Continuous Deployment with Docker (swarm mode)
This project shows a web application built using a microservices architecture.
There are two microservices:
- rest-count implemented in Python (Flask microframework) using a Redis database
- rest-ip implemented in Node.js (Express framework) using a MongoDB database
deployed as docker services on a swarm cluster
You can find additional information on my Slideshare presentation "Always be shipping"
(see legacy tag if you're interested into the previous version for Docker 1.11 using docker-compose and consul.)
run on a single node docker swarm cluster the following commands in terminal (the first time you have to wait for a few minutes to download the Docker base images), :
docker stack deploy -c docker-compose.yml cd
open your browser to http://127.0.0.1:8080/
you can check the services on http://127.0.0.1:8080/visualizer/
rest-count/main.py (for example, you can increase the version to 1.1)
docker build -t francescou/rest-count:1.1 rest-count/ docker service update --image francescou/rest-count:1.1 cd_restcount
the updated microservice will be deployed with no downtime.
You can also modify the rest-ip microservice in the same way (see
this section will explain how to can scale up and down the docker services.
open your browser to http://127.0.0.1:8080/visualizer/. There you will find a rest-count service, running on two replicas. Execute
docker service scale cd_restcount=4
check again http://127.0.0.1:8080/visualizer/ to ensure that there are now four rest-count instance.
Make a few requests to http://127.0.0.1:8080/api/v1/count and then run
docker service logs cd_restcount to see how requests are processed by different rest-count instances.
You can now scale down the rest-count service without having any down time, e.g.:
docker service scale cd_restcount=2
Again, you can check http://127.0.0.1:8080/visualizer/ to see that there are now only two instances.