Run Chaos Experiment

Inject controlled failures to test and improve system resilience.

When to Use

• Before major product launches (load testing)
• After architecture changes (validate resilience)
• During GameDays or disaster recovery drills
• To validate assumptions about failure modes
• As part of SRE maturity program

Inputs

• Required: Kubernetes cluster (for Litmus or Chaos Mesh)
• Required: Steady-state definition (what "normal" looks like)
• Required: Hypothesis to test (e.g., "API stays available if one pod crashes")
• Optional: Observability stack (Prometheus, Grafana) to measure impact
• Optional: Rollback plan