EC2 Container Service Networking

My team is exploring different toolsets (Docker Datacenter, DC/OS, etc.) for a CaaS Platform offering. We’re currently looking at Amazon EC2 Container Service (ECS) as a possible solution, but are having trouble translating our docker-compose files into ECS tasks.

Specifically, we’re not sure how to approach networking within a task in ECS. I was under the impression that a task was sort of analogous to a compose, but it seems to handle networking in a very different way.

  • postgres pg_basebackup can't find .pg_pass file
  • Is it possible to pin the container to a remote machine instead of a new container instance?
  • Google Kubernetes storage in EC2
  • AWS ECR image list
  • Does AWS support Kubernetes?
  • Selenium Grid with Docker containers on different hosts
  • E.g. the https://github.com/docker/example-voting-app expects two networks, ‘front-tier’ and ‘back-tier’ each of which seems to be stock overlay networks. How does one create these in ECS?

  • how I can execute scripts on docker-machine create
  • Is it desirable to use Ansible for creating a docker image
  • Log management on Docker
  • Artifactory exception : There should be 10 default repository layouts
  • Mysql installed and persisting data in docker images
  • Unable to install node 7.5.0
  • One Solution collect form web for “EC2 Container Service Networking”

    ECS supports docker compose style container deployment using
    cmd-ecs-cli-compose.

    Please refer to the following documenation
    http://docs.aws.amazon.com/AmazonECS/latest/developerguide/cmd-ecs-cli-compose.html

    The example you have shared uses docker networking, which was added to docker in 1.9.
    Amazon ECS still does not support docker networking options when defining tasks, so you will not get the example to work as is.

    Of course you can redefine the task and use legacy docker link features to open access between containers.

    Docker will be the best open platform for developers and sysadmins to build, ship, and run distributed applications.