bruce / docker_pydata_example

Example showing how to run a Python script (with pandas, numpy, etc) using Docker

Docker + Python (w/ pandas, numpy, etc) Example

This is a small example illustrating a Docker-based Python application that loads/updates/persists a state file.

The example merely appends a timestamp and a random integer to a persisted pandas dataframe on every run.

Installation

Install Docker for your system.

Clone this repository using Git:

$ git clone https://github.com/bruce/docker_pydata_example.git

Building it

Change into the directory:

$ cd docker_pydata_example

Build the docker image with make:

$ make build

(This may take awhile the first time, as it needs to grab the base images.)

Running it

Run it with make:

$ make run

You'll see something like this:

$ make run
mkdir -p data
docker run -v /home/bruce/code/github/bruce/docker_pydata_example/data:/app/data pydata_example
      timestamp  value
0  1.531632e+09   73.0

Try this several times, and you'll watch the output grow (the script adds an entry to its state on every run):

$ make run
mkdir -p data
docker run -v /home/bruce/code/github/bruce/docker_pydata_example/data:/app/data pydata_example
      timestamp  value
0  1.531632e+09   73.0
1  1.531632e+09   90.0
2  1.531632e+09   48.0
3  1.531632e+09   15.0
4  1.531633e+09   75.0

Exploring it

See:

• The Dockerfile, a whopping three lines that defines the base image (FROM), what files to add to the image (ADD), and what default command to execute (CMD).
• The application itself, under example/ (it's a pretty simple application, more functional then elegant or well-structured!)
• The Makefile, which defines how make builds the image and runs the container (creating and mounting a data directory, data/, as persistent storage across runs)
• After running it, the data/ directory, containing state.pkl.

Modifying it

Change the application (example/) however you'd like, then just run:

$ make

This will run the default task (all) which does a make build and make run in quick succession.

Resetting the data

You can clear the data with one command:

$ make clobber

Then your uses of make run will start from a clean slate (an empty dataframe, that is).