This is a proof of concept deployment of a simple Python stub and MongoDB, intended to showcase containerization of my fantasy football webapp sleeperPy.
Digital Ocean graciously supported this project with a donation to charity.
A simple Python docker container grabs player data from Sleeper Fantasy Football and imports it into a MongoDB collection. I hope to eventually pair this with a web frontend of Kubernetes use, ideally seperated out into microservices. The current stub would be nice with the Kubernetes CronJob object.
Using a kubeconfig file from your cloud provider, run the deployment.
cd deployments
kubectl apply -f .
I also used kompose to assist with creating the initial deployment file.
To continue development, after changes are made:
docker-compose build
docker-compose push
k apply -f deployments/
This is a standard Kubernetes multi-container pod deployment:
spec:
replicas: 3
selector:
matchLabels:
io.kompose.service: sleeperpy
template:
metadata:
annotations:
kompose.cmd: kompose convert
kompose.version: 1.26.0 (40646f47)
creationTimestamp: null
labels:
io.kompose.service: sleeperpy
spec:
containers:
- image: wbollock/sleeperpy:latest # my image
name: sleeperpy
- image: mongo:latest
name: mongodb
When making changes to our application, all we need to do is run a docker-compose build and docker-compose push that upload our latest image to Docker Hub. Re-applying the deployment with k apply -f . pulls the subsequent upload automatically and does a RollingRestart deploy method.
import_data.py is a simple script to pull NFL player data and import it into a MongoDB collection.
with urllib.request.urlopen(url) as url:
data = json.loads(url.read().decode())
with open(playersFile, 'w') as f:
json.dump(data,f)
f.close()
Curiously, I needed to manually install mongodb tools to obtain mongoimport, as PyMongo doens't seem to have the ability to import JSON data into a remote database. Perhaps I'm missing something.
RUN wget https://fastdl.mongodb.org/tools/db/mongodb-database-tools-debian10-x86_64-100.5.1.deb
RUN apt install ./mongodb-database-tools-debian10-x86_64-100.5.1.deb
It's important to note with a multi-container pod setup, localhost with actually be the database hostname, not something like the container_name in a regular docker-compose setup.
While eventually a Flask component or similar would be nice to display this data and fully rewrite sleeperPy to be container-compatible, for now, here is data from the Kubernetes pods.
I didn't structure this data perfectly, but it's clear that this is a distributed NoSQL deployment.
We're running a kubectl exec command here to query data from a single MongoDB container in one pod.
k exec --stdin --tty <one pod> -c mongodb -- mongo localhost:27017/sleeperPy --eval "db.players.find().pretty()" | head -50
Lastly, to see what our application is doing, we can query logs for it.
There isn't much to log here, but we can see that we're connecting to localhost successfully and importing our document.