v0.23.0

Fixes the lifecycle scripts for SageMaker Studio and SageMaker Notebook Instances to point to the right release artifact.

v0.22.0

JupyterLab 3.x support

The GUI plugin is now updated to support JupyterLab 3.x and, therefore, supports the current versions of SageMaker Notebook Instances and SageMaker Studio notebooks.

As of this release, the JupyterLab extension is no longer compatible with JupyterLab 2.x. If you're using JupyterLab 2.x, please use release v0.21.0.

Additional small features

• Add a real lifecycle script for SageMaker Studio notebooks so you can install the plugin permanently in your domain.
• Support for China regions (thanks to @yosefbs).
• Update default container environment to Python 3.10.

v0.21.0

This release is just security patches in the dependencies.

v0.20.0

This release provides support for JupyterLab 2.x.

As of this release, the JupyterLab extension is no longer compatible with JupyterLab 1.x. If you're using JupyterLab 1.x, please use release v0.19.0.

v0.19.0

This release has a number of small improvements and bug fixes.

After upgrading to this release, we recommend that you delete and recreate your infrastructure with the following commands:

1. aws cloudformation delete-stack --stack-name sagemaker-run-notebook
2. run-notebook create-infrastructure

This will pick up the new managed policies (see under [Bug Fixes])(#bug-fixes) below) which can not be updated with the --update option.

Bug fixes

• Fixed an install failure on versions >= 3.10. (Fixes issue #35)
• Use managed policies so that we can include the policies in other roles (See Change policies to managed policies by @dmoser04).
• Use a single permission statement on the Lambda function for all EventBridge rules. This prevents us overflowing the number of separate permissions when we have many scheduled notebook runs. This also means that we're not creating permissions at schedule time. (Fixes issue #9)
• Correctly handle scheduled runs with no supplied parameters. (Fixes issue #25)
• Update various JS dependencies for security fixes.

v0.18.0

This is a documentation only release:

• It adds the documentation tarball (docs.tar.gz) to the release.
• It adds reasonable docstrings to all the exported functions in sagemaker_run_notebook.

v0.16.0

Users moving to this version should update their Lambda function and permissions by running:

$ run-notebook create-infrastructure --update

Features

• Added the ability to run notebooks connected to an EMR cluster using SparkMagic. See the EMR examples for more information on using this feature.
• Add the permissions to the default execution policy and role that allow notebook executions to attach to user VPCs via the --extra option.

Bug fixes

• Pass environment variables specified in --extra parameters through to the notebook execution.
• Restrict S3 permissions in the default role to buckets whose name contains the word "SageMaker".
• Unpin the version of the Python "requests" library because it is (a) no longer necessary and (b) cause a dependency error with the latest version of boto3.

v0.15.0

This release primarily fixes a regression in v0.14.0 which caused "No such kernel" errors when running the notebook (Issue #4).

Two other small changes:

• Added a -v flag to run-notebook to show the current version.
• Changed the install scripts for SageMaker notebooks to install directly from GitHub so you don't need to specify an S3 bucket.

v0.14.0

Added:

• R Support
• More tools for building containers
• Ability to use all arguments to SageMaker Processing Jobs.

See CHANGELOG.md for details.

v0.13.0

The initial public release of the SageMaker notebook execution and scheduling tools.