How do I get k8s going on AWS?

tatty

Hi,

I have some AWS credits but I am not sure how to set up k8s on AWS.

Ideally I would like to practice on a production-like environment but I thought I would start with a minicube on an EC2 instance. I tried this and this is what I got:-

minikube start
😄  minikube v1.17.1 on Ubuntu 20.04 (xen/amd64)
✨  Automatically selected the docker driver. Other choices: none, ssh
 
⛔  Exiting due to RSRC_INSUFFICIENT_CORES:  has less than 2 CPUs available, but Kubernetes requires at least 2 to be available

chrispokorni

Hi @tatty,

Minikube was not designed for production-like clusters. By design, it bootstraps a single node Kubernetes cluster. Multi-node support is experimental, and not yet recommended.
Minikube was not designed to run on VMs or on cloud virtual instances, such as EC2. Running minikube in a virtual environment may cause many issues.
The course is not based on minikube. Instead, it uses the default kubeadm tool to bootstrap a multi-node, production-like, Kubernetes cluster, which can run on EC2 instances, and other cloud and local VMs.

The free tier or credits on AWS come with limitations. From the error above, your EC2 instance does not have sufficient resources to meet Kubernetes' minimum requirements.

I would recommend revisiting the "Exercise 3.1: Install Kubernetes / Overview" section for VM sizing guide. The labs work on GCP GCE VMs, AWS EC2 instances, DigitalOcean droplets, VirtualBox VMs, provided that instances are sized accordingly and the networking is fully open between the nodes of the Kubernetes cluster.

Regards,
-Chris

tatty

Thank you Chris. I watched the video and now I see that it says to provision an EC2 instance of type m5a.large, with 2 virtual CPUs. The video I am talking about is at https://training.linuxfoundation.org/cm/LFS258/LabSetup-AWS.mp4 ... using the login given in the class.

I also followed your (and the video's) advice and have now used 2 EC2 VMs with 2 CPUs and 20BG RAM each within the same Auto Scaling Group and I have used kubeadm to create a Kubernetes cluster. I have to say, the instructions in the course are a little thin on how to do that... it's not a gentle "introduction". Perhaps this could be an area flagged for improvement for future iterations of this course.

fracalo

Hi all,
I am also provisioning the VMs for the cluster on AWS,
the easiest way I found to accomplish this is to use cloudformation,
Here's the link to the template I'm using https://gist.github.com/fracalo/50f825460ad4f7a2070b8cacdcbba162.

The command to provision the stack is
aws cloudformation create-stack --region eu-north-1 --stack-name k8s-stack --template-body file://instanceCKA.yml --parameters ParameterKey=KeyName,ParameterValue=your_saved_ssh_key_on_aws ParameterKey=Instance,ParameterValue=t3.medium

if you change the region you'll probably need to change the os-ami accordingly, from what i remember it changes by region.

After the stack is up you'll still need to update the docker engine daemon.json, add the dns for the CP, run kubeadm, and configure calico;

Hope it helps...