How to use the Docker image

Building

    $   cd  bigdl-tutorials/docker

    # to use default 'Dockerfile'
    # do not forget this DOT (.) at the end
    $   docker build  .

    #   to do a different build file
    # do not forget this DOT (.) at the end
    $   docker build  -f  Dockerfile-v3   .

    # to add a tag
    $   docker build  -f  Dockerfile-v3  -t bigdl/bigdl   .

To force a build provide --no-cache option

    $   docker build --no-cache   .

To see built images

    $   docker images
    $   docker image inspect  <IMAGE_ID>

Running it

$   docker images

Will list all images.

Option 1:

The simplest way to run the docker container is to use run-bigdl-docker.sh script at the project root directory. It will run the image and also mounts a working directory so all the work is saved automatically.

    $    ./run-bigdl-docker.sh   <image_id>

Option 2:

Running the Jupyter notebook by default (~/run_jupyter.sh)

    $   docker run -it  -p 8888:8888   IMAGE_ID

Option 3:

Shell access

    $   docker run -it  -p 8888:8888   IMAGE_ID   bash

In Docker container you can run Jupyter as follows

    $    ./run_jupyter.sh

Go to http://localhost:8888 in browser

Pushing to Dockerhub

    ## login
    $  docker login
    ## enter username, password

    ## tag the image
    $   docker tag IMAGE_ID  elephantscale/bigdl-sandbox:latest
    $   docker images

    ## Pushing
    $   docker  push  elephantscale/bigdl-sandbox:latest