Circling back to this: we’ve implemented something closer to Yuvi’s original plan, where we let Kubernetes handle basically everything. The upshot is that things are easier to deploy since there’s no database and there’s no separate state to manage.
GitHub - kbatch-dev/kbatch is the new home. The README has basic usage guidelines. https://github.com/kbatch-dev/kbatch/tree/main/kbatch-proxy has some deployment documentation if anyone is interested in trying to deploy it to your hub (I’ll publish a helm chart soon-ish, so things will be easier).
Yes, something more like that.