kured, a Kubernetes daemonset, watches for OS‑level reboot signals such as /var/run/reboot-required or a sentinel command and coordinates safe node restarts. It locks the API server so that only one node reboots at a time, preventing simultaneous disruptions.

The tool cordinates node draining and cordoning before reboot, then uncordons the node afterward. Optional logic defers reboots when active Prometheus alerts or specific pods are running, ensuring critical workloads stay online.

Developers can install kured via Helm or kubectl, and configuration is exposed through a ConfigMap. The project lives in the CNCF Sandbox, with documentation hosted at https://kured.dev and community support through Slack and a mailing list.

kured streamlines operational overhead for Kubernetes operators, automating what was once a manual, error‑prone process. By integrating tightly with OS package managers and cluster APIs, it delivers predictable, low‑impact reboots across large fleets.