跳到主要内容

Metrics Authentication and Authorization

Introduction​

Starting from Kubebuilder v4, Shifu's controller exposes a /metrics endpoint with built-in authentication and authorization. This endpoint provides Prometheus-compatible metrics for monitoring the controller's health and performance.

Key Features:

  • Authentication: All requests are authenticated using Kubernetes TokenReview
  • Authorization: Access is controlled via RBAC using SubjectAccessReview
  • HTTP by default: Metrics are served over HTTP (port 8080) to avoid certificate issues
  • Optional HTTPS: Can be enabled with cert-manager for encrypted transport
信息

Authentication and authorization are always enabled, regardless of whether you use HTTP or HTTPS. This ensures that only authorized users and service accounts can access metrics.

Default Configuration (HTTP)​

By default, Shifu serves metrics on HTTP port 8080 with authentication enabled. This configuration works out-of-the-box without additional setup and is suitable for most deployments where metrics scraping happens within a trusted cluster network.

Why HTTP by Default?​

When using HTTPS without cert-manager, controller-runtime generates self-signed certificates that only contain localhost and 127.0.0.1 as Subject Alternative Names (SANs). This causes x509 validation errors when Prometheus or other tools try to scrape metrics using the service DNS name (e.g., shifu-crd-controller-manager-metrics-service.shifu-crd-system.svc.cluster.local).

HTTP avoids these certificate issues while still maintaining security through authentication and authorization.

How Authentication Works​

The metrics endpoint uses Kubernetes' built-in authentication mechanisms:

  1. TokenReview: Verifies the bearer token sent by the client is valid
  2. SubjectAccessReview: Checks if the authenticated identity has permission to access /metrics

This means that every request must include a valid Kubernetes ServiceAccount token and the ServiceAccount must have the appropriate RBAC permissions.

Access Without Authorization​

If you try to access metrics without proper authorization:

$ kubectl exec -it deploy/prometheus -n monitoring -- \
  curl -k http://shifu-crd-controller-manager-metrics-service.shifu-crd-system.svc.cluster.local:8080/metrics

Unauthorized

Granting Metrics Access​

To allow a ServiceAccount to access metrics, you need to bind it to the shifu-crd-metrics-reader ClusterRole.

Step 1: Create a ServiceAccount​

Create a ServiceAccount in the namespace where your monitoring tool runs:

apiVersion: v1
kind: ServiceAccount
metadata:
  name: prometheus
  namespace: monitoring

Step 2: Create a ClusterRoleBinding​

Bind the ServiceAccount to the shifu-crd-metrics-reader ClusterRole:

apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: prometheus-metrics-reader
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: shifu-crd-metrics-reader
subjects:
- kind: ServiceAccount
  name: prometheus
  namespace: monitoring

Apply both resources:

kubectl apply -f prometheus-sa.yaml
kubectl apply -f prometheus-metrics-binding.yaml

Step 3: Configure Prometheus to Use the ServiceAccount​

Update your Prometheus deployment to use the ServiceAccount:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: prometheus
  namespace: monitoring
spec:
  template:
    spec:
      serviceAccountName: prometheus  # Use the ServiceAccount we created
      containers:
      - name: prometheus
        image: prom/prometheus:latest
        # ... rest of configuration

Step 4: Add Scrape Configuration​

Configure Prometheus to scrape the Shifu controller metrics:

scrape_configs:
  - job_name: 'shifu-controller'
    kubernetes_sd_configs:
      - role: endpoints
        namespaces:
          names:
          - shifu-crd-system
    relabel_configs:
      - source_labels: [__meta_kubernetes_service_name]
        regex: shifu-crd-controller-manager-metrics-service
        action: keep
      - source_labels: [__meta_kubernetes_endpoint_port_name]
        regex: http
        action: keep
    scheme: http

Verification​

Test that the ServiceAccount can now access metrics:

$ kubectl exec -it deploy/prometheus -n monitoring -- \
  curl -H "Authorization: Bearer $(cat /var/run/secrets/kubernetes.io/serviceaccount/token)" \
  http://shifu-crd-controller-manager-metrics-service.shifu-crd-system.svc.cluster.local:8080/metrics

# HELP go_gc_duration_seconds A summary of the pause duration of garbage collection cycles.
# TYPE go_gc_duration_seconds summary
go_gc_duration_seconds{quantile="0"} 2.5875e-05
...

Optional: Enabling HTTPS with Cert-Manager​

For production environments requiring encrypted metrics, you can enable HTTPS with cert-manager. This configuration:

  1. Serves metrics on port 8443 with TLS
  2. Generates valid certificates with correct DNS names as SANs
  3. Maintains authentication and authorization

Prerequisites​

Install cert-manager in your cluster:

kubectl apply -f https://github.com/cert-manager/cert-manager/releases/download/v1.18.2/cert-manager.yaml

Wait for cert-manager to be ready:

kubectl wait --for=condition=Available --timeout=300s \
  deployment/cert-manager -n cert-manager

Enable HTTPS in Shifu​

Edit pkg/k8s/crd/config/default/kustomization.yaml and uncomment the cert-manager sections:

resources:
# Uncomment the following lines
- ../certmanager

patches:
# Uncomment the following patches
- path: cert_metrics_manager_patch.yaml
  target:
    kind: Deployment
- path: cert_metrics_service_patch.yaml
  target:
    kind: Service
    name: controller-manager-metrics-service

