1. UPGRADE KUBERNETES CLUSTER

Below are the steps we need to perform for upgrading the Kubernetes cluster.

Note: This document is only from the exam point of view and for upgrading production-based clusters, we need to consider more points.

Get detailed information about the upgrade process from the below-mentioned website.

Kubernetes

UPGRADE CONTROLPLANE NODE

# Update the repositiry

sudo apt update



# Check for the lastest kubernetes release versions

sudo apt-cache madison kubeadm

UPGRADE KUBEADM

# Unholds to upgrade kubeadm package

sudo apt-mark unhold kubeadm


# Install the new version of kubeadm, here the version is 1.20.4-00

sudo apt update && sudo apt install -y kubeadm=1.20.4-00


# Hold the package to prevent accidential upgrade

sudo apt-mark hold kubeadm


# Fetches the control plane component versions to which we can update to.

sudo kubeadm upgrade plan


# Upgrade kubeadm

sudo kubeadm upgrade apply v1.20.4

UPGRADE KUBECTL AND KUBELET

# Drain the control plane node

kubectl drain manager --ignore-daemonsets



# Unholds to upgrade for kubelet and kubectl

sudo apt-mark unhold kubelet kubectl



# Install the new version of kubelet and kubectl, here the version is 1.20.4-00

sudo apt-get update && sudo apt-get install -y kubelet=1.20.4-00 kubectl=1.20.4-00



# Hold the package to prevent accidential upgrade

sudo apt-mark hold kubelet kubectl



# Will reloads the systemd manager configuration

sudo systemctl daemon-reload



# Will restart the kubelet service

sudo systemctl restart kubelet



# Check the status of the service

sudo systemctl status kubelet



# Uncordon the cntrol plane node.

kubectl uncordon manager

UPGRADE WORKER NODES

Note: worker nodes should not be upgraded simultaneously.

UPGRADE KUBEADM

On Worker node

# Update the repositiry

sudo apt update



# Unholds to upgrade kubeadm

sudo apt-mark unhold kubeadm



# Install the new version of kubeadm, here the version is 1.20.4-00

sudo apt update && sudo apt install -y kubeadm=1.20.4-00



# Hold the package to prevent upgrade

sudo apt-mark hold kubeadm



# Upgrade the local configuration

sudo kubeadm upgrade node

UPGRADE KUBECTL AND KUBELET

On the control plane node

# Drain the worker1 node

kubectl drain worker1 --ignore-daemonsets --delete-emptydir-data

On Worker node

# Unholds to upgrade kubelet and kubectl

sudo apt-mark unhold kubelet kubectl



# Install the new version of kubelet and kubectl, here the version is 1.20.4-00

sudo apt-get update && sudo apt-get install -y kubelet=1.20.4-00 kubectl=1.20.4-00



# Hold the package to prevent upgrade

sudo apt-mark hold kubelet kubectl



# Will reload the systemd manager configuration

sudo systemctl daemon-reload



# Will restart the kubelet service

sudo systemctl restart kubelet

On the control plane node

# Uncordon the worker node to bring it online and run the below command on control plane node.

kubectl uncordon worker1

Perform the above steps on all the worker nodes to upgrade kubeadm, kubelet and kubectl.

On the control plane node

# Verify the cluster to list the updated nodes.

kubectl get nodes

1. Backing up and Restoring data

Backup a specific resource:

kubectl get <resource> <resource_name> -o yaml > backup.yaml

For example, to backup a deployment named my-deployment in the default namespace, you would run:

kubectl get deployment my-deployment -n default -o yaml > backup.yaml

Backup all resources in a namespace

kubectl get all -n <namespace> -o yaml > backup.yaml

For example, to backup all resources in the default namespace, you would run:

kubectl get all -n default -o yaml > backup.yaml

Restore a specific resource from a backup:

kubectl apply -f backup.yaml

Restore all resources from a backup:

kubectl apply -f backup.yaml --prune

The --prune flag will remove any resources that are not in the backup.

Backup and restore persistent volumes:

To backup a persistent volume, you can use your storage provider's tools or utilities such as velero. For example, to backup a persistent volume named my-pv, you would run:

velero backup create my-backup --include-namespaces <namespace> --selector key=value

To restore the persistent volume, you would run:

velero restore create --from-backup my-backup

1. Scaling the Cluster

Scale a deployment:

kubectl scale deployment <deployment_name> --replicas=<number_of_replicas>

For example, to scale a deployment named my-deployment to 3 replicas, you would run:

kubectl scale deployment my-deployment --replicas=3

Scale a statefulset:

kubectl scale statefulset <statefulset_name> --replicas=<number_of_replicas>

For example, to scale a statefulset named my-statefulset to 3 replicas, you would run:

kubectl scale statefulset my-statefulset --replicas=3

Scale a deployment using autoscaling:

kubectl autoscale deployment <deployment_name> --min=<minimum_number_of_replicas> --max=<maximum_number_of_replicas> --cpu-percent=<cpu_utilization_percentage>

For example, to create an autoscaling deployment named my-deployment with a minimum of 2 replicas, a maximum of 5 replicas, and a target CPU utilization of 80%, you would run:

kubectl autoscale deployment my-deployment --min=2 --max=5 --cpu-percent=80

Check the current scaling status of a deployment:

kubectl get deployment <deployment_name> -o=jsonpath='{.spec.replicas}'

For example, to check the current scaling status of a deployment named my-deployment, you would run:

kubectl get deployment my-deployment -o=jsonpath='{.spec.replicas}'

Thank you for reading this Blog. Hope you learned something new today! If you found this blog helpful, please like, share, and follow me for more blog posts like this in the future.