Kyverno is a handy policy engine for Kubernetes. For example, I use them to improve security by enabling user namespaces, or fix compatibility issues. I frequently use to validate or mutate resources that are created.

However, the YAML format can be a bit tricky for some of my more complicated policies. LLMs do alright creating many policies, but can struggle with my more complicated policies because they have to one-shot the policy and create it in one pass with no feedback or validation.

In this post, I’m going to show how I configure OpenWebUI to help me draft Kyverno policies.

As I switch from GitHub to Forgejo for my various project repositories, I find myself looking for a way to securely deploy new services into my Kubernetes cluster without having to keep updating Kubernetes auth tokens and storing them as a workflow secret.

GitHub Actions and, as of Forgejo v15, Forgejo Actions also supports providing temporary JWT tokens to workflow steps that securely state what repository and other commits details a given build is for. When a workflow is running, it can pass it to Kubernetes and other supported services and authenticate as a secure client with all the identity information of that repository and workflow. This is called OpenID Connect or OIDC.

This is super powerful because can federate identity and minimize the number of secrets required–great the next time a widely used workflow step gets compromised and steps all your secrets. Having fewer secrets helps.

In this post, I’m going to show how I configured Nix, Kubernetes, and Forgejo to securely update a deployment.

In my Kubernetes, I sometimes try to run a pod on a specific worker node. Maybe one of them has a folder that I need or a specific hardware characteristic. Historically, I’ve used Pod spec.nodeName: srv5 However, when that node becomes unavailable, say because it’s run out of disk space and has DiskPressure on it, then Kubernetes will continually try to spin up thousands of pods on it.

A screenshot of CPU usage growing, then Prometheus falls over and can’t scrap anymore.

In my previous post, I showed how to provision a Kubernetes node in NixOS maintaining compatibility with RKE1 (Rancher Kubernetes Engine v1), but switching to the Kubernetes nixpkg. In this post, I’m going to show how to take an Ubuntu worker and replace it with a NixOS based worker without rebuilding the cluster.

Everybody’s doing it. I guess I need to do AI, too. In my home, I have a few different tools that use generative AI and LLMs. I talk to my Home Assistant Voice Preview voice assistants which leverage a self-hosted Ollama running llama3.2. I use Open WebUI, tried Tabby as an experimental coding assistant. I use DeepInfra for larger models that don’t fit on my own GPU.

However, my problem is that each program supports different providers and models. Some support OpenAI style APIs to any provider, some only support Ollama APIs. If I wanted to forward my Home Assistant queries to DeepInfra, it wasn’t easy to do because there wasn’t an integration. If I wanted to change the model that Tabby uses between different models, I had to redeploy the service.