Kele's current implementation, for the most part, makes the simplistic and naive assumption that the user has full permissions to do whatever they want with the cluster in question. This is, more often than not, not the case.
Many of Kele's existing bugs and feature gaps are exactly due to this assumption (#191, #152). While some work has been done to streamline UX w.r.t. auth (#184 and #181), there are many "unknown unknowns" w.r.t. auth.
The Task
Create a solution for spinning up integration-test clusters with varying degrees of auth. For example:
A cluster where the user has certain kind-verb pairs forbidden;
A cluster where the user has absolutely no permissions whatsoever
Context
Kele's current implementation, for the most part, makes the simplistic and naive assumption that the user has full permissions to do whatever they want with the cluster in question. This is, more often than not, not the case.
Many of Kele's existing bugs and feature gaps are exactly due to this assumption (#191, #152). While some work has been done to streamline UX w.r.t. auth (#184 and #181), there are many "unknown unknowns" w.r.t. auth.
The Task
Create a solution for spinning up integration-test clusters with varying degrees of auth. For example: