Skip to content

Grafana in KOF#

Grafana installation and automatic configuration are now disabled in KOF by default.

Explore the Using KOF guide showing how to use KOF without Grafana.

Install and enable Grafana#

If you want to install Grafana manually and enable its support in KOF, apply the next steps:

  • If you're using kof-operators chart version 1.5.0 or less, run: 
    kubectl apply --server-side --force-conflicts -f \
https://github.com/grafana/grafana-operator/releases/download/v5.18.0/crds.yaml
    This is required because helm upgrade -i does not install immutable CRDs when a dependency is enabled in an already installed chart. KOF 1.6.0 uses auto-upgradable CRDs.
  • Apply kof-operators step of the Management Cluster section, adding: 
    --set grafana-operator.enabled=true
  • Apply kof-mothership step of the Management Cluster section, adding: 
    --set grafana.enabled=true
  • Install Grafana manually, for example: 
    kubectl apply -f - <<EOF
apiVersion: grafana.integreatly.org/v1beta1
kind: Grafana
metadata:
  name: grafana-vm
  namespace: kof
  labels:
    dashboards: grafana
spec:
  version: 10.4.18-security-01
  disableDefaultAdminSecret: true
  persistentVolumeClaim:
    spec:
      accessModes:
        - ReadWriteOnce
      resources:
        requests:
          storage: 200Mi
      # storageClassName: openebs-hostpath
  deployment:
    spec:
      template:
        spec:
          securityContext:
            fsGroup: 472
          volumes:
            - name: grafana-data
              persistentVolumeClaim:
                claimName: grafana-vm-pvc
          containers:
            - name: grafana
              env:
                - name: GF_SECURITY_ADMIN_USER
                  valueFrom:
                    secretKeyRef:
                      key: GF_SECURITY_ADMIN_USER
                      name: grafana-admin-credentials
                - name: GF_SECURITY_ADMIN_PASSWORD
                  valueFrom:
                    secretKeyRef:
                      key: GF_SECURITY_ADMIN_PASSWORD
                      name: grafana-admin-credentials
                - name: GF_INSTALL_PLUGINS
                  value: "victoriametrics-logs-datasource 0.21.0,victoriametrics-metrics-datasource 0.19.4"
EOF
  • You may optionally add features like dex and ingress from this example.
  • Wait for Grafana installation to complete successfully: 
    kubectl wait grafana -n kof grafana-vm \
  --for='jsonpath={.status.stage}=complete' \
  --for='jsonpath={.status.stageStatus}=success' \
  --timeout=5m
  • Get access to Grafana: 
    kubectl get secret -n kof grafana-admin-credentials -o yaml | yq '{
  "user": .data.GF_SECURITY_ADMIN_USER | @base64d,
  "pass": .data.GF_SECURITY_ADMIN_PASSWORD | @base64d
}'

kubectl port-forward -n kof svc/grafana-vm-service 3000:3000
  • Login to http://127.0.0.1:3000/dashboards with the username/password printed above.
  • Check the Dashboards - Cluster Monitoring - Kubernetes / Views / Global, it should show all clusters you collect metrics from.

collect-from-3-cluster-roles

grafana-demo

Cluster Overview#

From here you can get an overview of the cluster, including:

  • Health metrics
  • Resource utilization
  • Performance trends
  • Cost analysis

Logging Interface#

The logging interface will also be available, including:

  • Real-time log streaming
  • Full-text search
  • Log aggregation
  • Alert correlation

Traces#

You can view and analyze traces through Grafana Explore:

  1. Open Grafana in your browser.
  2. Navigate to "Explore" (compass icon in the left sidebar).
  3. Select the "Jaeger" type datasource from the dropdown at the top (not "VictoriaTraces").
  4. Use the query builder to search for traces by service name, operation, tags, or trace ID.

Dashboard Categories#

KOF ships with dashboards across:

  • Infrastructure: Provides infrastructure-related metrics, such as kube clusters, nodes, API server, networking, storage, or GPU.
  • Applications: Provides metrics for applications, such as VictoriaMetrics, VictoriaLogs, Jaeger and OpenCost.
  • Service Mesh: Provides metrics for service mesh, such as Istio control-plane and traffic.
  • Platform: Provides metrics for the platform itself, including KCM, Cluster API, and Sveltos.

Dashboard Lifecycle (GitOps Workflow)#

All dashboards are managed as code to keep environments consistent. To add or change a dashboard, follow these steps:

Add a new dashboard

  1. Create a YAML file under charts/kof-dashboards/files/dashboards/ with the new dashboard definition.
  2. Commit and push the change to Git.
  3. Your CI/CD pipeline applies the Helm chart to the target cluster.

Update an existing dashboard

  1. Edit the corresponding YAML file.
  2. Commit and push changes.
  3. CI/CD will roll out the update automatically.

Delete a dashboard

  1. Remove the YAML file.
  2. Commit and push changes.
  3. CI/CD pipeline removes the dashboard from Grafana.

Warning

Avoid editing dashboards directly in the Grafana UI. Changes will be overwritten by the next Helm release.

Single Sign-On#

Port forwarding, as described above, is a quick solution.

Single Single-On provides better experience. If you want to enable it, please apply this advanced guide: SSO for Grafana.

Uninstall Grafana#

If you have Grafana installed manually, and now you want to uninstall it, run:

kubectl delete --wait grafana -n kof grafana-vm