Configuration
Users with Administrator permission can access MLSpace’s Configuration page. This page is divided into three sections:
Dynamic Configuration
Admins can manage several features of MLSpace without needing to make code changes or redeploy the application. The sections below will explore the different type of features that can be configured.
Using the actions button on the right side of the page, an admin can:
- Import Configuration: upload a JSON file with the correct schema and it will be deployed as the current dynamic configuration for MLSpace.
- Export Configuration: downloads a JSON file containing all of the current dynamic condifigurations for MLSpace. This is useful if making a new deployment of MLSpace and you wish to re-use the current configurations.
- Expand All: expands all collapsed sections on the page so everything is in its expanded state.
At the bottom of this page is a "Save Changes" button. Upon clicking this, the user will be shown a brief summary of the fields that have been changed and will be updated when this new dynamic configuration is uploaded.
Allowed Instance Types
Modify the Amazon EC2 instance types that you wish to make available to users within the following resources: Notebook instance, Training and HPO jobs, Transform jobs, and Endpoints. IAM permissions that control access to these instance types within the MLSpace user interface and Jupyter Notebooks will automatically update.
Note: if Secure User Scoped Roles are not being used (MANAGE_IAM_ROLES is set to false) then all instance types will always be allowed and this feature will be unavailable.
Activated Services
Activate or deactivate services within MLSpace. IAM permissions that control access to these services within the MLSpace user interface and Jupyter Notebooks will automatically update. Deactivated services will no longer appear within the MLSpace user interface or be available for use within Notebooks. Deactivating services will suspend all active corresponding jobs and instances associated with the service.
Note: if Secure User Scoped Roles are not being used (MANAGE_IAM_ROLES is set to false) then all services will always be active and this feature will be unavailable.
EMR Config
Configure the settings that EMR Clusters deployed through the MLSpace user interface are deployed with.
Applications
Manage the applications available to EMR clusters that users create through the MLSpace user interface.
Cluster Types
Define or modify the cluster types that user can choose from when they create EMR clusters through the MLSpace user interface.
Auto Scaling Policy
An automatic scaling policy for a core instance group or task instance group in an Amazon EMR cluster. The automatic scaling policy defines how an instance group dynamically adds and terminates Amazon EC2 instances.
Project Creation
All users can create new projects in MLSpace by default. Restrict project creation to users with Admin permissions by activating “Admin Only”.
System Banner
When active, a banner will be displayed along the top and bottom of the MLSpace user interface. The color of the text and the banner itself can be selected as well.
Dynamic Configuration History
Admins can see the history of dynamic configurations for MLSpace. Version 0 is the default configuration deployed with MLSpace. When an update is made to the dynamic configuration, a new row is created in this table and the Version value is incremented by 1. If at any point an admin wishes to revert to a prior dynamic configuration they can click the "Rollback" for that version, and that version will become the current dynamic configuration for the application.
Deployment Configuration
Administrators can view the current deployment-time configurations for MLSpace. If configuration settings need to be changed, a system admin developer will need to make those modifications and redeploy the application.