metacontroller VS sample-controller

It worth adopt proper controller development with something like Metacontroller, Kopf or Java Operator SDK.

There are many frameworks, like the Operator Framework (https://operatorframework.io/) to the MetaController (https://github.com/metacontroller/metacontroller) to KubeBuilder(https://github.com/kubernetes-sigs/kubebuilder) to the Kubernetes Operator Framework (kopf, https://kopf.readthedocs.io/en/stable/), among others.

You might like a little alternative to writing your own in the first instance:) take a look at https://metacontroller.github.io/metacontroller/

Codegen is not magic, but if you want to make it simple stupid, supportable and prod-usable I'd go for metacontroller. There's not much magic to it - just common admission/validation webhooks propagated to your controller implementation, alongside the simplest possible reconciliation cycle. We had adopted Knative driven metacontrollers and it's a pretty solid approach.

https://github.com/metacontroller/metacontroller - Sounds cool! I will look into it.

When I need a quick mutating controller, I like to use MetaController, which makes rapid prototyping really easy: https://metacontroller.github.io/metacontroller/

If you are more into lightweight and common use cases, look at metacontroller or the shell-operator.

In case you've not heard of it, the metacontroller aims to do exactly this. It provides the controller-lifecycle (registering your types, watching the API for changes, etc), you just need to provide a HTTP webhook to receive the JSON object, and respond with the modifications\resources you want.

- Sample controller: https://github.com/kubernetes/sample-controller

The most low-level approach I have seen is using the client-go package. Both kubebuilder and controller-runtime are based on this package. You can see an official example here: https://github.com/kubernetes/sample-controller/blob/master/docs/controller-client-go.md But even that sample requires some code generation on top of it https://cloud.redhat.com/blog/kubernetes-deep-dive-code-generation-customresources

I would still recommend gitOps tools or if you want to create your own abstraction (CRD) that encapsulates whatever behaviour you want to watch/act upon, take a look into kubernetes sample controller (https://github.com/kubernetes/sample-controller) or the kubernetes operator pattern https://kubernetes.io/docs/concepts/extend-kubernetes/operator/

Either should be fine. In case though you want something more basic then consider sample-controller: https://github.com/kubernetes/sample-controller

As far as Operator development goes, if you are comfortable with golang then I would recommend checking out sample-controller https://github.com/kubernetes/sample-controller It does not abstract the underlying client-go machinery and so provides a good way to understand how various parts of an Operator come together. On the other hand, if you don't want to get exposed to that much details then either Operator SDK or Kubebuilder will be good options. In case you are not into golang then there are SDKs for other languages as well (Python, Java, etc.).

Cool! Given that you mention you are a beginner in K8s operator development, here is a plan that I would like to suggest. 1. Get basic understanding of Kubernetes custom controllers first. The sample controller repo is a good place to start. https://github.com/kubernetes/sample-controller

the sample controller using client-go to learn about informers: https://github.com/kubernetes/sample-controller