Add docs about how to avoid hanging finalizers during shut down

[kaganjd]

Description

If you are running the operator in the same namespace as your application, and you delete the entire namespace, you may end up with hanging finalizers like this:

  NamespaceContentRemaining                    True    Tue, 07 Jan 2025 14:55:19 -0800  SomeResourcesRemain   Some resources are remaining: domains.ingress.k8s.ngrok.com has 1 resource instances, httpsedges.ingress.k8s.ngrok.com has 1 resource instances, ingresses.networking.k8s.io has 1 resource instances, tunnels.ingress.k8s.ngrok.com has 1 resource instances
  NamespaceFinalizersRemaining                 True    Tue, 07 Jan 2025 14:55:19 -0800  SomeFinalizersRemain  Some content in the namespace has finalizers remaining: k8s.ngrok.com/finalizer in 4 resource instances

We should add docs suggesting:

1. running the ngrok-operator in its own namespace
2. shut down ngrok-operator namespace first, then application namespace
3. if this isn't possible, run a script to delete all the finalizers on ngrok custom resources. @Alice-Lilith had the idea to make this a Helm pre-delete hook

script:

1. first, set your namespace: namespace=set-your-namespace
2. then run:

declare -a crds=("domains" "ingresses" "tunnels" "httpsedges")

for i in "${crds[@]}"; do for j in $(kubectl get $i -n $namespace -o name | cut -d '/' -f 2-); do kubectl patch $i $j -n $namespace -p '{"metadata":{"finalizers":[]}}' --type=merge; done; done

Use Case

No response

Related issues

No response