SystemCrash

It's very interesting, but why the Kubernetes overhead? Imagine adapting the client to receive the list of available channels from the cluster, and the cluster scaling them based on current traffic (with an horizontal pod autoscaler maybe). Cool, right? Having all configurations in a ConfigMap and making them shared across all pods/channels is something I'd really like... Even more interesting would be integrating some custom CRDs for various utilities. Also, there would be the possibility to distribute it via Helm, so you could have a ready-to-use server with a simple "helm install metin2".

@ Exynox, first of all, congratulations on your work, it's amazing. I was modifying server files back in 2010, and today I’ve become a cloud architect, starting my adventure with Metin... As a cloud architect, I was thinking: what do you think about making it scalable on a Kubernetes cluster? A few weeks ago, I returned to the Metin world, and the first idea that came to my mind was: 'What if the server could auto-scale and create channels based on traffic needs?' The first big obstacle, of course, was migrating the server files to Linux, but it seems like you’ve already done that. Let me know