Launch a Session

Sessions allow you to perform actions such as run R or Python code. They also provide access to an interactive command prompt and terminal. This topic demonstrates how to launch a new session.

  1. Navigate to your project's Overview page.
  2. Click New Session.
    New projects now default to using ML Runtimes. Legacy Engines are deprecated in the current release. However, you can change the default engine later in this task.
  3. Check the settings for your session:
    If your project is using ML Runtimes, you will see the following settings:
    Editor
    Selects the Editor; currently only Workbench is supported and therefore the selector is static.
    Kernel
    Selects the Kernel, for example Python 3.7, R4.0.
    Edition
    Selects the Runtime Edition. Initially only Standard variants are supported.
    Version
    Selects the ML Runtimes version.
    If your project is using Legacy Engines, you will see the following settings:
    Editor
    Selects the Editor; currently only Workbench is supported and therefore the selector is static.
    Kernel
    Selects the Kernel. Initially only Python Runtimes are supported.
    Engine Image
    Displays the Advanced tab in Project Settings and allows you to set environment variables and the shared memory limit.
  4. If your project is using Legacy Engines, you can modify the engine image used by this session:
    1. By Engine Image, click Configure.
      Cloudera Machine Learning displays the Project Settings page.
    2. Select the Runtime/Engine tab.
    3. Next to Default Engine, select ML Runtime or Legacy Engine.
    4. Click Save Engine.
  5. Specify your Resource Profile.
    The minimum configuration is 1vCPU and 2 GB memory.
  6. Click Start Session.
    The command prompt at the bottom right of your browser window will turn green when the engine is ready. Sessions typically take between 10 and 20 seconds to start.