Container Compute Service:Use the computing power of ACS in ACK Pro clusters

How to use the computing power of ACS in ACK Pro clusters

Container Compute Service (ACS) is a cloud computing service that provides container compute resources that comply with the container specifications of Kubernetes. ACS adopts a layered architecture to implement Kubernetes control and computing power. The compute resources layer schedules and allocates resources to pods. The Kubernetes control layer manages workloads, such as Deployments, Services, StatefulSets, and CronJobs.

The computing power of ACS can be implemented in Kubernetes clusters by using virtual nodes. This way, Kubernetes clusters are empowered with high elasticity and are no longer limited by the computing capacity of cluster nodes. After you use ACS to take over infrastructure management for pods, the Kubernetes cluster no longer needs to schedule or launch individual pods. In addition, the Kubernetes cluster no longer needs to be concerned about the resources of underlying VMs. ACS can meet the resource requirements of pods at any time.

Container Service for Kubernetes (ACK) is one of the first services to participate in the Certified Kubernetes Conformance Program in the world. ACK provides high-performance containerized application management service. ACK is integrated with the virtualization, storage, network, and security capabilities provided by Alibaba Cloud. ACK simplifies your cluster setup and scaling and allows you to focus on containerized application development and management.

Before you can create ACS pods in an ACK Pro cluster, you must deploy virtual nodes in the cluster. If you need to scale out your ACK Pro cluster, you can create ACS pods on virtual nodes, without the need to plan the resource capacities of the virtual nodes. ACS pods can communicate with pods on physical nodes in the cluster. We recommend that you deploy long-lived applications whose workloads periodically fluctuate on virtual nodes. This improves resource utilization, reduces resource costs, and accelerates the scaling process. When the workload of your application decreases, you can remove pods from virtual nodes to reduce resource costs. Pods on virtual nodes run in a secure and isolated environment that is built on top of ACS. In this case, a pod is referred to as an ACS pod. For more information, see ACK cluster overview.

Install ACK Virtual Node to implement the computing power of ACS

The computing power of ACS can be implemented in ACK clusters by using virtual nodes. This way, Kubernetes clusters are empowered with high elasticity and are no longer limited by the computing capacity of cluster nodes. The following section describes how to transfer files by using SFTP.

1. Log on to the ACK console. In the left-side navigation pane, click Clusters.

2. On the Clusters page, find the one you want to manage and click its name. In the left-side navigation pane, click Add-ons.

3. On the Core Components tab, select ACK Virtual Node and click Install to install the component or click Update to update the component to the required version.

   

4. If the console prompts you to activate and grant permissions to ACS when you install ACK Virtual Node, follow the on-screen instructions to activate and grant permissions to ACS. Click OK.

5. After you install the component, choose Nodes > Nodes in the left-side navigation pane of the cluster details page. By default, the names of virtual nodes are prefixed with virtual-kubelet-.

   

Example

After you install the required version of ACK Virtual Node or update the component to the required version as described in the Prerequisites section, you can create ACS pods and elastic container instances.

To implement the computing power of ACS in an ACK cluster, perform the following steps:

1. Configure node selectors, affinity and anti-affinity rules, ResourcePolicies, and the alibabacloud.com/acs: "true" label to schedule pods to virtual nodes. For more information, see Node affinity scheduling.

   Note

   The alibabacloud.com/acs: "true" label does not apply to ACK Serverless clusters. It applies to the following clusters: ACK managed clusters, ACK dedicated clusters, ACK One registered clusters, and ACK Edge clusters.

2. When you create an ACS pod, add the alibabacloud.com/compute-class:Compute class label to the pod to specify the compute class of the pod. For more information about the compute classes of ACS pods, see ACS pod overview.

The following section describes how to transfer files by using SFTP.

1. Create a Deployment.

   Important

   If you schedule pods to virtual nodes by using the alibabacloud.com/acs: "true" pod label, the WaitForFirstConsumer StorageClass is not supported. Therefore, when you use ACS pods that are mounted with disks in ACK clusters, use the nodeSelector or create a ResourcePolicy to schedule pods to virtual nodes. For more information about configuring a ResourcePolicy, see ACK Pro clusters support colocated scheduling of ECS instances and ACS computing power.

   NodeSelector

   1. Run the following command to query the labels of a virtual node. Replace virtual-kubelet-cn-hangzhou-k in the following command with the actual virtual node name.

      kubectl get node virtual-kubelet-cn-hangzhou-k -oyaml

      The following expected output displays only the content related to labels:

      apiVersion: v1
      kind: Node
      metadata:
        labels:
          kubernetes.io/arch: amd64
          kubernetes.io/hostname: virtual-kubelet-cn-hangzhou-k
          kubernetes.io/os: linux
          kubernetes.io/role: agent
          service.alibabacloud.com/exclude-node: "true"
          topology.diskplugin.csi.alibabacloud.com/zone: cn-hangzhou-k
          topology.kubernetes.io/region: cn-hangzhou
          topology.kubernetes.io/zone: cn-hangzhou-k
          type: virtual-kubelet # Each virtual node has this label. If you want to schedule a pod to a virtual node, you can configure this label as the node selector of the pod. 
        name: virtual-kubelet-cn-hangzhou-k
      spec:
        taints:
        - effect: NoSchedule
          key: virtual-kubelet.io/provider
          value: alibabacloud 

