This workshop has been deprecated and archived. The new Amazon EKS Workshop is now available at www.eksworkshop.com.
Before we start making changes to VPC CNI, let’s make sure we are using latest CNI version
Run this command to find CNI version
kubectl describe daemonset aws-node --namespace kube-system | grep Image | cut -d "/" -f 2
Here is a sample response
Edit aws-node DaemonSet and add AWS_VPC_K8S_CNI_CUSTOM_NETWORK_CFG environment variable to the node container spec and set it to true
Note: You only need to set one environment variable in the CNI daemonset configuration:
kubectl set env ds aws-node -n kube-system AWS_VPC_K8S_CNI_CUSTOM_NETWORK_CFG=true
kubectl describe daemonset aws-node -n kube-system | grep -A5 Environment
Terminate worker nodes so that Autoscaling launches newer nodes that come bootstrapped with custom network config
Use caution before you run the next command because it terminates all worker nodes including running pods in your workshop
INSTANCE_IDS=(`aws ec2 describe-instances --query 'Reservations[*].Instances[*].InstanceId' --filters "Name=tag-key,Values=eks:cluster-name" "Name=tag-value,Values=eksworkshop*" --output text` )
for i in "${INSTANCE_IDS[@]}"
do
echo "Terminating EC2 instance $i ..."
aws ec2 terminate-instances --instance-ids $i
done
Enabling a custom network effectively removes an available network interface (and all of its available IP addresses for pods) from each node that uses it. The primary network interface for the node is not used for pod placement when a custom network is enabled. Determine the maximum number of pods that can be scheduled on each node using the following formula.
In your production setup, replace your existing nodegroup with a new nodegroup and apply the value of maxPodsPerNode option. For simplicity, this workshop continues with the existing nodegroup without custom maxPodsPerNode value.