TLDR
-
Kured is a Kubernetes daemonset that automates safe node reboots triggered by OS package managers, using API server locks to serialize reboots.
Key Takeaways
-
Watches for
/var/run/reboot-required or a custom sentinel command to detect reboot need from the underlying OS.
-
Uses a Kubernetes API server lock to guarantee only one node reboots at a time, preventing cluster-wide disruption.
-
Cordons and drains nodes before rebooting, then uncordons after, following standard Kubernetes safe-eviction patterns.
-
Can defer reboots when active Prometheus alerts are firing or specific pods are present, adding operational safety hooks.
-
CNCF Sandbox project with dedicated docs at kured.dev, a Slack channel, and monthly community meetings.
Hacker News Comment Review
-
No substantive HN discussion yet.
Original | Discuss on HN
