilience.
- Network latency fault
- A fault that adds delay to network packets between services, testing timeout handling, retry logic, and user-perceived performance under slow conditions.
- Pod-kill fault
- A state fault that deletes a Kubernetes pod, testing restart policies, replica set rescheduling, and service availability during pod loss.
- Experiment abort condition
- A predefined threshold or probe failure that automatically halts a chaos experiment to prevent unacceptable impact, enforcing safety during fault injection.
- ChaosEngine resource
- A Harness Chaos Engineering Kubernetes custom resource that defines an experiment run, including target selectors, probes, and engine state.
- ChaosHub
- A Harness repository of reusable chaos experiment templates and fault definitions that can be customized and versioned for different environments.
- Resilience probe
- A Harness chaos component that validates steady state by querying metrics, endpoints, or commands before, during, and after fault injection.
- Target selector
- A label, namespace, or kind-based filter in a chaos experiment that scopes which pods or nodes receive the injected fault, controlling blast radius.
- Chaos pipeline stage
- A Harness CI/CD pipeline stage that runs a chaos experiment, allowing resilience validation to gate deployments automatically.
- Experiment scheduling
- Cron-based recurring execution of chaos experiments in Harness, enabling continuous resilience validation rather than one-off tests.
- Observability triad
- Metrics, logs, and traces correlated by experiment ID and target labels, providing comprehensive visibility into fault impact across services.
- Distributed tracing
- A technique that follows requests across service boundaries, revealing how faults propagate and which downstream services are affected.
- Prometheus probe
- A Harness chaos probe that queries Prometheus metrics to validate steady-state thresholds during an experiment, enabling automated pass/fail decisions.
- Alert suppression
- Configuring monitoring to ignore experiment-induced anomalies using maintenance windows or label filters, preventing false incident escalations.
- Log correlation
- Injecting experiment metadata into application logs so fault-induced errors can be filtered, grouped, and analyzed alongside other observability data.
- Baseline period
- A pre-experiment observation window that captures normal metric variance, used to distinguish fault impact from natural fluctuations.
- Recovery validation
- Post-injection verification that the system returns to baseline metrics within an expected time, confirming resilience and absence of lingering degradation.
- Composite steady-state hypothesis
- A hypothesis combining multiple metrics such as availability, latency, and correctness, providing broader coverage than a single-metric assertion.
- Probe failure action
- The configured behavior when a steady-state probe fails: halt the experiment, mark it failed, or continue with a warning, based on risk tolerance.
- Continuous validation
- Running steady-state probes throughout fault injection, not just before and after, to detect transient deviations and threshold breaches in real time.
- Target mode
- A Harness ChaosEngine setting (one, all, fixed, percent) that determines how many pods or nodes receive the fault, directly limiting blast radius.
- Canary experiment
- A chaos run scoped to a single canary instance or small subset of replicas, used to observe impact before expanding to the full target set.
- Experiment timeout
- A time limit that automatically stops a chaos experiment, preventing indefinite fault injection and limiting blast radius over time.
- Kill switch
- A manual or automated mechanism to immediately stop a chaos experiment, such as setting ChaosEngine engineState to stop.
- Dependency mapping
- Identifying upstream and downstream services affected by a chaos target, ensuring validation covers the full blast radius of potential impact.
- Resource quota
- A Kubernetes limit on CPU, memory, or object counts that prevents a chaos experiment from consuming cluster-wide resources and expanding blast radius.
- Namespace isolation
- Scoping chaos experiments to a specific Kubernetes namespace to contain faults and prevent impact on unrelated workloads in other namespaces.
- Chaos Service Account
- A Kubernetes service account referenced in the ChaosEngine spec that grants the chaos runner permissions to inject faults on target resources.