CNCF Atlas

English

日本語

English

日本語

Appearance

Getting Started ​

Verified against v0.9.1. Commands assume a running Kubernetes cluster, kubectl pointed at it, plus helm v3 and git on your machine.

Prerequisites ​

  • A Kubernetes cluster you can reach with kubectl (the host that will run Clusterpedia).
  • helm v3, git, and kubectl installed locally.
  • Credentials (a kubeconfig, or an apiserver URL plus a token or client certificate) for at least one member cluster you want to import.

Install ​

The official install path uses the Helm chart, which depends on the Bitnami PostgreSQL subchart. Clone the chart and resolve its dependencies first.

bash
git clone https://github.com/clusterpedia-io/clusterpedia-helm.git
helm dependency build ./clusterpedia-helm

A first working setup ​

  1. Install Clusterpedia with the default PostgreSQL storage. Set installCRDs=true so the PediaCluster CRD is created, and pin the database to a node so its PersistentVolume binds.

    bash
    helm install clusterpedia ./clusterpedia-helm \
  --namespace clusterpedia-system \
  --create-namespace \
  --set installCRDs=true \
  --set persistenceMatchNode=<your-node-name>

  2. Wait for the components to come up. You should see the apiserver, the clustersynchro-manager, the controller-manager, and the PostgreSQL pod.

    bash
    kubectl -n clusterpedia-system get pods

  3. Describe a member cluster as a PediaCluster. Save this as pediacluster.yaml, filling in the apiserver address and credentials for the cluster you are importing.

    yaml
    apiVersion: cluster.clusterpedia.io/v1alpha2
kind: PediaCluster
metadata:
  name: cluster-example
spec:
  apiserver: "https://10.30.43.43:6443"
  caData:
  tokenData:
  syncResources:
    - group: apps
      resources:
        - deployments
    - group: ""
      resources:
        - pods

  4. Apply it to register the cluster and start synchronization.

    bash
    kubectl apply -f pediacluster.yaml

  5. Confirm the cluster is being synced. The STATUS column should report the synchro running.

    bash
    kubectl get pediaclusters

Verify it works ​

Clusterpedia registers as an Aggregated API, so you can query it through your existing kubeconfig. List the resource API groups it serves:

bash
kubectl get --raw "/apis/clusterpedia.io/v1beta1/resources/apis/apps" | jq

To search Deployments across every imported cluster, query the Clusterpedia resource endpoint and read the items it returns:

bash
kubectl get --raw "/apis/clusterpedia.io/v1beta1/resources/apis/apps/v1/deployments" | jq '.items | length'

Where to go next ​

  • Installation for production storage choices and high availability.
  • Import Clusters for authentication options and automatic import from platforms such as Karmada.
  • Sync Cluster Resources for selecting which resources to collect, including syncAllCustomResources.