Helm deployment

The repository ships a Helm chart under helm/. It deploys a single Deployment running chaos_zookoo, mounts the scenarios as a ConfigMap, and optionally sidecars a Grafana Alloy exporter for OTLP-based metric shipping.

Install

helm install chaos-zookoo ./helm \
  --namespace chaos-system --create-namespace \
  --values my-values.yaml

For the full installation walkthrough (credentials, testkit env vars, extraSecrets), see Deploy with Helm.

Minimal values.yaml

k8s:
  host: "https://api.my-cluster.example.com"
  clusterCert: "<base64-CA>"
  token: "<bearer-token>"

configs:
  - kind: Killing
    name: kill-api
    namespace: api
    schedule:
      interval: 60s
    selector:
      labels:
        app: api
    minAvailable: 2

Each entry under configs is a scenario object — the chart serialises it and concatenates all documents with --- into /app/config.yaml.

Full values.yaml reference

image:
  repository: neryolab/chaos-zookoo
  tag: "0.1.0"

# ServiceAccount: create=true lets the chart generate the SA.
# The chart never creates a Role or RoleBinding — add them separately.
serviceAccount:
  create: false
  name: ""  # existing SA name; defaults to "default" when empty

# Kubernetes API credentials (stored in a Secret by the chart).
k8s:
  host: ""
  clusterCert: ""
  token: ""

# Secrets created in the release namespace (e.g. for GRAFANA_TOKEN).
extraSecrets: []

# Extra env vars injected into the container.
# Required for testkit: GRAFANA_URL and GRAFANA_TOKEN.
extraEnv: []

extraContainers: []
extraVolumes: []
extraVolumeMounts: []

resources:
  limits:   {cpu: 100m, memory: 128Mi}
  requests: {cpu: 100m, memory: 128Mi}

readinessProbe: {}
livenessProbe:  {}

monitoring:
  prometheus:
    port: 9090           # maps to METRICS_ADDR, exposed on the Service
  exporter:
    enabled: false       # flip on to sidecar grafana/alloy for OTLP
    scrapeInterval: 30s
    image:
      repository: grafana/alloy
      tag: latest
      pullPolicy: IfNotPresent
    endpoint:
      url: http://otel-collector:4317
      insecure: false
    resources:
      limits:   {cpu: 300m, memory: 256Mi}
      requests: {cpu: 100m, memory: 128Mi}

configs: []

What the chart creates

Template	Kind	Purpose
templates/serviceaccount.yml	ServiceAccount	Created only when serviceAccount.create: true.
templates/secret.yml	Secret	Holds K8S_HOST, K8S_CLUSTER_CERT, K8S_TOKEN.
templates/extra-secrets.yml	Secret	One per extraSecrets entry — only when list is non-empty.
templates/configmap.yml	ConfigMap	Concatenated scenarios mounted at /app/config.yaml.
templates/service.yml	Service	Exposes the /metrics port for scraping.
templates/deployment.yml	Deployment	Runs the agent, mounts the ConfigMap, injects credentials.
templates/alloy-configmap.yml	ConfigMap	Alloy sidecar config — only when monitoring.exporter.enabled.

ServiceAccount and RBAC

The chart never creates a Role or RoleBinding. Set serviceAccount.create: true to generate the ServiceAccount object, then bind it manually to a Role scoped to the namespaces you actually target.

caution

Cluster-wide ClusterRole bindings are intentionally out of scope. Chaos permissions deserve explicit review by the platform team.

See RBAC requirements for the minimal permission set.

Upgrading scenarios without restart

The pod mounts ConfigMap data via subPath, so changes do not propagate without a pod restart. After editing values.configs:

helm upgrade chaos-zookoo ./helm -f my-values.yaml
kubectl -n chaos-system rollout restart deployment/chaos-zookoo