CDK solution to update SageMaker endpoints with auto-scaling
In brief
- Updating SageMaker endpoints with auto-scaling fails via CloudFormation/CDK without prior deregistration.
- A CDK construct automates deregistration and re-registration of auto scaling during deployments.
- The solution uses three custom resources and a ConfigHash to synchronize parameters.
AWS CloudFormation and AWS CDK deployments fail when updating an Amazon SageMaker AI endpoint configured with Application Auto Scaling. According to the announcement, SageMaker blocks the update as long as the scalable target remains registered, causing deployment errors.
This post introduces a reusable CDK construct that combines CloudFormation custom resources to automatically deregister and re-register auto scaling during endpoint updates. The approach uses three custom resources managing operation order through explicit dependencies in CDK.
This solution prevents deployment failures and corrupted states, ensuring auto scaling remains active after updates. A sample stack is available for immediate implementation or integration into existing projects.
What it changes for you
- Public SectorPublic administrations using SageMaker for public services must update their CI/CD pipelines to prevent service disruptions during endpoint updates.
This summary gives you the essentials. The full article, its figures and quotes are on Artificial Intelligence.
Read the original articleAI-generated summary of the source, checked by rules. How we write · Report an error