Experiment with Kubernetes with a custom Docker Image

README.md

1. Build a custom docker image using Dockerfile:

   docker build -t python/server .

2. Test the image:

   docker run -i -t -p 8000:8000 python/server
   curl http://localhost:8000

3. Deploy on Kubernetes

kubectl create -f myapp.yaml

4. Look around:

[dims@ubuntu-brix 17:30] ~/Downloads ⟩ kubectl get pods
NAME          READY     STATUS    RESTARTS   AGE
myapp-jxd9f   1/1       Running   0          47s
[dims@ubuntu-brix 17:30] ~/Downloads ⟩ kubectl get services
NAME         CLUSTER-IP   EXTERNAL-IP   PORT(S)    AGE
kubernetes   10.0.0.1     <none>        443/TCP    3h
myappsvc     10.0.0.74    nodes         8000/TCP   52s
[dims@ubuntu-brix 17:30] ~/Downloads ⟩ kubectl get replicationcontrollers
NAME      DESIRED   CURRENT   AGE
myapp     1         1         56s

5. Test the app:

curl -kvs 10.0.0.74:8000

6. Alternate to step #3:

kubectl run myapp-node --image=python/server --port=8000
kubectl expose deployment myapp-node --port=8000 --name=myapp-node-http
kubectl describe pods

Dockerfile

FROM  python:3  
EXPOSE  8000  
CMD ["python", "-m", "http.server"]

myapp.yaml

apiVersion: v1
kind: Service
metadata:
  name: myappsvc
  labels:
    app: myapp
spec:
  type: NodePort
  ports:
  - port: 8000
    protocol: TCP
    name: http
  selector:
    app: myapp
---
apiVersion: v1
kind: ReplicationController
metadata:
  name: myapp
spec:
  replicas: 1
  template:
    metadata:
      labels:
        app: myapp
    spec:
      containers:
      - name: myapp
        image: python/server
        ports:
        - containerPort: 8000

Thanks for the gist. This helped a lot.

I encountered errimagepull since I wanted to load from worker node but the image was in my master node.
I got it working by running a local image register.

docker run -d -p 5000:5000 --restart=always --name registry registry:2
docker tag python/server localhost:5000/server
docker push localhost:5000/server

and I loaded it with image: localhost:5000/server
based on this post

Where do you actually store your custom images for use with Kubernetes?