-
Notifications
You must be signed in to change notification settings - Fork 96
New Streamlined Authentication flow in Azure ML ‐ Remote extension
Shantnu Suman edited this page Jan 12, 2024
·
18 revisions
The Azure ML and Azure ML - Remote extensions starting version 0.44.0 have undergone an authentication update to improve security and flexibility. The old flow, which eventually depended on the Azure Account extension to obtain authentication tokens has been replaced with native VSCode authentication providers.
The native VSCode authentication has multiple benefits:
- It eliminates our dependency on Azure Account extension, therefore eliminating the dependency of remote connections on Azure ML extension, enabling us to support MSAL authentication provider in both the extensions.
- It allows the user to reuse the VSCode sign in sessions for all the extensions i.e. if you are already signed in to your account (e.g. for settings sync), you can use the same for your extensions.
- Make sure you have the latest version of VSCode installed.
- Make sure you have the Azure ML and Azure ML - Remote extension's version updated to a minimum of
0.44.0
.
- Open VS Code and ensure you are signed into your account. You may sign in using the
Azure ML: Sign In
command (accessible via the command palette). - Run
Azure ML: Sign In to Directory
. - Select a tenant from the list of displayed tenants.
- Complete the additional sign in steps.
- Open VSCode settings
- Navigate to
Workspace -> Extensions -> Microsoft Sovereign Cloud
. - Alternatively, you can search for setting
microsoft-sovereign-cloud.environment
in the settings search bar. - Select your cloud under the setting
Microsoft-sovereign-cloud: Environment
underUser
settings.
- If you are already signed in with an Azure Account, signing in with a new Azure account without signing out from the existing one makes the subscriptions inaccessible from the extensions. See here
- When changing the
microsoft-sovereign-cloud.environment
setting, the tree view is not refreshed automatically and you have to refresh it manually. Additionally, the Sign In button in the tree view or the Sign In command in the command palette won't work immediately until, you have reloaded the window - Azure ML extension is unable to pick other Azure extensions' sign in and you have to sign in separately for this extension