Node/instance connection not working anymore after node memory overload

[aptroost]

Problem
When a node is memory overloaded, the state of that node will be defined as unknown. Meanwhile the instance is running fine when looking in the Hetzner dashboard. The node is not working in Kubernetes and all running pods are stuck in either terminated or restarting state.
A terraform plan would return no changes.

Tested solutions that do not work
I tried a drain and delete for the node and removed the Hetzner instance manually.
After that I retried terraform plan and apply, that would successfully recreate the Hetzner instance, but the instance was not joined to the kubernetes cluster (node not added to cluster).

To be researched

• Potentially when a Kubernetes node is running on the instance and crashes, this provider will not restart the node on that instance. This can be tested by memory overloading the node while checking the instance.
• When removing the node and instance manually, a kubectl join command seems not to be executed on the instance. This can be tested by manually executing this process while monitoring the processes on the new instance.