Example Django and Docker setup

I'm just playing around now. Currently, this is a minimal Django installation with Postgres.

How to run (with fig)

This project uses fig to coordinate things, so the easy way to get up an running is this:

$ pip install fig
$ fig up

This will turn on your db service and your django service. Leave this running. You also need to migrate your database, so you can do it like this:

$ fig run web syncdb

How to run (without fig)

If you don't want to use fig (for example, if you want to know what's going on), here's the commands that I ran to get everything working. This is a lot more involved than the fig example, but I learned a ton doing it.

First run:

$ docker build -t djangodocker_web .

This will build the web image. Internally, the web image relies on the google/python image, but it also installs libpq-dev package so that we can install psycopg2. I originally wanted to use google/python-runtime, but I don't think it would be possible [1].

Next we want to get our database container up and running, we can run this:

$ docker run --name postgres \
         -e USER=docker \
         -e PASS=docker \
         -e DB=docker \
         paintedfox/postgresql

Note

I chose to use paintedfox/postgresql instead of the "official" postgres from stackbrew image because the official one isn't marked as Trusted and also I would rather have Postgres installed from a package manager than compiled like the way that stackbrew creates their images.

After that we should probably migrate our database:

$ docker run --rm -ti \
         --link postgres:db_1 \
         djangodocker_web \
         syncdb

I called that using the --rm option because I don't want to keep littering everything with all of these one-off containers. I think that's correct. The -ti options would be optional in a normal migration, but since the first migration asks you if you want to create a superuser, you need to have those. Now we can run our web container:

$ docker run --rm \
         --link postgres:db_1 \
         -p 8000:8080 \
         djangodocker_web

At first I thought this was broken because I didn't see any of the server start up messages that Django normally spits out, but then I went to http://localhost:8000/ and saw that it was working. If you want to see the start up messages, you can pass -t to the run command and then Django will spit the messages out at you:

$ docker run --rm -t \
         --link postgres:db_1 \
         -p 8000:8080 \
         djangodocker_web

This was cool and all, but then I tried to edit my settings file when I realized that code inside the container wasn't being updated with my code. So I turned off the web service and rebuilt the image and restarted the web service and then my new settings were in the code. So I have to rebuild and restart the server every time I change my code? Turns out no. Duh, obviously not. Using this whole thing would be a waste if I had to do that. I just have to mount my source code into the image, which Docker makes quite easy.:

$ docker run --rm -t \
         --link postgres:db_1 \
         -p 8000:8080 \
         -v /abs/path/to/dir/:/app \
         djangodocker_web

Now whenever I change the code, the changes appear in the browser.

[1]	Let's see what happens with this issue: <GoogleCloudPlatform/python-runtime#7>.