Hi, I currently have a cluster running on AWS with Zero-to-JupyterHub-k8s installed on it. Currently the hub is not using any authentication and I want to start using the Native Authenticator. By doing this, will I lose the accounts and subsequently the PVC’s of the users already using the Hub? The users have important documents on their respective PV’s and cannot afford to lose those. If yes, is there a way to restore the PVC’s of those users to their new user accounts?
Does switching to a new authenticator on JupyterHub on Kubernetes (Zero-to-JupyterHub-k8s) delete all the old users?
Hi! Could you explain what you mean by “the hub is not using any authentication”? JupyterHub always has some form of authenticator- it has to in order to identify a user.
This is the auth part of my config.yaml file that I am using with Zero-to-JupyterHub-k8s helm installation
auth: admin: users: - X - Y - Z
As of now, These three users have admin access and users can log-in with their names. But passwords are not being saved (because I haven’t specified any authentication). Anyone can log-in by just specifying the username and leaving the password column blank. Upon further reading, I found out that this was the Dummy Authenticator without a password.
So I would like to retain all the users and shift to using the Native Authenticator. How do I do that? An example config.yaml would help as there doesn’t seem to be any documentation for zero-to-jupyterhub-k8s specifically.
If your usernames are identical across both authenticators your PVCs should still match up. As with all critical production systems it’s best to back things up if you’re not sure.
Since there’s no in-built configuration for the NativeAuthenticator you’re probably best off using the Auth0 example as a guide: