Hi,

Thanks for sharing this great idea. I tried replicating the result on Minikube (which is what we're using), but the steps from the readme do not produce a deployment at my end. I didn't change a single line in the checkout, but this is the result when I run the build and apply the echo resource:

...
/home/reinv/.cargo/bin/cargo run --color=always
   Compiling k8s-openapi v0.13.0
   Compiling kube-core v0.60.0
   Compiling kube v0.60.0
   Compiling kube-runtime v0.60.0
   Compiling operator v0.1.0 (/home/reinv/Documents/Git/rust-kubernetes-operator-example)
    Finished dev [unoptimized + debuginfo] target(s) in 17.45s
     Running `target/debug/operator`
Reconciliation successful. Resource: (ObjectRef { dyntype: (), name: "test-echo", namespace: Some("default") }, ReconcilerAction { requeue_after: Some(10s) })
Reconciliation successful. Resource: (ObjectRef { dyntype: (), name: "test-echo", namespace: Some("default") }, ReconcilerAction { requeue_after: Some(10s) })
Reconciliation successful. Resource: (ObjectRef { dyntype: (), name: "test-echo", namespace: Some("default") }, ReconcilerAction { requeue_after: Some(10s) })
Reconciliation successful. Resource: (ObjectRef { dyntype: (), name: "test-echo", namespace: Some("default") }, ReconcilerAction { requeue_after: Some(10s) })
...

But when I look in my Minikube dashboard, there's no deployment and no pods being deployed. Any ideas?

Thanks in advance!

1. Use a fully qualified finalizer name. Because of the way the finalizer code is written (clears all finalizers before delete), this will work with existing deployed echos, which isn't super important for an example app, but kind of neat. I discovered this when getting an error using this finalizer name on a deployment, and the kube builder book recommended adding a /finalizer suffix. https://book.kubebuilder.io/reference/using-finalizers.html
2. Make echo::delete a no-op if the deployment was manually deleted. Currently it just fails. Again, example app, but gives an example for doing it.
3. Check for existence of a deployment in the reconciliation path. If the deployment was manually deleted, recreate it. This one could certainly be removed from this PR for now. In the future I'd like to follow up with a PR that updates the replica count as the CRD changes.

A few small improvements. Wanted to offer them up. Feel free to close this PR if you're 👎 . Thanks!

Continuation of https://github.com/Pscheidl/rust-kubernetes-operator-example/pull/9

Use a fully qualified finalizer name. Because of the way the finalizer code is written (clears all finalizers before delete), this will work with existing deployed echos, which isn't super important for an example app, but kind of neat. I discovered this when getting an error using this finalizer name on a deployment, and the kube builder book recommended adding a /finalizer suffix. https://book.kubebuilder.io/reference/using-finalizers.html

@Pscheidl I'm so sorry for not responding to the other PR before. I completely missed the notification. Here's just the first commit from #9 .