Real-Time Collaboration in Jupyter

JupyterLab supports real-time collaboration, allowing multiple users to work simultaneously in the same notebooks and files. This feature can be useful for pair programming, live demonstrations, or training sessions where multiple people need to view and edit the same notebook at the same time.

Prerequisites

Before setting up real-time collaboration, you need to:

• Have a Jupyter server resource created (see Jupyter Servers)
• Install the Jupyter collaboration extension
• Configure access permissions for collaborators

Install Jupyter Collaboration Extension

To enable real-time collaboration, you need to install the jupyter-collaboration package into the environment that runs JupyterLab itself (not the data science environment). This must be done using the Start Script.

Add the following to your resource’s Advanced Settings > Start Script:

/opt/saturncloud/bin/pip install jupyter-collaboration

You can add this when creating a new resource, or by editing an existing resource:

1. Navigate to your Jupyter server resource page
2. Click on the resource name to edit settings
3. Expand Advanced Settings at the bottom of the form
4. In the Start Script (Bash) section, add the pip install command above
5. Click Update to save your changes
6. Restart your resource for the changes to take effect

Using /opt/saturncloud/bin/pip ensures the package is installed directly into the JupyterLab environment where it can be loaded.

Configure Access Permissions

Once the collaboration extension is installed, you need to grant other users access to your Jupyter server. There are two approaches:

Option 1: Add Viewers to Your Jupyter Server

If you own the Jupyter server, you can add specific users or groups as viewers:

1. Navigate to your Jupyter server resource page
2. Look for the Viewers section in the resource details
3. Click Add Viewer
4. Select the users or groups you want to grant access to
5. Save your changes

Option 2: Use a Group-Owned Jupyter Server

Instead of individually adding viewers, you can create a Jupyter server owned by a group. All members of the group will automatically have access to the server:

1. When creating a new Jupyter server, set the Owner field to a group instead of your user account
2. All group members will be able to start, stop, and access the server
3. Any group member can collaborate in real-time when the server is running

For more information about groups, see the Groups documentation.

Using Real-Time Collaboration

Once the extension is installed and access is configured, collaborators can work together:

1. The resource owner starts the Jupyter server
2. Collaborators navigate to the same resource page in Saturn Cloud
3. All users can open the same notebooks and files
4. Changes made by any user appear in real-time for all collaborators
5. Each user’s cursor and selections are shown with different colors

Collaboration Features

• Simultaneous editing: Multiple users can edit the same notebook cells
• Live cursors: See where other users are working with colored cursors
• Automatic sync: Changes are synchronized automatically across all users
• Shared execution: Code execution happens on the shared resource

Best Practices

• Communication: Use a separate communication tool (Slack, Zoom, etc.) to coordinate with collaborators
• Resource size: Ensure the Jupyter server has sufficient resources for multiple users
• Save frequently: Although changes sync in real-time, regular saves are recommended
• Git integration: Combine real-time collaboration with Git for version control
• Clear ownership: Establish clear guidelines for who can start/stop group-owned resources

Limitations and Considerations

• Performance: Real-time collaboration adds overhead. Use appropriately sized instances for your team
• Network dependency: Requires stable internet connections for all collaborators
• Merge conflicts: Simultaneous edits to the same cell can be confusing. Coordinate with collaborators
• Not a replacement for Git: Real-time collaboration doesn’t provide version control, code review, or persistent history

When to Use Real-Time Collaboration

Real-time collaboration works best for:

• Pair programming sessions
• Live training or demos
• Debugging sessions with team members
• Interactive code reviews
• Short-term collaborative analysis

For asynchronous collaboration, code reviews, and production workflows, use Git and resource recipes instead.