# Uncomment the replacements section
replacements:
- source:
    kind: Service
    version: v1
    name: controller-manager-metrics-service
    fieldPath: metadata.name
  targets:
    - select:
        kind: Certificate
        group: cert-manager.io
        version: v1
        name: metrics-certs
      fieldPaths:
        - spec.dnsNames.0
        - spec.dnsNames.1
      options:
        delimiter: '.'
        index: 0
        create: true
# ... (rest of replacements)

Redeploy Shifu​

Apply the updated configuration:

cd pkg/k8s/crd
make deploy IMG=edgehub/shifu-controller:nightly

Update Prometheus Configuration​

Update the scrape configuration to use HTTPS:

scrape_configs:
  - job_name: 'shifu-controller'
    scheme: https
    tls_config:
      ca_file: /var/run/secrets/kubernetes.io/serviceaccount/ca.crt
      insecure_skip_verify: false  # Certificate is now valid!
    kubernetes_sd_configs:
      - role: endpoints
        namespaces:
          names:
          - shifu-crd-system
    relabel_configs:
      - source_labels: [__meta_kubernetes_service_name]
        regex: shifu-crd-controller-manager-metrics-service
        action: keep
      - source_labels: [__meta_kubernetes_endpoint_port_name]
        regex: https  # Changed from 'http' to 'https'
        action: keep

Troubleshooting​

"Unauthorized" Error​

Problem: Metrics endpoint returns Unauthorized

Solution: This means authentication failed. Ensure your request includes a valid bearer token:

# Include the ServiceAccount token in the request
TOKEN=$(cat /var/run/secrets/kubernetes.io/serviceaccount/token)
curl -H "Authorization: Bearer $TOKEN" http://metrics-service:8080/metrics

"Forbidden" Error​

Problem: Metrics endpoint returns Forbidden

Solution: This means authentication succeeded, but authorization failed. The ServiceAccount doesn't have permission to access /metrics. Create a ClusterRoleBinding:

apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: my-app-metrics-reader
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: shifu-crd-metrics-reader
subjects:
- kind: ServiceAccount
  name: my-app
  namespace: my-namespace

x509 Certificate Errors (HTTPS)​

Problem: x509: certificate is valid for localhost, not shifu-crd-controller-manager-metrics-service.shifu-crd-system.svc.cluster.local

Solution: This happens when you enable --metrics-secure=true without cert-manager. Either:

  1. Use HTTP (recommended for most cases): Keep the default configuration
  2. Enable cert-manager: Follow the HTTPS setup guide above

Connection Refused​

Problem: Connection refused when trying to access metrics

Solution: Check that the Shifu controller is running:

kubectl get pods -n shifu-crd-system
kubectl logs -n shifu-crd-system deploy/shifu-crd-controller-manager

Check the service exists:

kubectl get svc -n shifu-crd-system shifu-crd-controller-manager-metrics-service

Complete Example: Prometheus with Metrics Access​

Here's a complete example of deploying Prometheus with metrics access:

1. Create Monitoring Namespace​

kubectl create namespace monitoring

2. Create ServiceAccount and RBAC​

# prometheus-rbac.yaml
apiVersion: v1
kind: ServiceAccount
metadata:
  name: prometheus
  namespace: monitoring
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: prometheus-metrics-reader
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: shifu-crd-metrics-reader
subjects:
- kind: ServiceAccount
  name: prometheus
  namespace: monitoring

3. Create Prometheus ConfigMap​

# prometheus-config.yaml
apiVersion: v1
kind: ConfigMap
metadata:
  name: prometheus-config
  namespace: monitoring
data:
  prometheus.yml: |
    global:
      scrape_interval: 15s
    scrape_configs:
      - job_name: 'shifu-controller'
        kubernetes_sd_configs:
          - role: endpoints
            namespaces:
              names:
              - shifu-crd-system
        relabel_configs:
          - source_labels: [__meta_kubernetes_service_name]
            regex: shifu-crd-controller-manager-metrics-service
            action: keep
          - source_labels: [__meta_kubernetes_endpoint_port_name]
            regex: http
            action: keep
        scheme: http

4. Deploy Prometheus​

# prometheus-deployment.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
  name: prometheus
  namespace: monitoring
spec:
  replicas: 1
  selector:
    matchLabels:
      app: prometheus
  template:
    metadata:
      labels:
        app: prometheus
    spec:
      serviceAccountName: prometheus
      containers:
      - name: prometheus
        image: prom/prometheus:latest
        ports:
        - containerPort: 9090
        volumeMounts:
        - name: config
          mountPath: /etc/prometheus
      volumes:
      - name: config
        configMap:
          name: prometheus-config

5. Apply All Resources​

kubectl apply -f prometheus-rbac.yaml
kubectl apply -f prometheus-config.yaml
kubectl apply -f prometheus-deployment.yaml

6. Verify Metrics Collection​

Port-forward to Prometheus:

kubectl port-forward -n monitoring deploy/prometheus 9090:9090

Open http://localhost:9090 and check that the shifu-controller target is up and scraping metrics successfully.

Summary​

  • Metrics are always authenticated and authorized, regardless of HTTP or HTTPS
  • Default configuration uses HTTP (port 8080) to avoid certificate issues
  • Grant access by binding ServiceAccounts to the shifu-crd-metrics-reader ClusterRole
  • HTTPS is optional and requires cert-manager for valid certificates
  • Authentication errors (Unauthorized) mean the token is missing or invalid
  • Authorization errors (Forbidden) mean the ServiceAccount lacks permissions

For more information about the metrics system, see the controller-runtime metrics documentation.