The Huawei Karpenter Provider enables node autoprovisioning using Karpenter on your CCE cluster. Karpenter improves the efficiency and cost of running workloads on Kubernetes clusters by:

• Watching for pods that the Kubernetes scheduler has marked as unschedulable
• Evaluating scheduling constraints (resource requests, node selectors, affinities, tolerations, and topology-spread constraints) requested by the pods
• Provisioning nodes that meet the requirements of the pods
• Removing the nodes when they are no longer needed
• Consolidating existing nodes onto cheaper nodes with higher utilization per node

• On-demand only: Only on-demand capacity type is supported. Spot instances are not yet available.

• A Huawei Cloud CCE cluster running Kubernetes 1.26 - 1.34
• Huawei Cloud credentials (AK/SK) with permissions for:
  • ECS - Elastic Cloud Server (flavor listing, server tagging)
  • CCE - Cloud Container Engine (node create/delete/list)
  • VPC - Virtual Private Cloud (subnet discovery)
  • BSS - Billing (on-demand pricing queries)
• Helm v3

helm install karpenter-provider-huawei charts/karpenter-provider-huawei \
  --namespace karpenter-provider-huawei-system \
  --create-namespace \
  --set-string credentials.accessKey=<your-access-key> \
  --set-string credentials.secretKey=<your-secret-key> \
  --set-string credentials.region=<region-id> \
  --set-string clusterInfo.clusterID=<cce-cluster-id>

The chart creates a huawei-credentials Secret by default and loads it into the controller. To use an existing Secret instead, set credentials.create=false and credentials.existingSecret=<secret-name>.

CCENodeClass is a cluster-scoped resource that defines Huawei Cloud-specific node configuration:

apiVersion: karpenter.k8s.huawei/v1alpha1
kind: CCENodeClass
metadata:
  name: default
spec:
  subnetSelectorTerms:
    - id: "<subnet-uuid>"                  # Your VPC subnet ID
  imsSelector:
    imsFamily: "Huawei Cloud EulerOS 2.0"  # Example value verified on a live CCE cluster
  blockDeviceMappings:
    k8s:
      volumeSize: 120
      volumeType: SAS
    root:
      volumeSize: 120
      volumeType: SAS
    users:
      - volumeSize: 100
        volumeType: SAS
  runtimeConfiguration:
    type: containerd
  login:
    userPassword:
      username: root
      password: "<salted-and-encrypted-password>"

After creation, wait for the SubnetsReady condition to become True before the NodeClass can be used for provisioning:

kubectl get ccenodeclass default -o jsonpath='{.status.conditions}'

Create a Karpenter NodePool that references your CCENodeClass:

apiVersion: karpenter.sh/v1
kind: NodePool
metadata:
  name: default
spec:
  template:
    spec:
      nodeClassRef:
        group: karpenter.k8s.huawei
        kind: CCENodeClass
        name: default
      requirements:
        - key: karpenter.sh/capacity-type
          operator: In
          values: ["on-demand"]
        - key: kubernetes.io/arch
          operator: In
          values: ["amd64"]
  disruption:
    consolidationPolicy: WhenEmptyOrUnderutilized
    consolidateAfter: 1m

Deploy a workload with resource requests. Karpenter will automatically provision right-sized nodes:

kubectl apply -f - <<EOF
apiVersion: apps/v1
kind: Deployment
metadata:
  name: inflate
spec:
  replicas: 5
  selector:
    matchLabels:
      app: inflate
  template:
    metadata:
      labels:
        app: inflate
    spec:
      containers:
        - name: inflate
          image: nginx:latest
          resources:
            requests:
              cpu: "1"
              memory: 1Gi
EOF

• Go 1.25+
• Docker (or Podman)
• kubectl
• Helm v3
• Kind (for e2e tests)

make build                                              # Build controller binary
make docker-build IMG=<your-registry>/controller:<tag>  # Build Docker image
make docker-push IMG=<your-registry>/controller:<tag>   # Push Docker image
make docker-buildx IMG=<your-registry>/controller:<tag> # Cross-platform build

make test      # Unit tests
make test-e2e  # E2E tests (uses Kind)

make lint      # Run golangci-lint
make lint-fix  # Run with auto-fix

make manifests  # Generate CRD manifests
make generate   # Generate DeepCopy methods

make helm-lint       # Lint the chart
make helm-template   # Render templates locally
make helm-install    # Install
make helm-upgrade    # Upgrade
make helm-uninstall  # Uninstall

See docs/ROADMAP.md for the project roadmap and progress.

Contributions are welcome! Please feel free to submit issues and pull requests.

1. Fork the repository
2. Create a feature branch (git checkout -b feature/my-feature)
3. Make your changes and ensure tests pass (make test && make lint)
4. Submit a Pull Request following the PR template

• Issues: GitHub Issues
• Karpenter Upstream: karpenter.sh | Karpenter GitHub

This project follows the CNCF Community Code of Conduct.

This project is licensed under the Apache License 2.0.