Container Service for Kubernetes:Schedule pods on ACS using a virtual node

How it works

Virtual nodes integrate ACS computing power into your ACK One registered cluster. After you install the ack-virtual-node component, a virtual node is created in your cluster. When you schedule a pod to the virtual node, ACS automatically provisions serverless compute resources for the pod. The pod runs in an isolated environment on ACS and can communicate with other pods in the cluster. Virtual nodes are ideal for workloads with variable or unpredictable traffic patterns because they scale on demand without requiring you to provision or manage physical nodes.

For more information about virtual nodes, see Overview of registered clusters.

Prerequisites

• You have created an ACK One registered cluster and connected it to a Kubernetes cluster. Kubernetes 1.24 or later is required. For more information, see Create an ACK One registered cluster.

• The ack-virtual-node component version 2.13.0 or later is installed in the registered cluster. For more information, see Install the ack-virtual-node component.

Configure RAM permissions for the ack-virtual-node component

Example: Use ACS CPU computing power

After you install or upgrade the ack-virtual-node component to version 2.13.0 or later, the component supports both ACS and Elastic Container Instance (ECI) computing power.

Perform the following steps to use ACS CPU computing power with an ACK One registered cluster:

1. Update the security group configuration of the registered cluster.

2. On the Basic Information page of the cluster, click the ID of the Control Plane Security Group.

3. On the security group details page, click Add Rule. Configure the rule by using the following values.

   Rule Type	Protocol	Port Range	Source IP Range	Description
   Inbound	TCP	80	CIDR block of the IDC cluster, for example, 192.168.1.0/24.	For configuring ACS endpoint scenarios.
   Inbound	TCP	443	CIDR block of the IDC cluster, for example, 192.168.1.0/24.	For configuring ACS endpoint scenarios.
   Inbound	TCP	10250	CIDR block of the IDC cluster, for example, 192.168.1.0/24.	The port listened on by the serverless kubelet service.

4. Create a Deployment that uses ACS computing power.

5. Create a file named nginx.yaml and copy the following content to the file:

   apiVersion: apps/v1
   kind: Deployment
   metadata:
     name: nginx
     labels:
       app: nginx
   spec:
     replicas: 2
     selector:
       matchLabels:
         app: nginx
     template:
       metadata:
         labels:
           app: nginx 
           alibabacloud.com/acs: "true" # Configure to use ACS computing power.
           alibabacloud.com/compute-class: general-purpose # Configure the computing power type for the ACS pod.
           alibabacloud.com/compute-qos: default # Configure the computing power quality for the ACS pod.
       spec:
         containers:
         - name: nginx
           image: mirrors-ssl.aliyuncs.com/nginx:stable-alpine
           ports:
             - containerPort: 80
               protocol: TCP 
           resources:
             limits:
               cpu: 2
             requests:
               cpu: 2

6. Run the following command to create the nginx application:

   kubectl apply -f nginx.yaml 

7. Run the following command to check the deployment status:

   kubectl get pods -o wide

   Expected output (simplified):

   NAME                     READY   STATUS    RESTARTS   AGE     IP               NODE                            NOMINATED NODE   READINESS GATES
   nginx-54bcbc9b66-****   1/1     Running   0          3m29s   192.168.XX.XXX   virtual-kubelet-cn-shanghai-l   <none>           <none>
   nginx-54bcbc9b66-****   1/1     Running   0          3m29s   192.168.XX.XXX   virtual-kubelet-cn-shanghai-l   <none>           <none>

   The output shows that the two pods are scheduled to the node labeled type=virtual-kubelet.

8. Run the following command to view the details of an nginx pod:

   kubectl describe pod nginx-54bcbc9b66-**** 

   Expected output:

   Annotations:  ProviderCreate: done
                 alibabacloud.com/instance-id: acs-uf6008giwgjxlvn*****
                 alibabacloud.com/pod-ephemeral-storage: 30Gi
                 alibabacloud.com/pod-use-spec: 2-2Gi
                 kubernetes.io/pod-stream-port: 10250
                 network.alibabacloud.com/enable-dns-cache: false
                 topology.kubernetes.io/region: cn-shanghai

   If the output contains Annotation alibabacloud.com/instance-id: acs-uf6008giwgjxlvn*****, the pod is an ACS pod instance.

Example: Use ACS GPU computing power

...     
     labels:
        # Declare the ACS GPU resource requirements in the label.
        alibabacloud.com/compute-class: gpu     # If the type is GPU, set this to gpu.
        alibabacloud.com/compute-qos: default   # The QoS type for computing. The meaning is the same as for normal ACS computing power.
        alibabacloud.com/gpu-model-series: GN8IS  # The GPU model. Replace this with the actual model.
...

Example: Use ACS GPU HPN computing power

The process for using ACS GPU HPN computing power is similar to using ACS CPU computing power, but has the following requirements:

• You must purchase a GPU-HPN capacity reservation in advance and associate it with the cluster.

• ACS GPU HPN computing power requires upgrading the ack-virtual-node component. The required component version is currently in invitational preview. To enable this feature, submit a ticket.

...     
labels:
  # Declare the ACS GPU resource requirements in the label.
  alibabacloud.com/compute-class: gpu-hpn     # Set to the gpu-hpn type.
  alibabacloud.com/compute-qos: default    # The computing QoS type. The meaning is the same as for regular ACS computing power.
  alibabacloud.com/acs: "true"           # The label to configure the use of ACS computing power.
...