Since drain/eviction process should not depend on the sysadmin's laptop. The idea is to write a controller that would manage the maintenance mode from server side.
We also want the ability to fire node maintenance from UI - which couldn't be achieved by invoking kubectl drain command (not possible).
The purpose of the controller (wrapped as an operator in the current project):
The purpose of the node-maintenance-operator is to watch for new or deleted custom resources called NodeMaintenance which indicate that a node in the cluster should either:
--
You received this message because you are subscribed to the Google Groups "kubernetes-sig-node" group.
To unsubscribe from this group and stop receiving emails from it, send an email to kubernetes-sig-...@googlegroups.com.
To post to this group, send email to kubernete...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/kubernetes-sig-node/0da1cb27-ab10-47dc-aef9-8038254ad69a%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.