Import Alibaba Cloud Cluster

Preconditions

1. Obtain the Alibaba Cloud Access Key and Access Key Secret and have the following permissions. For the specific application process, please refer to Create Access Key.

   Services	Permissions	Description
   Cloud Server (ECS)	ecs:DescribeDisks ecs:DescribeInstances ecs:DescribeRegions	This permission is required to obtain disk, instance and region information
   Container Service Kubernetes version (CS)	cs:DescribeClusterNodePools cs:DescribeClusterNodePoolDetail cs:DescribeClusters cs:DescribeClusterUserKubeconfig cs:DescribeClusterNodes	This permission is required to obtain the cluster, node group, Node, Kubeconfig and other information

   The permission policy file is described as follows. You can import the policy directly in the Alibaba Cloud console. For details, please refer to Creating a custom permission policy through script editing mode

   {
     "Version": "1",
     "Statement": [
       {
         "Effect": "Allow",
         "Action": [
           "ecs:DescribeDisks",
           "ecs:DescribeInstances"
           "ecs:DescribeRegions"
         ],
         "Resource": [
           "*"
         ],
         "Condition": {}
       },
       {
         "Effect": "Allow",
         "Action": [
           "cs:DescribeClusterNodePools",
           "cs:DescribeClusterNodePoolDetail",
           "cs:DescribeClusters",
           "cs:DescribeClusterUserKubeconfig",
           "cs:DescribeClusterNodes"
         ],
         "Resource": [
           "*"
         ],
         "Condition": {}
       }
     ]
   }
   

2. Alibaba Cloud Kubernetes cluster Kubeconfig. For cluster versions 1.22 and above, it is recommended to use ACK cluster. For details, please see Container Service Kubernetes.

3. Kubeconfig requires the following permissions.

   Resources	Permissions	Description
   Node	List Get Watch	Get node information
   Pod	List Get Watch	Get Pod information
   Namespace	List Get Watch	Get Namespace information
   Service	List Get Watch	Get Service information
   Persistent Volume Persistent Volume Claim	List Get Watch	Get Persistent Volume, Persistent Volume Claim information
   Deployment StatefulSet DaemonSet ReplicaSet ReplicationControllers	List Get Watch	Get Deployment, StatefulSet, DaemonSet, ReplicaSet , ReplicationContoller information
   Job CronJob	List Get Watch	Get Job and CronJob information

   The permissions file is as follows:

   apiVersion: rbac.authorization.k8s.io/v1
   kind: ClusterRole
   metadata:
     name: optimizer
   rules:
     - apiGroups:
         - ""
       resources:
         -nodes
         -pods
         -pods/status
         - persistentvolumes
         - persistentvolumeclaims
         - services
         - namespaces
       verbs:
         - get
         - list
         - watch
     - apiGroups:
         - "apps/v1"
       resources:
         -deployments
         -replicasets
         -replicationcontollers
         - statefulsets
         -daemonsets
       verbs:
         - get
         - list
         - watch
     - apiGroups:
         - "batch/v1"
       resources:
         - jobs
         - cronjobs
       verbs:
         - get
         - list
         - watch
     - apiGroups:
         - "storage/v1"
       resources:
         -storageclasses
       verbs:
         - get
         - list
         - watch
   

4. Prometheus monitoring or Thanos monitoring. This monitoring is used to provide monitoring data for nodes, containers, etc. in the cluster.

Import steps

Add Alibaba Cloud account

1. Select the Settings menu and select the Cloud Account tab to enter the cloud account management page. By default, this page will display all cloud accounts managed by the platform in pages, including names and corresponding cloud service providers. It also provides an operation bar to facilitate users to perform corresponding operations, as shown in the figure below.

   

2. On the cloud account management page, click the Add button to add a cloud account. The interface is as shown below. In this interface, the user enters the cloud account name, selects the corresponding cloud service provider, enters the Access Key and Access Key Secret, and clicks the OK button to complete adding the cloud account. After adding parameters, users can also click the Test Connectivity button to test whether the cloud account is correct. The platform currently supports Alibaba Cloud, which can be selected from the cloud service provider drop-down box. Access Key and Access Key Secret is the account information for accessing the cloud service. Please obtain it from the corresponding cloud service provider.

   

Add Alibaba Cloud cluster

1. Select the Settings menu and select the cluster tab to enter the cluster management page. By default, this page will display all clusters managed by the platform in pages, including name, cluster ID, cluster region, Access Token name, corresponding cloud service provider and Prometheus address. It also provides an operation bar to facilitate users to perform corresponding operations. As shown below.

   

2. Click the Import Cluster button to pop up the import dialog box, where you can import the cluster. There are 5 steps to import a cluster. Currently, only Alibaba Cloud clusters are supported.

   1. Set Alibaba Cloud access credentials
   2. Connect to the Kubernetes cluster
   3. Connect to Prometheus
   4. View configuration
   5. Complete

3. The interface for setting Alibaba Cloud access credentials is as shown in the figure below. In the Alibaba Cloud access credentials interface, set the Alibaba Cloud access credentials by using existing credentials. Here, select the Alibaba Cloud account added in Add Alibaba Cloud Account, as shown in the figure below.

   

4. After setting the Alibaba Cloud access credentials, click the Next button to enter the page for connecting to the Kubernetes cluster, as shown in the figure below. The platform supports connecting to two types of Kubernetes clusters, including managed clusters and self-built clusters. The Connect Hosting Cluster page is shown in the figure below. Here we choose the managed cluster. The platform automatically obtains the user cluster through the cloud account. The user selects the corresponding cluster through the cluster drop-down box, enters the description field, selects whether to use the API Server private network address, selects whether to automatically refresh KubeConfig and the KubeConfig life cycle. Finally enter the node group label.

   

5. The connection page to Prometheus is as shown in the figure below. After the user enters information on this page, click the Next button to enter the view configuration page. The input information details are shown in the table below.

   

   The parameter description for connecting to Prometheus is as shown in the table:

   Parameters	Description
   Whether Thanos	If the backend cluster is Thanos, this should be checked.
   Prometheus address	Prometheus address, must be accessible.
   Limit access rate	If you need the current platform’s access rate to Prometheus, you should check this option.
   Maximum number of concurrencies	The maximum number of concurrencies for the platform to access Prometheus.
   Cluster label	Cluster label is used to filter the monitoring indicators of the cluster from Prometheus or Thanos
   Cluster tag value	Cluster tag value is used to filter the monitoring indicators of the cluster from Prometheus or Thanos

6. View the configuration page as shown in the figure below. This page displays Alibaba Cloud access credentials, detailed parameters for connecting to the Kubernetes cluster and connecting to the Prometheus cluster, which is convenient for users to confirm. When the user completes the confirmation, click the Import button to enter the completion page.

   

7. The completed page is as shown below, and the submission status will be displayed on the page. Click Finish to complete the cluster import.

   

Next steps

After importing the cluster, wait for a period of time (waiting for the optimization strategy to complete scheduling and execution), and then you can view optimization suggestions for the cluster in multiple dimensions, including:

• Nodegroup
• Node
• GPU Node
• Disk
• Persistent volume
• Application