My First Cluster

So around 2022, I made my server (old desktop pc), so I could run my own Jellyfin server using NGINX reverse proxy and docker. I had the idea, that I could automate the process of downloading new movies, without having to manually look for them myself. I still pay for streaming services like Netflix and regularly attend my local cinema, but this would allow me to watch things all from one place.

Discovering The Movie Database API (TMDB)

I made a NodeJS project that would periodically check for new episodes, and movies from TMDB API. It would take the items found that were above 70% rating, then place them into JSON database file.

Walking the plank

I then created an other NodeJS project (plankwalker), to take the json file and search to the best possible torrents online. Plankwalker would then generate a list of magnets to submit to transmission.

Transmission Loud and Clear

Transmission is a Fast, Easy and Free open source Bittorrent Client. After learning that transmission had a web mode, I discovered that its RPC API was also well documented. So plankwalker could automate submitting new magnets to be downloaded to transmission. I then made a script that could have it organise the downloads into a folder that Jellyfin could pick up automatically. Seems perfect right?

My First Raspberry Pi

Around winter time, I got really into youtube channels like @JeffGeerling. So i bought the cheapest raspberry pi zero kit I could find on amazon. After realising how little amount of energy it took to power it and how much you can do with it, I began to consider moving Jellyfin over to a cluster of Raspberry Pis. Compared to the 600 watts my "server" was consuming, I could run 4 Raspberries at around 100 watts, saving 500 watts of energy.

K3S For the Win!

K3s is a highly available, certified Kubernetes distribution. It runs super well on Raspberry Pi. Plus it works really well with rancher, a web based management service for Kubernetes. K3s enabled this DevOps hating engineer, setup his very own cluster, with minimal hassle. I was able to easily port my docker containers over to it. Jellyfin ran well, but due to the low compute power of a raspberry pi, i could not use it to transcode the video in real time. So i would have to convert my videos into mp4 h264 AAC, in order to view them in any browser.

The results were better than I expected. It turned out, that my cluster only uses 30 watts of power, via Power over Ethernet. I am hooked!