Target Pods to VNodes Using nodeSelector or nodeName - Elastic Container Instance - Alibaba Cloud - Elastic Container Instance

In self-managed Kubernetes clusters that use both real nodes and virtual nodes (VNodes), you can schedule pods to a VNode so they run as elastic container instances (ECIs). This topic covers two manual scheduling methods: configuring nodeSelector and tolerations, or specifying nodeName directly.

Both methods require modifying individual workload resources. For production environments, use the eci-profile component instead—it uses a label selector to automatically route matching pods to VNodes across all workloads, without per-workload changes. See Use eci-profile to schedule pods to a VNode.

Choose a scheduling method

Method	How it works	When to use
`nodeSelector` + `tolerations`	Matches pods to VNodes by label and tolerates the VNode taint	Schedule a group of pods to any available VNode (recommended for most cases)
`nodeName`	Pins pods to a specific VNode by name	Testing or debugging on a known VNode

Method 1: Configure nodeSelector and tolerations

By default, VNodes carry the following label and taint:

labels:
  k8s.aliyun.com/vnode: "true"
taints:
- effect: NoSchedule
  key: k8s.aliyun.com/vnode
  value: "true"

To schedule pods to VNodes, set nodeSelector to match the label and add a toleration for the NoSchedule taint.

Prerequisites

Before you begin, ensure that you have:

A self-managed Kubernetes cluster with at least one VNode (verify with kubectl get nodes)
kubectl configured to connect to the cluster

Steps

Save the following content as deploy-vnode-test1.yaml.

apiVersion: apps/v1
kind: Deployment
metadata:
  name: vnode-nginx-test1
spec:
  replicas: 2
  selector:
    matchLabels:
      app: uid
  template:
    metadata:
      labels:
        app: uid
    spec:
      nodeSelector:          # Match VNodes by label
        k8s.aliyun.com/vnode: "true"
      tolerations:           # Tolerate the VNode NoSchedule taint
      - key: k8s.aliyun.com/vnode
        operator: "Equal"
        value: "true"
        effect: "NoSchedule"
      containers:
      - name: nginx
        image: registry-vpc.cn-hangzhou.aliyuncs.com/eci_open/nginx:1.14.2

Apply the Deployment.

kubectl apply -f deploy-vnode-test1.yaml

Method 2: Specify nodeName

Specifying nodeName bypasses the Kubernetes scheduler and pins pods directly to a named VNode.

Prerequisites

Before you begin, ensure that you have:

A self-managed Kubernetes cluster with at least one VNode
kubectl configured to connect to the cluster
The exact name of the target VNode

Steps

Get the VNode name.

kubectl get nodes

VNode names follow the format <region>.<VNode ID>. Sample output:

NAME                                    STATUS   ROLES                  AGE    VERSION
cn-hangzhou.vnd-2ze2qhf8lmp5kgsx****   Ready    agent                  132m   v1.20.6
k8s-master                             Ready    control-plane,master   169m   v1.20.6

Save the following content as deploy-vnode-test2.yaml. Replace cn-hangzhou.vnd-2ze2qhf8lmp5kgsx**** with your VNode name.

apiVersion: apps/v1
kind: Deployment
metadata:
  name: vnode-nginx-test2
spec:
  replicas: 2
  selector:
    matchLabels:
      app: uid
  template:
    metadata:
      labels:
        app: uid
    spec:
      nodeName: cn-hangzhou.vnd-2ze2qhf8lmp5kgsx****   # Replace with your VNode name
      containers:
      - name: nginx
        image: registry-vpc.cn-hangzhou.aliyuncs.com/eci_open/nginx:1.14.2

Apply the Deployment.

kubectl apply -f deploy-vnode-test2.yaml

Next steps

Use eci-profile to schedule pods to a VNode — the recommended approach for production environments