Improvements and Progress
In terms of cluster reliability, we have added a second router to be highly available with the already existing one; If the master router fails for any reason, then the backup one will take its place until the master comes back online. Furthermore, we configured the routers to act as TFTP servers so that they we may use Ubuntu netboot to load Ubuntu on all nodes from scratch. As compared to the previous method, this way is more effective and utilizes existing infastructure better.
In the Works
For the future, we are planning to develop a Ceph cluster to replace our NFS-client-provisioner as the storage solution for users on the JupyterHub. We also intend to extend the cluster with GPU nodes and provide more specialized computing enviornments. Feel free to use our Binder plugin on libretexts.org, and check out our JupyterHub if you are a UC Davis member!