README.md

How to run a data science container

In order to start a container you just need ./run.sh

Jupyter notebook config file is located in ./config directory. You might want to change config, at least the password hash to access the notebook.

Prerequisites

You need docker, GPU drivers and nvidia-docker installed.

Settings

Before running a container you might set some environment variables:

DS_NAME

default: ds-py3

DS_NOTEBOOKS_DIR

default: ./notebooks

Directory in the host system which is mapped to a container /notebooks directory where all the notebooks are stored

DS_DATA_DIR

default: ./data

Directory in the host system which is mapped to a container /data directory where all the data are stored

DS_CONFIG_DIR

default: ./config

Directory in the host system where jupyter_notebook_config.py is stored. It is mapped to /jupyter/config directory in the container.

DS_SECRET_DIR

default: ./secret

Directory in the host system where TLS certs are stored. It is mapped to /jupyter/secret directory in the container.

DS_PORT

default: 8888

Host port where jupyter notebook is listening.

DS_IMAGE

default: analysiscenter1/ds-py3

Docker image to run in a container.

DS_EXTRA_PORTS

default: 2

Number of extra ports to open in a container, starting from the DS_PORT + 1. If equals to 0, no extra ports are mapped.
For example, if DS_PORT=8892 and DS_EXTRA_PORTS=2, then the container is started with additional port mapping 8893:8893, 8894:8894.

Examples

DS_PORT=8889 ./run.sh - to run a container which can be accessed at http://localhost:8889 DS_NOTEBOOKS_DIR=/notebooks ./run.sh - to store notebooks in the host directory /notebooks

You can pass additional docker options, for instance:
DS_PORT=8889 ./run.sh -it - to run a container interactively.

To see docker options enabled by default, see the runfile.