The application can be launched on your local laptop in development mode using a Minikube k8s cluster. We have also included Terraform scripts, here, that can help you spin up a k8s cluster in Amazon cloud EKS to highlight a few integration points between your k8s resident application and the underlying cloud provider.
You can download the slides accompanying the code.
Your Kubernetes Learning Plan
- Browse through the slides in this tutorial to get a condensed start to the concepts around Docker and k8s, and understanding how a Python/Django application can live inside k8s.
- If you want to try out k8s without installing anything or creating a cloud provider account, spend time on the official Kubernetes tutorials; these include interactive tools to let you actually play with a k8s cluster without spinning one up on your own machine/cloud account:
- If you are ready to take the next step - and commit to installing a k8s-all-in-one VM on your PC, head over to the Minikube installation page to install the single-node k8s instance on your PC. This is useful for development work. You can then try to deploy the django application included in this tutorial on your local PC k8s cluster.
- The next step is moving toward Kubernetes for production. We strongly recommend creating a cloud-provider-managed k8s cluster unless you have a strict requirement for an in-house production-grade do-it-yourself k8s cluster.
- We have included Terraform scripts to create a Amazon cloud EKS cluster. You will need AWS credentials and a credit card - warning: running a k8s cluster is not cheap (see our slides for an example).
What is Where?
Here are links to key resources in this repository
- Slides for the tutorial.
- Demo video - Jenkins + Kubernetes demo to help you visualize a devops/infrastructure-as-code approach to Python development and operations-as-code using Kubernetes.
- Minikube installation instructions
- Docker setup and commands and the Dockerfile to create the django application container image from a standard base python image.
- Amazon AWS EKS k8s setup - you will need an AWS account (credit card required).
- Using the k8s cluster This is the meat of the tutorial - things to do with the k8s cluster.
- Amazon AWS RDS setup to create a simple postgres database instance - you will need an AWS account (credit card required). We separated the terraform scripts for EKS and RDS setup because you probably don’t want to create and tear down your production database and EKS cluster at the same time!
- Sample django poll application, the settings.py may be of particular interest. The README describes the docker commands to package the application into a container.
- The Jenkins pipeline files contain shell scripts, executed in stages by Jenkins. This gives you ideas about how to use a CI/CD tool with Kuberetes workflows; see the demo video to see these Jenkins pipelines in action.
Why did you choose AWS EKS and not Google or Azure as the managed k8s cluster provider? And why not Flask instead of Django?
K8s evolved from an internal project in Google, and it is widely believed that the Google cloud managed k8s product is the most polished and easy to use (as of late 2018). Google has also done a phenomenal job of creating documentation for Python developers using their k8s product (see: a Python Flask app on Google cloud k8s).
We decided to go with AWS EKS for this tutorial to complement Google’s K8s + Flask example. In addition, there is a large installed base of AWS users who may find this tutorial helpful.
What are the costs associated with running the AWS EKS cluster?
You pay for the k8s management plane (~$0.20 per hour), the number of worker nodes (EC2 pricing), and the load balancer cost. If you use other services like RDS or S3, those are extra.
Sachin Agarwal is a computer systems researcher and the founder of BigBitBus.
BigBitBus is on a mission to bring greater transparency in public cloud and managed big data and analytics services. Talk to us about how you can architect your cloud IT assets to minimize public-cloud stickiness and lock-in.