version: 1
id: chaos-experiment-basics
title: Plan a Chaos Experiment
summary: Write a payments-api chaos experiment with hypothesis, blast radius, abort criteria, and a single safe failure mode.
difficulty: intermediate
estimatedMinutes: 25
prerequisites: [burn-rate-alerts, kubernetes-deploy]
image: alpine:3.21
shell: /bin/sh
setup:
  - "mkdir -p /workspace/chaos /workspace/docs"
  - "printf 'Ticket: validate payments-api survives a single replica kill.\\nNeed hypothesis, blast radius, abort criteria, and experiment.yaml.\\n' > /workspace/chaos/TICKET.md"
tasks:
  - id: experiment-spec
    title: Author the experiment sketch
    description: 'Create /workspace/chaos/experiment.yaml with name payments-api-pod-kill, target Deployment payments-api, action kill-pod (or pod-kill), and steadyStateHypothesis mentioning availability or ready. Include blastRadius mentioning 1 replica or one pod.'
    hints:
      - 'Tip code ONE_FAILURE: start with a single-pod kill before cascading faults.'
      - 'Tip code STEADY_STATE: declare what healthy looks like before you inject failure.'
      - 'Include payments-api and kill-pod (or pod-kill).'
    checks:
      - type: file
        name: Experiment named
        path: /workspace/chaos/experiment.yaml
        value: payments-api-pod-kill
      - type: file
        name: Targets payments-api
        path: /workspace/chaos/experiment.yaml
        value: payments-api
      - type: command
        name: Action is pod kill
        command: "grep -Eqi 'kill-pod|pod-kill' /workspace/chaos/experiment.yaml"
      - type: command
        name: Steady state declared
        command: "grep -Eqi 'steadyState|steady.state|hypothesis' /workspace/chaos/experiment.yaml"
      - type: command
        name: Steady state mentions availability
        command: "grep -Eqi 'availability|ready' /workspace/chaos/experiment.yaml"
      - type: command
        name: Blast radius limited
        command: "grep -Eqi 'blastRadius|blast radius|1 replica|one pod' /workspace/chaos/experiment.yaml"
  - id: abort-and-hypothesis
    title: Document hypothesis and abort
    description: 'Write /workspace/docs/EXPERIMENT.md with headings Hypothesis, Blast radius, and Abort. Hypothesis must mention survive or recover. Blast radius must mention one pod or single replica. Abort must mention error budget or freeze or rollback. Write /workspace/chaos/STATUS.txt containing CHAOS_OK.'
    hints:
      - 'Tip code ABORT_BUDGET: abort when burn accelerates or freeze gate trips.'
      - 'STATUS.txt should include CHAOS_OK.'
      - 'Keep headings scannable for the game-day facilitator.'
    checks:
      - type: file
        name: Hypothesis section present
        path: /workspace/docs/EXPERIMENT.md
        value: Hypothesis
      - type: file
        name: Blast radius section present
        path: /workspace/docs/EXPERIMENT.md
        value: Blast radius
      - type: file
        name: Abort section present
        path: /workspace/docs/EXPERIMENT.md
        value: Abort
      - type: command
        name: Hypothesis mentions survival
        command: "grep -A3 '^#* *Hypothesis' /workspace/docs/EXPERIMENT.md | grep -Eqi 'survive|recover'"
      - type: command
        name: Blast radius is small
        command: "grep -A3 '^#* *Blast radius' /workspace/docs/EXPERIMENT.md | grep -Eqi 'one pod|single replica|1 replica'"
      - type: command
        name: Abort ties to budget
        command: "grep -A3 '^#* *Abort' /workspace/docs/EXPERIMENT.md | grep -Eqi 'error budget|freeze|rollback'"
      - type: file
        name: Chaos status ok
        path: /workspace/chaos/STATUS.txt
        value: CHAOS_OK
limits: {cpus: "0.5", memory: 128m, pids: 64, timeout: 1800, network: false}