informatics-lab / aml-jupyterhub

Code work and experiments to integrate Azure Machine Learning with JupyterHub.
6 stars 4 forks source link

Allow users to select workspace / VM size #30

Closed nbarlowATI closed 4 years ago

nbarlowATI commented 4 years ago

After the user logs in via Azure AD, would be good to offer them a choice of which Workspace to use (or if they want to create a new one). Can also offer a list of VM sizes for them to choose from.

One thing to consider is that while it is not too difficult for the Service Principal to get a list of all Workspaces in the Resource Group, it is not quite so simple to filter this list to only the ones that the user has access to.

pierocor commented 4 years ago

The option to create a new workspace has been added to the form page. When selecting this option the redirection to the progress page could take minutes and, in most cases, an error appears:

ngrok gateway error
The server returned an invalid or incomplete HTTP response.

However, the logs show that both the WS and the CI are created successfully. The user has to manually go to $HOST/hub to be redirected to the CI.

tam203 commented 4 years ago

Hi @pierocor - I was very impressed with your demo on this last week. My suggestions to streamline the process from a 'scientist who doesn't know or care about Azure' point of view would be:

Two options on the spawner page:

They can revisit this at any time at <jupyterhub-url>/hub/home if we enable named servers

We discussed that what projects are available would be ascertained by what permissions the use had (i.e. what projects/resource group/workspace they could access).

The above permission management would be simplified by a few scrips we could write maybe: add user to project, create project, soft delete project (remove the user's permissions but don't delete the resources(resource group)) and hard delete (delete all the project assets.

nbarlowATI commented 4 years ago

Closing, as outstanding part of this Issue is specified in more detail in #40