Glia assessment

Python FastAPI

This repository contains a Python FastAPI server containing the following endpoints:

/jumble/ - Takes a word and re-arranges the characters randomly
/audit - Return the last 10 API calls made to the server

The solution uses a mysql container as database, which in turn uses a persistent volume claim and persistent volume, and an nginx ingress controller. These components are deployed as dependencies in the helm chart, so we don't need to manage and create extra charts.

The application is built as a container image. To build the image we can proceed with:

docker build -t progerjkd/glia-jumble .

Scan the image for vulnerabilities:

docker scan progerjkd/glia-jumble

Push image to docker hub

docker push progerjkd/glia-jumble

App architecture

Once deployed, the helm chart will provision the above architecture. There will be one service for the glia-jumble-app pod, and one service for the mysql pod. The the glia-jumble-app deployment will import a secret containing the database username and password. The mysql deployment will provision a persistent volume, and a persistent volume claim to store the database metadata directory.

Deploy the solution to minikube

The diagram below ilustrates the kubernetes resources used in the helm chart.

We can launch the application in a local minikube cluster, by using the provided helm chart.

# start minikube
minikube start

# install the helm chart
helm install glia-jumble-app -f helm/glia-jumble/values.yaml helm/glia-jumble/

The above command creates a helm release called glia-jumble-app of the glia-jumble helm chart.

We can fetch the minikube IP and the service’s NodePort:

minikube service glia-jumble-app --url

http://127.0.0.1:64270
❗  Because you are using a Docker driver on darwin, the terminal needs to be open to run it.

As the mysql server may take some time to be ready, the application may fail to connect until it establishes the database connection. We can check for the pods status by using kubectl.

kubectl get pods  -w

Testing

To test the application we can access the exposed service (using the port from the above command):

http://127.0.0.1:64270/jumble/word_to_be_jumbled

We get:

{"jumbled":"te_owdb__ubodmjler"}

We can visit the /audit endpoint to check the last requests to the server:

http://127.0.0.1:64270/audit

[{"id":13507,"timestamp":"2023-02-23T04:33:31","url":"http://10.244.0.85:8000/","payload":"/","status":200,"duration":4.485},{"id":13506,"timestamp":"2023-02-23T04:33:31","url":"http://10.244.0.85:8000/","payload":"/","status":200,"duration":0.246},{"id":13505,"timestamp":"2023-02-23T04:33:26","url":"http://127.0.0.1:64270/audit","payload":"/audit","status":200,"duration":96.827},{"id":13504,"timestamp":"2023-02-23T04:33:21","url":"http://10.244.0.85:8000/","payload":"/","status":200,"duration":16.78},{"id":13503,"timestamp":"2023-02-23T04:33:21","url":"http://10.244.0.85:8000/","payload":"/","status":200,"duration":0.32},{"id":13501,"timestamp":"2023-02-23T04:33:11","url":"http://10.244.0.85:8000/","payload":"/","status":200,"duration":0.237},{"id":13502,"timestamp":"2023-02-23T04:33:11","url":"http://10.244.0.85:8000/","payload":"/","status":200,"duration":5.041},{"id":13500,"timestamp":"2023-02-23T04:33:01","url":"http://10.244.0.85:8000/","payload":"/","status":200,"duration":4.437},{"id":13499,"timestamp":"2023-02-23T04:33:01","url":"http://10.244.0.85:8000/","payload":"/","status":200,"duration":0.249},{"id":13498,"timestamp":"2023-02-23T04:32:51","url":"http://10.244.0.85:8000/","payload":"/","status":200,"duration":4.557}]

The log entries are composed by the following fields:

id - incremental value
timestamp - date and time of the request
url - the requested URL
payload - the arguments passed to the API
status - HTTP status code
duration - processing time in miliseconds

We may see requests to / due to the kubernetes readiness probe.

We can optionally scale the deployment glia-jumble-app:

kubectl scale --replicas=2 deployment glia-jumble-app

Extending the solution

One can extend the API by rebuilding the docker image, pushing to the docker hub, and launching a new revision for the helm chart, or by using the build.sh script:

docker build -t progerjkd/glia-jumble .
docker push progerjkd/glia-jumble

helm upgrade glia-jumble-app -f helm/glia-jumble/values.yaml helm/glia-jumble/

progerjkd/glia-assessment