As a admin, I need to make sure when users start notebook servers of different sizes, they can be supported by being placed on appropriate worker nodes. For example, I might have 1-2 worker nodes that are large enough to accommodate an XL notebook server. I want to ensure when a user attempts to start an XL notebook server, it gets placed on a node with an appropriate taint to accept XL notebooks. Other smaller notebooks can share the resources across the other worker nodes that might not have a taint for an XL notebook. This can help ensure users that need to do work with large notebook sizes, the resources will be available to support them.

This story covers a broader, more generic need to utilize taints & tolerations for notebook pods to ensure notebook sessions land on the appropriate worker nodes.

This was feedback from Beta tester - Boehringer-Ingelheim.

Notes on this story:

• Want to enable customers to dedicate designated machine pools to notebook pods only.
• We will do this after we move to the Kubeflow notebook controller
• Expectation is that this will be an admin UI config option

Req:

• The system must allow RHODS admins to define tolerations for notebook pods via an admin UI.