the ingress, canal, and kube-proxy pods are running on the worker, it's just the apply-system-agent-upgrader that keeps restarting over and over