Docker Service Stack deployment

To work with Docker service stack deployment we should have an orchestrator either Swarm or Kubernetes. Here, in this post, I will be experimenting with the Docker swarm cluster.

Prerequisites

Docker installed machine

Docker swarm initialized and active

Create the manifestation file that describes the task, which will be available in the service definition. We can use the declarative definitions for each service in a YAML file. 

Docker Swarm Service Stack Deployment

I read the book: Learn Docker - Fundamentals of Docker 19.x, where I found the nice explanation given in the 

 

Let's run the docker swarm visualizer container, which will show the Swarm cluster node status, container service status it will be very clear to understand how the orchestrator capabilities work for Docker containers.

docker run -it –d \
  -p 8080:8080 \
  -v /var/run/docker.sock:/var/run/docker.sock  \   
  dockersamples/visualizer

Alternatively, you can use the docker-compose file as a replacement for the container run command:

version '3'

services:
  viz:
    image:  dockersamples/visualizer
    volumes:
      - "/var/run/docker.sock:/var/run/docker.sock"
    ports:
      - "8080:8080"

You can enhance it's usability with your own parameters to run the containers.

When no deployments swarm nodes contain empty:

Initial Swarm Visualization 

Visualizer shows us the number of master and worker swarm nodes participating in the cluster. The green glow-filled circle indicates the active Node of the Docker Swarm cluster. 

Docker stack service deployment

The docker swarm service logs can be viewed in two ways using the parameter as service ID or using service NAME
 

version: "3.7"
services:
  whoami:
    image: training/whoami:latest
    networks:
      - test-net
    ports:
      - 81:8000
    deploy:
      replicas: 6
      update_config:
        parallelism: 2
        delay: 10s
      labels:
        app: sample-app
        environment: prod-south

networks:
  test-net:
    driver: overlay

The menifestation tells about the desired state  of the application, where we mentioned what is the port we want to use on the host and corresponding container port. Here 81 is host port, 8000 is container port. This service uses test-net network resource. The deploy section mentioned as replicas: 6, that means 6 containers will be created on Docker Swarm clsuter. You wish to run the multiple containers run parallel on the same node. We can use labels to identify and remove the services in runtime that will help in monitoring the applicaiton, It is also useful to terminate the services by descovering with these labels.

Run the following docker Stack deployment command as:

docker stack deploy whoami --compose-file docker-service-whoami.yml

docker stack ls

Stack deployment automatically creates overlay network

docker network ls
docker service ls
docker container ls
curl http://192.168.33.250:81

The execution will produce output as follows:

Curl command output from the Docker swarm cluster

Now let's see the deployment of whoami sample:

docker service logs whoami_whoami
docker service logs kiw9ckp68hu5

After the deployment of 'whoami' service see that is distributed amont multiple machines which built-in capability of the Docker swarm.

Docker Swarm Visualization shows replicas=6 on multi-host deployment

Docker Cluster Resilience Test case

Here we can do the test for the resilience two levels:

• container level
• node level

Use case 1: Container level fail-over 

Step 1: Determine the running containers list

Step 2: Remove a container from the stack 

Step 3: The service deployment will automatically create the new container to maintain the DESIRED state as 6 when a container removed ACTUAL State gone to 5, to maintain it Docker Swarm will work to birng up the service.

docker container ls
  docker container rm -f whoami_whoami.3.selectyourown-container
docker ps

Use case 1: Remove a container Swarm service recreate the replacement container

This concludes containers high availability using service deployment using stack on swarm cluster. The actual real usecase that node level will give HA in tne next.

Use case 2: When the node fails - Service Migration

Brought down the Machine and see the outcome on the Swarm Visualizer:

One your Vagrant machines you can stop one of the node to see the impact how it will recover or maintaine the DESIRED State of replicas=6, Here in our example, the service which is running on node1 continue running on available nodes (mstr, node2) by migrating the service to them. This migration will be taken care by the Docker Swarm automatically. We can also define choices in the YAML file. 

Swarm Visualization of migration of services to other worker: Vagrant node1 halt

The Visualization tool shows clear picture of how the Docker Swarm capable of migrate the service when there is issue/failure on node.

 
It will clearly show how High-Availability can be achieved with the Docker Swarm orchestrator.
Jai Hind!!