Prometheus & Grafana¶

The observability stack is deployed as a single kube-prometheus-stack HelmRelease in the monitoring namespace.

Components¶

Component	Purpose
Prometheus	Scrapes and stores metrics from all cluster components
Grafana	Dashboards and visualization
kube-state-metrics	Exposes K8s object metrics (pod status, deployment replicas, etc.)
node-exporter	Exposes node-level metrics (CPU, memory, disk, network)
prometheus-operator	Manages Prometheus config via CRDs (`ServiceMonitor`, `PodMonitor`)

Access Grafana¶

Via minikube service (simplest)Via Ingress

minikube service kube-prometheus-stack-grafana -n monitoring

Minikube opens a tunnel and launches the browser automatically.

echo "127.0.0.1  grafana.local" | sudo tee -a /etc/hosts
# Open: http://grafana.local  (requires sudo minikube tunnel)

Login: admin / admin

Change the password

The default password is committed as plaintext for local dev only. Phase 4 will replace this with SOPS-encrypted secrets.

Pre-installed Dashboards¶

The stack ships with dashboards for:

Kubernetes cluster overview
Node resource usage (CPU, RAM, disk)
Pod and container metrics
Namespace resource quotas

Access them via Dashboards → Browse in Grafana.

Adding Your Own Metrics (Phase 3+)¶

Apps expose metrics at /metrics (Prometheus format). To auto-discover them, create a ServiceMonitor:

apiVersion: monitoring.coreos.com/v1
kind: ServiceMonitor
metadata:
  name: my-app
  namespace: monitoring
spec:
  selector:
    matchLabels:
      app: my-app
  endpoints:
    - port: http
      path: /metrics
      interval: 30s

Prometheus will start scraping automatically — no config file edits needed.

Resource Tuning¶

Current limits (optimised for MacBook):

Component	Memory Request	Memory Limit
Prometheus	256 Mi	1 Gi
Grafana	128 Mi	256 Mi
prometheus-operator	64 Mi	128 Mi
kube-state-metrics	32 Mi	64 Mi
node-exporter	32 Mi	64 Mi

Retention: 24h / 2 GB (local sandbox — data does not persist across minikube delete).

See ADR-007 for the full decision record.