Filip Hoffmann

Pilkanahali (PNH) is a Polish web platform that gathers sports enthusiasts and enables them to team up and play soccer, basketball, volleyball, and other team sports. Users must register for every single game, and for popular locations, there are a lot of people competing for the spots. I love to play football, and I was very frustrated that signing up for a game in my neighborhood was always very difficult due to the number of people interested in playing. Therefore, I wrote an application that waits for the new sign-ups to get posted and registers me automatically.

The application was built and deployed to the Docker Hub repository and run on a GCP instance created with Terraform script. It queried the PNH portal for the new events within a specified date range and location, and if there were available spots, it would sign the user up.

The subscription processing API was deployed as a pyramid application on two EC2 instances sitting behind the load balancer. My team had experienced some random EC2 instance crashes before. Hence, another senior developer and I created the failover procedure that destroyed the malfunctioning server and set up a new server bound to the same elastic IP address that the crashed server used. We described the failover instance in Terraform and left the code commented, so in case of failover, the on-duty developer had to comment out the code for the crashed instance and uncomment the code for the failover instance, rerun the Terraform script and deploy the application to a new server with standard deployment procedure. The whole procedure was described and very easy to follow, reducing the human error possibility during the server crash.

In this project, I was responsible for describing the failover infrastructure in the Terraform scripts and writing the manual for on duty developer.

The procedure proved very useful as one of our servers crashed the day after we finished it. We got it back online in around 10 minutes.

The client needed a universal tool that would automatically verify if all of the clickable links on their websites return 200 status code so they can test if the website works correctly after some back-end reworks. Me and two other developers created a Python application that used Scrapy and pytest to recursively go through all of the URLs on a website, starting from the root URL and returning the test report with the URL mapped to its corresponding response code. As some of the client websites were difficult to scrape 100% automatically, we included the possibility to extend the test case with the following scenario: a set of requests to the website that were made before Scrapy started its run to prepare the website and its content (for example, create the new user and submit some data).

In this project, I was responsible for designing the whole application from scratch, implementing the scraper, creating Bazel build script, and helping my colleagues whenever necessary.