Docker “Sharing Dependencies”

Along the readings with Docker, I stopped couple of times with the fact that Docker containers not only share the host kernel, but If possible they share common binaries and libraries.

What I understand from that, is that If I’m running the same docker Image twice on the same host, and this image is using some files x,y,z (say libraries / bins .. anything). These files will also be shared among the 2 lunched containers ? What is even more is that if I’m running two different images, they still could share these common dependencies. What I’m asking for is just two things …

  • Docker missing log4net ConsoleAppender logs
  • Loopback handle connection issues (and startup crashes)
  • Docker ADD is failing with relative directory
  • Vagrant or Docker for environment development
  • node js os.networkInterfaces() not returning any results when called through docker entrypoint on AWS
  • Installing and using Gradle in a docker image/container
  • 1- Verification / Explanation –> Is that true / false + explanation (how does that happen)
    2- If true –> Is there a practical example, that I can run 2 containers (of the same / diff images) and verify they are seeing the same files / libs.

    I hope my question is clear and someone has an answer 🙂

  • NGINX Reverse Proxy failing with Linked Docker Containers
  • docker exec does not open bash
  • Change Docker machine location - Windows
  • docker tomcat Log issue
  • Can you change the CMD of the node layer - Docker
  • Curl in Docker entrypoint script
  • One Solution collect form web for “Docker “Sharing Dependencies””

    Yes, answer is “true” to both questions.
    If you start 2 (or more) containers on the same host, all using the same base image, the whole content of the base image will be shared.

    What is called as an “image” is, in fact, multiple images called “layers” with parent-child relationships, stacked together.

    Now, If you start multiple containers with different images, it may happen that these images share some common layers, depending on how they were built.

    At the system level, Docker mounts each image layer on top of the other up to the final/top image. each layer overwrites its parent content if it overlaps. To do that, it uses what is called an “union filesystem” (Aufs), or even volume snapshots. More information here.

    The images are never modified, they are read-only. On top of the last/upper image, an extra, writeable layer, is added, it will contain changes/additions made by the running container.

    That means that this writeable layer can also be turned into an image layer, and you can start other containers from this new image.

    To see layers sharing “with your own eyes”, just run the following examples:

    docker run ubuntu:trusty /bin/bash
    

    Then:

    docker run ubuntu-upstart:trusty /bin/bash
    

    Docker will tell you that it already has some layers and will thus download them all.

    Check the documentation about writing a Dockerfile (image build script), that should give you a good vision about how all this works.


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