Installation¶

Prerequisites¶

Kubernetes ≥ 1.24
Helm ≥ 3.10
Prometheus with PromQL-compatible endpoint
Loki with LogQL endpoint
Alertmanager v2 endpoint

What this chart does NOT install

Prometheus, Loki, and Alertmanager must already be running in your cluster. The anomaly detection service connects to them as data sources.

Add the Helm repository¶

helm repo add staffops https://staffops.github.io/helm-charts/
helm repo update

Quick install¶

helm install ad staffops/staffops-anomaly-detection \
  --namespace monitoring \
  --create-namespace \
  --set clusterName=my-cluster \
  --set datasources.victoriametrics.url=https://vm.example.com/select/0/prometheus \
  --set datasources.loki.url=https://loki.example.com \
  --set datasources.alertmanager.url=https://alertmanager.example.com

dryRun is enabled by default

The controller starts in dryRun=true mode — anomalies are logged but alerts are NOT sent to Alertmanager. Disable with --set controller.dryRun=false when ready for live alerting.

Production install¶

Create a values file:

values-prd.yaml

clusterName: prd-eks

datasources:
  prometheus:
    url: https://vm.internal/select/0/prometheus
  loki:
    url: https://loki.internal
  alertmanager:
    url: https://alertmanager.internal

controller:
  replicaCount: 2
  dryRun: false

redis:
  enabled: false
  external:
    addr: redis-prd.cache.amazonaws.com:6379
    existingSecret: redis-credentials

vmServiceScrape:
  enabled: true

vmRule:
  enabled: true

grafanaDashboard:
  enabled: true

links:
  grafanaBaseUrl: https://grafana.example.com
  runbookBaseUrl: https://docs.example.com/runbooks

helm install ad staffops/staffops-anomaly-detection \
  --namespace monitoring \
  --create-namespace \
  -f values-prd.yaml

Verify the install¶

# Check pods
kubectl get pods -n monitoring -l app.kubernetes.io/name=staffops-anomaly-detection

# Controller readiness
kubectl exec -n monitoring deploy/ad-staffops-anomaly-detection-controller -- \
  wget -qO- http://localhost:8080/readyz

# Metrics endpoint
kubectl port-forward -n monitoring svc/ad-staffops-anomaly-detection-controller 8080:8080 &
curl -s localhost:8080/metrics | grep staffops_ad_controller_cycles_total

Upgrade¶

helm repo update
helm upgrade ad staffops/staffops-anomaly-detection -n monitoring -f values-prd.yaml

Baseline preservation on upgrade

If redis.persistence.enabled=true or an external Redis is used, baselines survive upgrades. With in-cluster ephemeral Redis, baselines warm up again from scratch (~30 minutes).

Uninstall¶

helm uninstall ad -n monitoring

If Redis persistence was enabled, delete the PVC manually:

kubectl delete pvc -n monitoring -l app.kubernetes.io/name=staffops-anomaly-detection