Step-by-Step Guide to Upgrade a Kubernetes Cluster from v1.17.9 to v1.18.9

Upgrade Overview

Use a kubeadm‑provisioned Kubernetes cluster.

Minor upgrades are supported; major upgrades are allowed but skipping two major versions is not recommended.

Back up all cluster resources before proceeding.

Upgrade Target

Upgrade the cluster from v1.17.9 to v1.18.9 .

# kubectl get nodes
NAME            STATUS   ROLES   AGE   VERSION
ecs-968f-0005   Ready    node    102d  v1.17.9
k8s-master      Ready    master  102d  v1.17.9

Backup the Cluster

kubeadm upgrade

does not affect workloads, but backing up the cluster is recommended. The guide uses the open‑source script k8s-backup-restore :

(1) Download the script

$ mkdir -p /data
cd /data
git clone https://github.com/solomonxu/k8s-backup-restore.git

(2) Run the backup

cd /data/k8s-backup-restore
./bin/k8s_backup.sh

To restore:

# mkdir -p /data/k8s-backup-restore/data/restore
$ cp devops_deployments_gitlab.yaml ../../restore/
$ cd /data/k8s-backup-restore
./bin/k8s_restore.sh

Upgrade the Cluster

Master Upgrade

(1) Choose the target version

$ yum list --showduplicates kubeadm --disableexcludes=kubernetes

(2) Upgrade kubeadm to 1.18.9

$ yum install -y kubeadm-1.18.9-0 --disableexcludes=kubernetes

(3) Verify the version

$ kubeadm version
kubeadm version: &version.Info{Major:"1", Minor:"18", GitVersion:"v1.18.9", ...}

(4) Drain the master node

$ kubectl cordon k8s-master
$ kubectl drain k8s-master

(5) Run the upgrade plan

$ kubeadm upgrade plan

(6) Apply the upgrade

$ kubeadm upgrade apply v1.18.9 --config kubeadm.yaml

(7) Uncordon the master

# kubectl uncordon k8s-master

Node Upgrade

(1) Upgrade kubeadm on the node

$ yum install -y kubeadm-1.18.9-0 --disableexcludes=kubernetes

(2) Cordon and drain the node

$ kubectl cordon ecs-968f-0005
$ kubectl drain ecs-968f-0005

(3) Upgrade the node control plane

$ kubeadm upgrade node

(4) Upgrade kubelet and

kubectl

$ yum install -y kubelet-1.18.9-0 kubectl-1.18.9-0 --disableexcludes=kubernetes
$ systemctl daemon-reload
$ systemctl restart kubelet

(5) Uncordon the node

$ kubectl uncordon ecs-968f-0005

Verify the Cluster

Check node status

$ kubectl get no
NAME            STATUS   ROLES   AGE   VERSION
ecs-968f-0005   Ready    node    102d  v1.18.9
k8s-master      Ready    master  102d  v1.18.9

Check certificate expiration

$ kubeadm alpha certs check-expiration
... (output omitted for brevity) ...

Note: kubeadm upgrade automatically renews certificates it manages on the node. To skip renewal, add the flag --certificate-renewal=false .

Failure Recovery

If an upgrade fails without rollback, you can re‑run kubeadm upgrade or force it with kubeadm upgrade --force. Backup files are stored under /etc/kubernetes/tmp: kubeadm-backup-etcd-* – etcd data backup. kubeadm-backup-manifests-* – static pod YAML manifests.

Restore etcd data by copying the backup into the etcd data directory, and restore static pod manifests by copying them back to /etc/kubernetes/manifests.