To be able to reboot, make changes/upgrades on the OS without have downtime you can use the commands bellow:
-
kubectl drain nodename- Recreates all the PODs in other Nodes and mark the Node as unscheduled, but keep in mind that if you don't have more replicas your service have some downtime.
-
kubectl cordon nodename- Mark the Node as unscheduled
-
kubectl uncordon nodename- It will make the Node schedulable again.
-
To ignore daemonsets, just add:
--ignore-daemonsets