Workbench

Workbench is a hosted JupyterLab environment that combines a ready-to-go environment with custom integrations to the Descartes Labs Platform APIs.

Starting a Server

When you log in to https://workbench.descarteslabs.com for the first time, you’ll be presented with the hub home, where you can choose to start (and later stop or connect to) your jupyerlab server. Clicking the Start My Server button presents you with a choice of the type of compute instance you would like to use – either a Model exploration/Development or a Model Training instance. Choose the environment that is best suited for your work. You can always stop and start a different type of server if you decide to change. Your home directory will persist across instance types.

Getting Started

To get started you will need to go through a one-time initialization of your environment in order to set up the proper authentication token for future API calls with the Python client library.

A notebook named "01_initial_setup.ipynb" is located in the "example_notebooks/" directory in your filesystem browser. Open the notebook and follow the instructions to add your credentials to the home directory of your JupyterLab environment. Alternatively, you can open a terminal in JupyterLab, execute the command descarteslabs auth login, and follow the instructions listed there.

Pre-loaded Examples

Upon server startup, a set of examples are copied to the example_notebooks folder in your home directory. These examples are a good starting point for exploring and understanding how you can use the Descartes Labs Platform APIs and visualization tools. These examples are refreshed at server start time, and will overwrite contents in the example_notebooks directory. Avoid saving work to that directory.

Custom Integrations

The Workflows client library is integrated with JupyterLab via custom jupyter widgets. For example, a workflows.Image can be displayed on an ipyleaflet-based map by running .visualize(...) and displayed by evaluating descarteslabs.workflows.map at the end of a code cell.

Server Lifecycle

Your JupyterLab server will occasionally be shut down if it is deemed inactive (based on interactions with the JupyterLab interface). To restart, simply go through the same steps to start a server. You will not need to re-initialize your environment. To manually shut down a server, navigate to the hub home at https://workbench.descarteslabs.com/hub/home and click the Stop My Server button. After a few seconds the server will shut down, at which point you could either log out or start a server of a different type, depending on what you intend to do next.

Data Persistence

All data in your ${HOME} (/home/jovyan) directory is persisted across servers. Any modifications to the rest of the system will lost at server shutdown/restart. Example notebooks are loaded at server start time, and will overwrite contents in the example_notebooks directory. Avoid saving work to that directory.