(copied from discussion on Marketplace dev GitLab organization)
Following on from a discussion with @Lukas_Koschmieder, it may prove beneficial to implement a load balancing system into S-R (i.e. Kubenetes) in order to support large groups of users (~200 at a time) for tutorials / online training.
AiidaLab are already using K8s and the following guide will be useful to implement something similar: https://zero-to-jupyterhub.readthedocs.io/en/latest/
It seems like at the very least this will require developing a new Spawner class, as well as ensuring that S-R can be ran on a stable JupyterHub Helm chart version: https://jupyterhub.github.io/helm-chart/
In which case we may expect it to be dependent on #552
(copied from discussion on Marketplace dev GitLab organization)
Following on from a discussion with @Lukas_Koschmieder, it may prove beneficial to implement a load balancing system into S-R (i.e. Kubenetes) in order to support large groups of users (~200 at a time) for tutorials / online training.
AiidaLab are already using K8s and the following guide will be useful to implement something similar: https://zero-to-jupyterhub.readthedocs.io/en/latest/
It seems like at the very least this will require developing a new Spawner class, as well as ensuring that S-R can be ran on a stable JupyterHub Helm chart version: https://jupyterhub.github.io/helm-chart/
In which case we may expect it to be dependent on #552