Custom DOCKER_HOST in Elastic Beanstalk Docker environment

I have an ElasticBeanstalk setup using Amazon’s Docker AMI. I have just started experimenting with setting up a new environment within my EB Application. It is running as Webserver with Docker.

What I am trying to do with this environment is have Docker daemon running that is running on TCP rather than a Unix socket. That works by modifying /etc/sysconf/docker and adding the -H flag. Unfortunately, that breaks that EB deployment scripts because they are expecting the Docker daemon on the default Unix socket.

  • Container healthcheck status via tasks json
  • Connecting from a docker container to the host?
  • Docker containers not running on VM
  • docker run --ulimit cpu=10 does not kill java process after timeout
  • Is it possible to access Hbase installed inside docker container to be accessed using java client on mac OSX?
  • stop a database container in docker automatically
  • I have tried to add a /etc/profile/docker.sh which contains the appropriate DOCKER_HOST variable. However, it doesn’t seem to be picked up by the EB deployment scripts.

    Did anyone try something similar? Where can I ‘globally’ set the DOCKER_HOST so it gets picked up by the EB deployment?

    I appreciate any hints and suggestions.

  • alpine linux — autossh (missing)
  • Docker Swarm with TLS: Compose behaves strangely using DOCKER_CERT_PATH
  • want to use the same host and port for ssh session
  • Basic Dockerfile not running CMD command
  • Script test to see if a tool is already loaded?
  • How can I create a Docker image based on a git tag in the public Registry?
  • One Solution collect form web for “Custom DOCKER_HOST in Elastic Beanstalk Docker environment”

    You can use multiple -H flags to have your docker daemon listening on both the unix socket and TCP socket.

    Wouldn’t that please the EB deployment script while still allowing you to control the docker daemon with TCP?

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