Skip to content

How do I delete/shut down ml.m5.2xlarge-Hosting instance from SageMaker AI?

0

Hi, I used SageMaker AI on May 30th to create a domain and Jupyter notebook instance. I used this model with single instance - Hosting ml.m5.2xlarge - US East (N. Virginia). I have since deleted the domain, the notebook and the model instance. I am still being charged everyday. How do I delete this hosting instance? Where can I find it under SageMaker AI? I checked the SageMaker AI dashboard and everything there is deleted including endpoint configurations. Is there any place outside Sagemaker that I need to check?

I don't see the ML Ops in the left navigation panel under Sagemaker Canvas. As I don't have a domain, I only see create domain options under SageMaker Canvas. I do remember deploying the application. Where do I find the deployment to delete it?

Please help. Thank you.

asked a year ago209 views

2 Answers
0

To delete a SageMaker AI hosting instance (ml.m5.2xlarge) that's still incurring charges, you need to ensure that all associated endpoints are properly deleted. Since you've already checked the SageMaker AI dashboard and confirmed that endpoints and configurations are deleted, here are additional steps to take:

  1. Check for any remaining model deployments in SageMaker Canvas:
  • Open the SageMaker Canvas application
  • Navigate to ML Ops in the left panel
  • Select the Deployments tab
  • If you find your deployment, select it, click the "More options" icon, and choose "Delete deployment"
  • Confirm deletion in the dialog box
  1. Check for Kubernetes-based hosting deployments:
  • If you used Kubernetes with SageMaker, you can delete hosting deployments using kubectl:
  • Use command: kubectl delete hostingdeployments <your-deployment-name>
  • Verify deletion with: kubectl get hostingdeployments
  1. Use AWS CLI to ensure all applications are deleted:
  • Delete any JupyterLab applications: aws --region <your-region> sagemaker delete-app --domain-id <your-domain-id> --app-name default --app-type JupyterLab --space-name <your-space-name>
  • Delete any spaces: aws --region <your-region> sagemaker delete-space --domain-id <your-domain-id> --space-name <your-space-name>
  1. Check AWS CloudFormation for any remaining stacks related to your SageMaker deployment and delete them.

If you've completed all these steps and are still seeing charges, I recommend using AWS Cost Explorer to identify exactly which resources are generating the charges. This will help pinpoint any resources that might have been missed during cleanup.

Remember that deleted endpoints should not incur any charges for SageMaker AI resources, so if charges continue, there's likely another resource still active.
Sources
Delete a model deployment - Amazon SageMaker AI
Where to shut down resources per SageMaker AI features - Amazon SageMaker AI
Use Amazon SageMaker AI Jobs - Amazon SageMaker AI
Delete unused resources - Amazon SageMaker AI

answered a year ago

  • I don't see the ML Ops in the left navigation panel under Sagemaker Canvas. As I don't have a domain, I only see create domain options under SageMaker Canvas. I do remember deploying the application. Where do I find the deployment to delete it?

0

I don't see the ML Ops in the left navigation panel under Sagemaker Canvas. As I don't have a domain, I only see create domain options under SageMaker Canvas. I do remember deploying the code. Where do I find the deployment so I can delete it? I had used Sagemaker Studio to create a single user domain. And I have deleted the domain that I had created.

answered a year ago

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.