A Virtual Kubelet provider for running systemd units that just works. This project is heavily inspired by systemk.
There is just a limited set of supported features, like just wrapping basic functionalities of dbus and journal,
exposing them via K8s API. That's because:
- I have a farm of no greater than 50 machines to manage, per cluster
- I have the full control of these machines, with 250+ GiB memory and 50+ TiB disks, for distributed storage
- I don't think Ansible-based automation is decently enough for administering
- I want to try some fancy yet battle-tested tooling
Todos:
- Fetching container logs
- Exposing metrics
- Ah yes, haven't tested all of these yet xD
MIT