Old-kubelet & kubelet on the node

Hi All,

I noticed all my nodes have old-kubelet and kubelet containers running (not very successfully), both using the same rancher/hyperkube:v1.13.5-rancher1 image. The problem is that they are running using the same ports and kubelet is always restarting (because ports are taken by old-kbelet).

I’m not sure why it ended up like this and also not sure if this is normal, but currently I have to manually login into each node and kill the ‘old-kubelet’ container after each node restart(!). What would be the proper way to remove the old-kubelet automatically (or at least persistent).