Getting Started

Verified against the main manifest cnpg-1.30.0-rc1.yaml. Commands assume a running Kubernetes cluster and a configured kubectl.

Prerequisites

• A Kubernetes cluster (a local one such as kind or minikube is fine for this walkthrough).
• kubectl configured to reach that cluster, with permission to apply cluster-scoped resources.
• A default StorageClass, so the Cluster can allocate a PersistentVolume.

Install

Apply the operator manifest. This example uses the release candidate manifest carried on main; for production use the matching YAML from the releases page of the stable line you want.

kubectl apply --server-side -f \
  https://raw.githubusercontent.com/cloudnative-pg/cloudnative-pg/main/releases/cnpg-1.30.0-rc1.yaml

Wait for the operator Deployment to roll out:

kubectl rollout status deployment \
  -n cnpg-system cnpg-controller-manager

A first working setup

1. Write a three-instance Cluster definition to cluster-example.yaml:

   apiVersion: postgresql.cnpg.io/v1
   kind: Cluster
   metadata:
     name: cluster-example
   spec:
     instances: 3
     storage:
       size: 1Gi

2. Apply it:

   kubectl apply -f cluster-example.yaml

3. Watch the Pods come up. The operator labels every object it manages with cnpg.io/cluster:

   kubectl get pods -l cnpg.io/cluster=cluster-example

   You should see three Pods, one primary and two replicas, reaching Running.

Verify it works

Install the cnpg kubectl plugin (the kubectl-cnpg binary built from cmd/kubectl-cnpg) and ask for cluster status:

kubectl cnpg status cluster-example

A healthy cluster reports a single primary, the replica instances streaming, and a Cluster in healthy state summary. You can also confirm directly from the resource:

kubectl get cluster cluster-example

The STATUS column should read Cluster in healthy state once reconciliation settles.

Where to go next

• Backup and point-in-time recovery to object storage, configured through the barman-cloud plugin.
• High availability tuning: synchronous replication quorum (minSyncReplicas / maxSyncReplicas) and the FailoverQuorum resource.
• Connection pooling with the Pooler CRD (PgBouncer) and monitoring with the generated Prometheus PodMonitor.

See the official documentation for these production concerns.