1.10 -> 1.11 upgrades

[timothysc]

Create guide on 1.11 upgrade process.

[timothysc]

Install kubeadm on 1.10
Fetch from head and upgrade to latest 1.11 using upgrade apply
Fetch latest debs/rpms and install them and verify it’s working
Check kubelet config / unit file changes

[liztio]

1. Follow the normal instructions to set up a 1.10 cluster.
2. Checkout HEAD on the branch you want to test.
3. Build a kubeadm binary: bazel build //cmd/kubeadm:kubeadm --platforms=@io_bazel_rules_go//go/toolchain:linux_amd64
4. scp the binary up to your master node
5. Upgrade the kubernetes master. sudo /tmp/kubeadm upgrade apply ci/<k8s version> --allow-experimental-upgrades
   Make sure that <k8s version> is a HEAD revision or the build artifacts won't exist.
   You can double-check that it does with e.g. gsutil ls gs://kubernetes-release-dev/ci/v1.11.0-beta.2.78+e0b33dbc2bde88.
6. scp the kubeconfig locally. Set the host address to kubernetes in /etc/hosts
7. In your local hosts file, point kubernetes to the master's IP.
8. Alternatively, if you haven't opened port 6443 to the internet, you can forward port 6443 with ssh <master> -L 6443:localhost:6443, and set kubernetes to 127.0.0.1 in /etc/hosts
9. Run the kubeadm e2e tests: go run test/e2e_kubeadm/runner/local/run_local.go -build -test-flags="--kubeconfig=<path to configi>" from the kubernetes/kubernetes directory
10. Get a kubelet binary from https://storage.googleapis.com/kubernetes-release-dev/bazel/<k8s version>/bin/linux/amd64/kubelet.deb
11. Install the binary with sudo dpkg -i on the master and all nodes
12. Restart the kubelet with systemctl restart kubelet
13. Check the node version with kubectl get nodes. Make sure it says Ready.
14. Repeat on all worker nodes.
15. Re-run the kubeadm e2e tests.

[luxas]

The PR for this has been merged 👍