- Goals
- Pre-reqs
- Overview
- Steps
- 01 - Run nginx in docker serving local files
- 02 - Run a git sync tool in docker cloning this repo
- 03 - Run both containers above in a k8s pod
- 04 - Run a fixed number of replicas for each of our pods, called deployment
- 05 - Create a service for each deployment that allows us to do service discovery across any number of pods running
- 06 - Create an ingress object that makes those pods available to the external world
This workshop will help you understand how the basic kubernetes primitives work.
You will understand how:
- pods work
- multiple containers
- volume sharing
- network sharing
- resource constraints
- deployments work with a pod that has:
- 2 containers
- exposed ports
- volume sharing
- resource contraints
- services work with different label selectors
- ingresses work with path based routing, SSL certificates and custom nginx configuration
This assumes that you can run docker locally and that you have a kubernetes cluster setup with:
- external-dns
- cert-manager
- nginx-ingress-controller
We will go through a journey running simple application which serves gifs of cats and dogs. The application has two components; a web server serving the images and a file puller pulling gifs from somewhere, in this case a git repository.
These are the steps we will go through during the workshop
In this step we will run our nginx server locally on docker serving gifs we have locally.
Run:
01-docker-nginx/run.sh
Thats nice but how do I get my local files in the instance docker runs ? Enter step 2
In this step we will run a separate container that pulls files from a remote git repository.
Run:
02-docker-git-sync/run.sh
Nice! so now we have a way to get our files over to the instance docker runs. But I need to combine both containers somehow..
In this step we will show how you can combine multiple containers of single concern into a pod and leverage that capability.
Run:
kubectl apply -f 03-pod/
So now we can put both containers in a pod and leverage the pod semantics across shared network and volume. What happens if the pod dies though ? How can i maintain a number of pods always ?
In this step we will create two deployment objects. One for cats and one for dogs. They will each have a fixed number of replicas such that we can ensure a fixed capacity irrespective of individual pods are dying.
Run:
kubectl apply -f 04-deployment/
How do I access all the pods you might ask ? Since i have more than 1, i need a way to do service discovery and load balance across all the available ones.
05 - Create a service for each deployment that allows us to do service discovery across any number of pods running
In this step we will create a Service object for each deployment. This will allow us to group all the pods of a deployment into a service discovery endpoint based on the labels we provided to the pods in the deployment.
Run:
kubectl apply -f 05-service/
How do i expose those to the outside world now ?
In this step we will create an Ingress object that exposes the pods under https://k8sprimitives1.k8sdemos.test-infra.ouzi.io
Run:
kubectl apply -f 06-ingress/