Studio CLI

A small light-weight cli to manage users and teams for hackathons using SageMaker Studio Spaces

Prerequisites

• A SageMaker Studio Studio Domain created, with IAM (not Identity Center) to manage access.
• Active AWS credentials in the current environment in the same account as the SageMaker Studio domain

Installation

You can install the CLI using pip, or use Docker to use the CLI in a containerized environment

Using pip

In the root of the directory, run

pip install .

Using Docker (Optional)

For convenience, and to avoid installing the CLI/Python dependencies on your machine you can optionally use Docker to run the CLI.

docker build . -t studio-cli # Build the Docker image and tag it as "studio-cli"
docker run -it --rm studio-cli # Run the CLI in a container (and remove the container when exiting)

This will launch an interactive shell inside the Docker container with the CLI pre-installed. From there you can run any of the CLI commands as documented below.

Example usage

Here's how to use some of the more useful commands

Configure the CLI

studio configure

Run this when

• you configure the tool before a hackathon

Persists the configuration, including region, Studio domain ID and DDB table name. It creates a DDB table if none exists.

Show current configuration

studio get-conf

Setup users and teams

studio setup-users users.csv

where users.csv is a csv with 2 columns, an email and a team number. See sample.

Run this when

• you have all participants and team divisions before an event, to create SM user profiles and bootstrap spaces.

Get presigned URLs

studio get-urls

Run this when

• you want to generate presigned URLs (valid for 5 minutes) to distribute to hackathon participants.

Note

This command is not necessary if you use the web-app to grant participants access to their environment.

Finish an event

studio purge

Tries to delete all hackathon-related resources.

Run this when

• The event is over to delete all SM user profiles and SM spaces and apps.

Known Issues

Warning

Below are some known issues

Purge: The purge command tries to use the API to delete all related resources. There are a lot of dependencies between the different resources, meaning that some resources can't be deleted untill others are. i.e a space cannot be deleted untill all apps runnning in that space have been deleted. To avoid having a command running for 30 minutes, some resource deleteions are skipped in case there are deletions in progress for resources it's depending on. This mean that the purge command may have to be run several times, preferably with some time in between.

Same Email: SM user profiles are created based on the email. If users share the first part of the email, i.e [email protected] and [email protected], the last occuring person in the list will not get a user profile created. This is known issue with the current implementation.

Limits

These are some, but not an extensive list of AWS service limits you need to be mindful of.

• Maximum number of Studio user profiles allowed per account. (This maps to the number of participants)
• Maximum number of Studio spaces allowed per account. (This maps to the number of teams or number of instances that will)
• Studio JupyterLab Apps running on instances. (These are the underlying instances. Note that the ml instances are different to the EC2 instances, with different qoutas.)
• Studio CodeEditor Apps running on instances. (These are the underlying instances. Note that the ml instances are different to the EC2 instances, with different qoutas.)

A general recommendation is to test well in advance of the hackathon to be able to hash out what limits need to be adjusted.

Web App

If you want hackathon participants to have a lightweigh web app where they can enter their email and get redirected to their teams SageMaker Studio Space, follow the instructions in web-app/README.md

Caution

The web-app is a passwordless frontend where participants simply enter their email to get access.