How to run docker containers on different machines

Question

I've setup several docker containers for my application:

1. container for mysql
2. container for application server (tomcat7)
3. container for nginx
4. container for python tornado application

On my local machine, using docker-machine, I've tried all these containers and them interactive with each other. I created one virtualbox machine with boot2linux and setup all these containers on it.

Now, for production I am going to run each one of these containers on a separate server. Is there something docker provides to quickly/easily do this? Or should I set up docker on each server, copy relevant images to that server, and then start docker-images manually on all servers?

Answer 1

You can use labels and selectors for this purpose.

First add labels to the nodes:

kubectl local nodes node-a podwants=feature1-nodea
kubectl local nodes node-b podwants=feature1-nodeb

A nodeSelector can then be set in the Pod definitions spec.

apiVersion: v1
kind: Pod
metadata:
  name: nginx
  labels:
    app: my-app
spec:
  nodeSelector:
    podwants: feature1-nodea
  container:
    - name: nginx
      image: nginx:1.8
      ports:
      - containerPort: 80

And a define a service in the following manner:

kind: Service
apiVersion: v1
metadata:
  name: web-svc
spec:
  selector:
    app: my-app
  ports:
  - protocol: TCP
    port: 80
    targetPort: 80

Add CoreDNS to your cluster to discover the available Services in other Pods.

 WEB_SVC_SERVICE_HOST=x.x.x.x
 WEB_SVC_SERVICE_PORT=80

Answer 2

set up docker on each server, copy relevant images to that server, and then start docker-images manually on all servers

also,

• create the relevant libnetworks in order for your containers to be registered in them
• create as first container a KV one (Key-Value: Consul, etc, Zookeeper, ...) that will monitor each containers in each network(/machine), allowing them to see each others.
  Note: you might want to use a docker 1.10 for adding a network-scoped alias to your containers.

Then you can create your containers on your machines.