This message was deleted.

Hmm I migrated to manipulating the

Longhorn node-drain-policy Setting Custom Resource

instead. This works better. Both in regards to manipulating the actual setting and it applies quicker. Further, I tested whether or it’s necessary to actually disable

DISK SCHEDULING

today. It’s not. • Cordoning workers to be removed and then • changing the

Longhorn node-drain-policy

is enough I see:

AutoEviction

on the worker nodes to be removed that contains volume replicas. This is nice!