docker

Docker Compute Environment

In this hands-on exercise, we will learn how to create and use a reproducible computational environment with Docker.

[TOC]

Installing Docker

You must already have Docker installed. See the Docker installation documention for more information.

In general this means having a properly configure kernel and the docker client and daemon installed on a Linux distribution. On Mac or Windows, the installation installation instructions will install a boot2docker utility to manage a VirtualBox virtual machine running the docker client and daemon.

Build the Docker Base Image

We need to build the Docker images, our reproducible computational environments. Docker images are built from a set of instructions, the Dockerfile.

After changing to the current directory scipy-tutorial-2014/environement/docker, build the base image via:

{{ d['build-and-run.sh|idio']['build-reproducible'] }}

It's also possible to use the pregenerated Docker image from the Docker Hub, via:

{{ d['hub-docker-image.sh|idio']['remove'] }}

Running Docker

To launch the base image and open a shell:

{{ d['enter-environment.sh|idio']['enter'] }}

You are now in a Docker container. The -i -t flags tells docker to start up a pseudo-terminal and keep stdin open. Enter exit to exit the shell.

This runs the Docker container in the background, makes the ports listening available on the host, mounts the current working directory, and gives the container the name ipython.

Non-Interactive Containers

We can also configure containers to perform unattended tasks.

Dexy Image

We can build other images on top of reproducible/base to handle common tasks.

{% set dimg = "Dockerfile-dexy" %}

The {{ dimg }} image is used to run the dexy command.

Here's the Dockerfile:

{{ d[dimg + "|idio"] }}

It's built via:

{{ d['build-and-run.sh|idio']['build-dexy'] }}

And run via:

{{ d['build-and-run.sh|idio']['run-dexy'] }}

This mounts the scipy-tutorial-2014/dexy directory within the container and then runs dexy. The dexy-generated output files are present on your local file system after the container exits because of the volume mounting.

IPython Image

{% set dimg = "Dockerfile-ipython" %}

The {{ dimg }} image is used to launch the IPython notebook server.

Here's the Dockerfile:

{{ d[dimg + "|idio"] }}

It's built via:

{{ d['build-and-run.sh|idio']['build-ipython'] }}

To launch a non-interactive container which will run the IPython notebook server and forward the port IPython is running on, do:

{{ d['build-and-run.sh|idio']['start-ipython'] }}

To get information on the containers running and the ports available from a given container,

{{ d['build-and-run.sh|idio']['ipython-ps'] }}

If this returns 0.0.0.0:49153, we will use the 49153 port. On Linux, point your browser to http://localhost:49153. On Mac on Windows, use boot2docker to find the correct IP address,

boot2docker ip

If this returns 192.168.59.103, then point your browser to http://192.168.59.103:49153.

To stop and remove the container:

{{ d['build-and-run.sh|idio']['ipython-stop'] }}

Docker Help

For a full list of docker commands or docker subcommand help, enter docker or docker subcommand with no arguments.

{{ d['help.sh|idio|shint|pyg']['main-help'] }}

{{ d['help.sh|idio|shint|pyg']['subcommand-help'] }}