I just upgraded the RAM from 16GB to 32GB for all VM nodes of existing cluster.
But kubectl get nodes still shows 16GB per node memory, how to make kubernetes re-read the sepcs?
Restart Kubelet does the trick but its not practial in case of larger clusters.
$ kubectl get nodes node-02 -o yaml | grep mem
memory: 16145384Ki
memory: 16247784Ki
message: kubelet has sufficient memory available
After restart:
$ kubectl get nodes node-02 -o yaml | grep mem
memory: 32922600Ki
memory: 33025000Ki
message: kubelet has sufficient memory available