When launching a notebook there is no information available to the end user to indicate what kind of resources are available such as GPU. If a cluster has 2 GPUs in it and both are claimed by other users/pods a request to start a notebook with a GPU will fail.

This information could be made available to users on the Jupyter launcher page how many GPUs and possibly which type of GPUs are available to be requested.

The limit shown could indicate the total number available in the clusters for each GPU with something like this:

Available GPUs: 1/3 V100, 0/2 A100

Information about quotas for the namespace could also be displayed in a similar fashion to give better feedback to a user when they are about to request something that will fail.