Update the SageMaker AI Distribution Image - Amazon SageMaker AI

Update the SageMaker AI Distribution Image

Important

This topic assumes that you've created a space and given the user access to it. For more information, see Give your users access to spaces.

Update the JupyterLab spaces that you've already created to use the latest version of the SageMaker Distribution Image to access the latest features. You can use either the Studio UI or the AWS Command Line Interface (AWS CLI) to update the image.

The following sections provide information about updating an image.

Updating the image involves restarting the JupyterLab space of your user. Use the following procedure to update your user's JupyterLab space with the latest image.

To update the image (UI)

  1. Open Studio. For information about opening Studio, see Launch Amazon SageMaker Studio.

  2. Choose JupyterLab.

  3. Select the JupyterLab space of your user.

  4. Choose Stop space.

  5. For Image, select an updated version of the SageMaker AI Distribution Image. For the latest image, choose Latest.

  6. Choose Run space.

This section assumes that you have the AWS Command Line Interface (AWS CLI) installed. For information about installing the AWS CLI, see Install or update to the latest version of the AWS CLI.

To update the image, you must the do the following for your user's space:

  1. Delete the JupyterLab application

  2. Update the space

  3. Create the application

Important

You must have the following information ready before you start updating the image:

  • domain ID – The ID of your user's Amazon SageMaker AI domain.

  • Application type – JupyterLab.

  • Application name – default.

  • Space name – The name specified for the space.

  • Instance type – The Amazon EC2 instance type that you're using to run the application. For example, ml.t3.medium.

  • SageMaker AI Image ARN – The Amazon Resource Name (ARN) of the SageMaker AI Distribution Image. You can provide the latest version of the SageMaker AI Distribution Image by specifying either sagemaker-distribution-cpu or sagemaker-distribution-gpu as the resource identifier.

To delete the JupyterLab application, run the following command:


aws sagemaker delete-app \
--domain-id your-user's-domain-id 
--app-type JupyterLab \
--app-name default \
--space-name name-of-your-user's-space

To update your user's space, run the following command:


aws sagemaker update-space \
--space-name name-of-your-user's-space \
--domain-id your-user's-domain-id

If you've updated the space successfully, you'll see the space ARN in the response:


{
"SpaceArn": "arn:aws:sagemaker:AWS Region:111122223333:space/your-user's-domain-id/name-of-your-user's-space"
}

To create the application, run the following command:


aws sagemaker create-app \
--domain-id your-user's-domain-id  \
--app-type JupyterLab \
--app-name default \
--space-name name-of-your-user's-space \
--resource-spec "InstanceType=instance-type,SageMakerImageArn=arn:aws:sagemaker:AWS Region:555555555555:image/sagemaker-distribution-resource-identifier"