Modes of configuring third-party Editors

The configuration for an Integrated Development Environment (IDE) depends on which type of editor you want to use.

In addition to the native Cloudera Machine Learning editor, you can configure Cloudera Machine Learning to work with third-party, browser-based IDEs, such as Jupyter, and also certain local IDEs that run on your machine, such as PyCharm.

Workbench editor

The Workbench editor is the built-in editor for Cloudera Machine Learning. No additional configuration is required to use it. When you launch a session, select the Workbench editor.

Third-party, browser-based IDEs

Browser IDEs are editors such as Jupyter or RStudio. When you use a browser IDE, it runs within a session and allows you to edit and run code interactively. Changes that you make in the editor are propagated to the Cloudera Machine Learning project. Base Engine Image v8 and higher ships with Jupyter preconfigured as a browser IDE. You can select it when you start a session or add a different browser IDE. For more information, see Configure a Browser IDE as an Editor.

Keep the following in mind when using browser IDEs:

Engine Version Requirements
- Browser-based IDEs require Base Engine Image v8 or higher.
When you are finished using a browser IDE, you must exit the IDE properly, including saving your work if necessary. Do not just stop the Cloudera Machine Learning session. Doing so will cause you to lose your session state. For example, if you want RStudio to save your state, including variables, to ~/.RData, exit the RStudio workspace using the power button in the top right of the RStudio UI.
Depending on the behavior of the browser IDE, multiple users within a project may overwrite each other's state. For example, RStudio state is persisted in /home/cdsw/.RData that is shared by all users within a project.
Browser IDEs do not adhere to the timeout set in IDLE_MAXIMUM_MINUTES. Instead, they use the timeout set in SESSION_MAXIMUM_MINUTES, which is 7 days by default. Cloudera recommends that users stop their session manually after using a browser-based editor. Running sessions continue to consume resources and may impact other users.
Logs for browser IDEs are available on the Logs tab of the session window. This includes information that the IDE may generate, such as error messages, in addition to any Cloudera Machine Learning logs.

Local IDE Editors on your machine that can use SSH-based remote editing

These editors, referred to as Local IDEs in the documentation, are editors such as PyCharm that run on your local machine. They connect to Cloudera Machine Learning with an SSH endpoint and allow you to edit and run code interactively. You must manually configure some sort of file sync and ignore list between your local machine and Cloudera Machine Learning. You can use functionality within the local IDE, such as PyCharm's sync, or external tools that can sync via the SSH endpoint, such as Mutagen.

Keep the following in mind before setting up local IDEs:

Local IDEs do not require a specific engine image, but Cloudera always recommends you use the latest engine image.
Site Administrators should work with IT to determine the data access policies for your organization. For example, your data policy may not allow users to sync certain files to their machines from Cloudera Machine Learning. Verify that users understand the requirements and adhere to them when configuring their file sync behavior.
Users should ensure that any IDEs that the IDEs they want to use support SSH. For example, VS Code supports "remote development over SSH," and PyCharm supports using a "remote interpreter over SSH."