Has anyone tried the new TensorWorks CI/CD solution with Kubernetes and Unreal Engine? Building Windows Unreal Engine projects in a Kubernetes cluster | TensorWorks
I’m interested in learning about user experiences, performance benchmarks, and pricing details. Specifically, I’m looking for insights on whether this solution is cost-effective compared to other options, especially considering the high costs of AWS.
Hey I am Alex from the video, I hope it was useful.
The workflow presented works local and for cloud deployments.
The complicated part is getting a version of UE5 from source and compiling it on windows with Linux support. Once you have your local server building, containerized and running on docker or podman the rest is about automating the deployment.
Deploying scalable and globally available game servers on GKE is very straight forward.
I have used the same approach to deploy game servers to Google, Azure, and other cloud K8 providers. That’s the magic of cloud native software, it will just run your container.
In this repo i added sample files for Lyra, you can try them out.
I’m looking at your example, which by the way, looks really good and seems pretty simple.
I’ve never heard of the agones API, which is a kubernetes extension made to run a type of application.
I find it fascinating. When I have time I’ll try it.
What I’m doing is a bit more complex. I’m creating the cluster myself with K3s. Cloud providers are fine for something in development if you pay 70 - 350 euros/month. If you want something serious you’re going to pay 3000 euros/month perfectly. That’s why I prefer to create the solutions myself.
In my case it would be to create the “cluster” with K3s and then look at deploying the solution you propose as agones. I have to look at a couple of things to do everything completely and especially the performance issue.
The complicated part is getting a version of UE5 from source and compiling it on windows with Linux support.
Totally agree. But I don’t think it’s very complicated. You just have to run the server as headless if I’m not mistaken.
That is
1 - Dedicated Server Compilation: First, you need to compile your UE5 project with dedicated server support. This generates the necessary binaries for the server.
2 - Create the Docker Image: In the Docker image, you only need to include the server binaries and the necessary configuration files. You don’t need to include all the UE5 source code.
3 - Configuration and Deployment: Configure your Dockerfile to copy the necessary binaries and files to the container. Then, you can deploy this image to your Kubernetes or K3s cluster.
I hope I have time, I’m with springboot, react, jenkins… too many things.