Install the AI suite

Pre-installed MySQL (testing only)

If you do not select ApsaraDB RDS, the cluster uses a built-in MySQL database by default.

When you deploy this component, the cloud-native AI suite creates a 120 GB disk as a PersistentVolumeClaim (PVC) using the StorageClass. ACK does not manage the disk lifecycle. Delete the disk manually when it is no longer needed. For more information, see Release a disk.

ApsaraDB RDS (production)

Use ApsaraDB RDS for production workloads.

If a connection error occurs, see Troubleshoot instance connection failures.

To change the data storage method later, uninstall and reinstall the cloud-native AI suite. If a Secret named kubeai-rds exists in the kube-ai namespace, delete it before reinstalling.

1. Purchase an ApsaraDB RDS instance and create a database and an account. For more information, see Quick start for ApsaraDB RDS. For billing details, see Billing overview.

2. Click Deploy Cloud-native AI Suite at the bottom of the deployment page.

3. In the ACK console, click the cluster name. In the left navigation pane, choose Configurations > Secrets.

4. From the Namespace drop-down list, select kube-ai.

5. Click Create from YAML.

6. Enter the following YAML template to create a Secret named kubeai-rds, then click Create.

   Parameter	Description
   MYSQL_HOST	The connection endpoint of your ApsaraDB RDS instance
   MYSQL_DB_NAME	The name of the database you created
   MYSQL_USER	The database account username
   MYSQL_PASSWORD	The database account password

      apiVersion: v1
      kind: Secret
      metadata:
        name: kubeai-rds
        namespace: kube-ai
      type: Opaque
      stringData:
        MYSQL_HOST: "Your RDS endpoint"
        MYSQL_DB_NAME: "Database name"
        MYSQL_USER: "Database username"
        MYSQL_PASSWORD: "Database password"

Pre-installed MinIO (testing only)

If you do not select OSS, the cluster uses a built-in MinIO instance by default.

When you deploy this component, the cloud-native AI suite creates a 20 GB disk as a PVC using the StorageClass. ACK does not manage the disk lifecycle. Delete the disk manually when it is no longer needed. For more information, see Release a disk.

Object Storage Service (testing and production)

Use Object Storage Service (OSS) for persistent, scalable workflow data storage.

1. If the kube-ai namespace does not exist, create it:

   The kube-ai namespace is created automatically when you deploy the cloud-native AI suite. Verify whether it already exists before creating it.

      kubectl create ns kube-ai

2. In the ACK console, click the cluster name. In the left navigation pane, choose Configurations > Secrets.

3. From the Namespace drop-down list, select kube-ai.

4. Click Create from YAML.

5. Enter the following YAML template and click Create. This creates a Secret named kubeai-oss.

   Parameter	Description
   ENDPOINT	The OSS endpoint for your region. The example uses the China (Beijing) endpoint. For the full list, see OSS regions and endpoints
   ACCESS_KEY_ID	The AccessKey ID of a RAM user with the AliyunOSSFullAccess permission. For more information, see Create an AccessKey pair
   ACCESS_KEY_SECRET	The AccessKey secret of the RAM user

      apiVersion: v1
      kind: Secret
      metadata:
        name: kubeai-oss
        namespace: kube-ai
      type: Opaque
      stringData:
        ENDPOINT: "https://oss-cn-beijing.aliyuncs.com"
        ACCESS_KEY_ID: "****"
        ACCESS_KEY_SECRET: "****"

6. After you create the Secret, verify that a bucket named mlpipeline-<clusterid> appears in the OSS console. This confirms that OSS is configured as the workflow data storage. For OSS billing, see Billing overview.

   

7. Return to the cloud-native AI suite deployment page and install the Kubeflow Pipelines component.