Upgrade hop

Kubernetes 1.16 → 1.17

1.17 is itself end of life. If you are on 1.16 today, plan the whole path to a supported version (at least 1.35) with the upgrade planner.

1 · Fix removed APIs first

No API versions are removed between 1.16 and 1.17.

2 · Upgrade in this order

1. Prepare. Move every node to 1.15 or newer if any are older: a 1.17 API server supports kubelets back to 1.15 only. Upgrade to the newest 1.16 patch, 1.16.15.
2. kube-apiserver to 1.17 (in HA clusters, one instance at a time; old and new may differ by one minor during the roll). Use the newest patch, 1.17.17.
3. kube-controller-manager, kube-scheduler, cloud-controller-manager to 1.17, once every API server is on 1.17.
4. kubelet on each node (drain first). Optional in this hop: kubelets from 1.15 to 1.17 all work with a 1.17 control plane.
5. kube-proxy: no newer than the API server, and within the same window as the kubelet it runs beside.
6. kubectl: 1.16 to 1.18 all talk to a 1.17 API server.

From the official version skew policy. Managed services run the control-plane steps for you; node pools are still yours to schedule.