Almost 3000 lines of code for automating draining nodes and rebooting it. And it requires that another component has already queued up an update that requires a reboot.

Looking at the issues, people try to shoehorn a thousand unique behaviours into a general purpose tool, just to avoid a bit of old school sysadmin-ing. There's a guy wanting to change TZ of the running cluster, and want "Kured" to support that use case so it's only updated during night - in an ever changing TZ.