   2. Create a YAML file named nginx.yaml based on the following content to provision two pods:

      apiVersion: apps/v1
      kind: Deployment
      metadata:
        name: nginx
        labels:
          app: nginx
      spec:
        replicas: 2
        selector:
          matchLabels:
            app: nginx
        template:
          metadata:
            name: nginx
            labels:
              app: nginx 
              alibabacloud.com/compute-class: general-purpose # The compute class of the ACS pod. Default value: general-purpose.
              alibabacloud.com/compute-qos: default # The QoS class of the ACS pod. Default value: default.
          spec:
            nodeSelector:
              type: virtual-kubelet # The node selector used to select a virtual node.
            tolerations:
            - key: "virtual-kubelet.io/provider" # The toleration used to tolerate virtual nodes. 
              operator: "Exists"
              effect: "NoSchedule"
            containers:
            - name: nginx
              image: registry.openanolis.cn/openanolis/nginx:1.14.1-8.6
              resources:
                limits:
                  cpu: 2
                requests:
                  cpu: 2

   3. Deploy an NGINX application and query the pods.

      1. Run the following command to deploy an NGINX application:

         kubectl apply -f nginx.yaml 

      2. Run the following command to check whether the NGINX application is deployed:

         kubectl get pods -o wide

         Expected results:

         NAME                    READY   STATUS    RESTARTS   AGE   IP               NODE                            NOMINATED NODE   READINESS GATES
         nginx-9cdf7bbf9-s****   1/1     Running   0          36s   10.0.6.68        virtual-kubelet-cn-hangzhou-j   <none>           <none>
         nginx-9cdf7bbf9-v****   1/1     Running   0          36s   10.0.6.67        virtual-kubelet-cn-hangzhou-k   <none>           <none>

         The result shows that the two pods are deployed on nodes that have the type=virtual-kubelet label, which is specified by the nodeSelector parameter in the Deployment configurations.

   Schedule pods based on pod labels

   1. 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" # Use the compute power of ACS.
              alibabacloud.com/compute-class: general-purpose # The compute class of the ACS pod. Default value: general-purpose.
              alibabacloud.com/compute-qos: default # The QoS class of the ACS pod. Default value: default.
          spec:
            containers:
            - name: nginx
              image: registry.openanolis.cn/openanolis/nginx:1.14.1-8.6
              resources:
                limits:
                  cpu: 2
                requests:
                  cpu: 2

   2. Deploy an NGINX application and query the pods.

      1. Run the following command to deploy an NGINX application:

         kubectl apply -f nginx.yaml 

      2. Run the following command to check whether the NGINX application is deployed:

         kubectl get pods -o wide

         Expected results:

         NAME                    READY   STATUS    RESTARTS   AGE   IP               NODE                            NOMINATED NODE   READINESS GATES
         nginx-9cdf7bbf9-s****   1/1     Running   0          36s   10.0.6.68        virtual-kubelet-cn-hangzhou-j   <none>           <none>
         nginx-9cdf7bbf9-v****   1/1     Running   0          36s   10.0.6.67        virtual-kubelet-cn-hangzhou-k   <none>           <none>

         The result shows that the two pods are deployed on nodes that have the type=virtual-kubelet label, which is specified by the nodeSelector parameter in the Deployment configurations.

2. Check whether ACS pods are created for the NGINX applications.

   1. Run the following command to query the details of a pod created for the NGINX application:

      kubectl describe pod nginx-9cdf7bbf9-s**** 

      The following expected output displays only the key information:

      Annotations:      ProviderCreate: done
                        alibabacloud.com/client-token: edf29202-54ac-438e-9626-a1ca007xxxxx
                        alibabacloud.com/instance-id: acs-2ze008giupcyaqbxxxxx
                        alibabacloud.com/pod-ephemeral-storage: 30Gi
                        alibabacloud.com/pod-use-spec: 2-4Gi
                        alibabacloud.com/request-id: A0EF3BF3-37E7-5A07-AC2D-68A0CFCxxxxx
                        alibabacloud.com/schedule-result: finished
                        alibabacloud.com/user-id: 14889995898xxxxx
                        kubernetes.io/pod-stream-port: 10250
                        kubernetes.io/preferred-scheduling-node: virtual-kubelet-cn-hangzhou-j/1
                        kubernetes.io/resource-type: serverless

      The output shows that the configurations of the pod include the alibabacloud.com/instance-id: acs-2ze008giupcyaqbxxxxx annotation, which indicates that the pod is an ACS pod.

Example

The procedure for using ACS GPU compute power is similar to that for using ACS CPU compute power. However, you also need to ensure that the scheduling components meet the version requirements and add some additional configurations.

...     
     labels:
        # Add labels to request ACS GPU resources.
        alibabacloud.com/compute-class: gpu     #Set to gpu if GPU compute power is used.
        alibabacloud.com/compute-qos: default   #The QoS class, which is the same as regular ACS compute power.
        alibabacloud.com/gpu-model-series: GN8IS  # The GPU model. Specify the actual model that you use.
...

Examples on using ACS GPU-HPN pods in ACK

You can use ACS GPU-HPN pods in the same way as using ACS CPU-accelerated pods. Make sure that the following requirements are met:

• You can use ACS GPU-HPN pods only in ACK managed clusters, ACK One registered clusters, and ACK One Kubernetes clusters for distributed Argo workflows.

• You need to first purchase GPU-HPN capacity reservations and associated them with your clusters.

• You need to update and configure the kube-scheduler and ACK Virtual Node components. The component versions are in invitational preview, submit a ticket to update to these versions.

...     
labels:
  # Add labels to request ACS GPU resources.
  alibabacloud.com/compute-class: gpu-hpn     #Set to gpu-hpn.
  alibabacloud.com/compute-qos: default   #The QoS class.
...