Access the Slice
This topic describes how the user (or team) can access a slice. The EGS admin allocates the slice to the user.
Across our documentation, we refer to the workspace as the slice workspace. The two terms are used interchangeably.
Get Access to a Workspace
A user (or a team) can have one or more workspaces allocated, where AI models are deployed. A workspace can be a single cluster scoped or span across multiple clusters.
The admin is responsible for creating a workspace with user details. The admin provides the user with login details to log in
to the portal. The user is given an access token to access the slice through the user portal. With the access token, the
user can log in and, if needed, download their kubeconfig file that will let them provision their applications
on the workspace/namespace.
To know more, see fetch a workspace token.
Starting with version 1.10.1, an owner role-based API token generated for a workspace also serves as a token to access the User Portal. To know more, see API token.
Login to the User Portal
The user can log in to the portal using an access token (or IDP credentials) or URL. The portal enables workflows to manage the life cycle of GPU provision requests (GPRs) and deep GPU observability for the user AI workloads.
View a Workspace
After a successful login, you see the GPU Requests page as the landing page, where you can view the list of workspaces. A slice is a workspace of namespaces where you can create a GPU request and deploy AI models. The workspace is allocated to a user by the admin.
To view the workspace:
-
Select GPU Requests on the left sidebar.
-
On the GPU Requests page, you see workspaces available to create the GPU requests.
