The ALB listeners are not auto removed when I delete the corresponding ingress from EKS

I have deployed an AWS ALB Controller and I create listeners with ingress resources in a EKS cluster.

The steps I followed are the following:

I had an ingress for a service named first-test-api and all where fine
I deploy a new Helm release [first] with just renaming the chart from test-api to main-api. So now is first-main-api.
Noting seems to break in terms of k8s resources , but...
the test-api.mydomain.com listener in the AWS ALB is stuck to the old service

Has anyone encounter such a thing before?

I could delete the listener manually, but I don't want to. I'd like to know what is happening and why it didn't happened automatically :)

EDIT:

The ingress had an ALB annotation that enabled the deletion protection.

Solution

I will provide some generic advice on things I would look at, but it might be better to detail a small example.

Yes, ALB controller should automatically manage changes on the backend.

I would suggest ignoring the helm chart and looking into the actual objects:

kubectl get ing -n <namespace> shows the ingress you are expecting?
kubectl get ing -n <ns> <name of ingress> -o yaml points to the correct/new service?
kubectl get svc -n <ns> <name of new svc> shows the new service?
kubectl get endpoints -n <ns> <name of new svc> shows the pod you are expecting?

And then gut feeling.

Check the labels in your new service are differents from the labels in the old service if you expect to both services serve different things.
Get the logs of the ALB controller. You will see registering/deregistering stuff. Sometimes errors. Especially if the role the node/service account doesn't have the proper IAM permissions.

Happy to modify the answer if you expand the question with more details.