Jupyter (formerly known as iPython Notebooks) is a wonderful tool to work with and analyze data using Python, providing the means to report and visualize by means of a nice web interface.
I started learning my way through using this by developing some business reporting, and in order for that to be worthwhile, I wanted to connect directly to live data. I have a few data stores I need to interact with, but most of what I needed was in Postgres and Oracle databases.
It's a slight hassle to get it all set up, so I thought I'd take advantage of Docker. The starting point comes right from the Jupyter's docker-stacks and does not modify any of their conventions. This Dockerfile is as simple as I could think to make it, but it does require that you have - in the same directory as this Dockerfile - the "basic" and "SDK" Oracle Instant Client (for 64-bit Linux), which you can only get from Oracle by means of a free developer account (try here). At time of writing this README, the files I have are:
- instantclient-basic-linux.x64-12.1.0.2.0.zip
- instantclient-sdk-linux.x64-12.1.0.2.0.zip
So verifying the right files are in place and building looks something like this...
mac:~ me$ cd Documents/kpmartin-jupyter
mac:kpmartin-jupyter me$ ls
Dockerfile instantclient-basic-linux.x64-12.1.0.2.0.zip instantclient-sdk-linux.x64-12.1.0.2.0.zip
mac:kpmartin-jupyter me$ docker build -t papakpmartin/jupyter .I strongly recommend configuring a Docker Volume so that you can persist your work outside of the Docker container. Something like this is what I do...
docker run -p 8888:8888 -v /SOME/LOCAL/DIRECTORY/jupyter/work:/home/jovyan/work papakpmartin/jupyter
It's really kind of magical to be able to spin this up and be running in seconds and see all of your work there, ready to go.
Note: I do have this up in Docker Hub at papakpmartin/jupyter-pyspark-plus-oracle-and-postgres but it won't build there because of the Oracle dependencies, so I'm not yet sure that's a good place to get this. (I'm pretty new to Docker right now.)