Not compliant when creating Gateway referring infrastructure

[hafe]

Describe the bug
If a gateway is created and referring to an NginxProxy through the infrastructure.parametersRef, it is not rejected if the NginxProxy does not exist.

The specification states in https://gateway-api.sigs.k8s.io/reference/1.4/spec/#gatewayinfrastructure:

"If the referent cannot be found, refers to an unsupported kind, or when
the data within that resource is malformed, the Gateway SHOULD be
rejected with the "Accepted" status condition set to "False" and an
"InvalidParameters" reason."

This was detected because the NodePort assigned was not expected but a random port. Thus no traffic could be routed through the dataplane.

No logs showing any problem.

To Reproduce
Just follow the getting started guide but create gateway first.

Expected behavior
Gateway rejected as spec state
Logs indicating the problem

Your environment
Kind
NGF 2.2.1
Kubernetes v1.34.0

Additional context
Add any other context about the problem here. Any log files you want to share.

[nginx-bot]

Hi @hafe! Welcome to the project! 🎉

Thanks for opening this issue!
Be sure to check out our Contributing Guidelines and the Issue Lifecycle while you wait for someone on the team to take a look at this.

[sjberman]

The spec states that the Gateway SHOULD be rejected, but not MUST be rejected. This was intentional on our part because we didn't want an update to the NginxProxy resource that was invalid to take down an entire set of infrastructure. That could be very disruptive to a user if they put in a simple typo.

Could you share a little more detail on the NodePort error/configuration you had? Trying to understand what the issue is and if we can be more descriptive.

[sjberman]

I'm assuming the condition was at least set on the Gateway status to show the error? That would be the main source of truth when there are problems with the spec.

[hafe]

I deployed Gateway and NginxProxy resources in the same "transaction" (using kustomize).
That created a service with the wrong NodePort that never got corrected.

Yes the status of the Gateway indicated the problem.

When I first create Gateway and then NginxProxy the service NodePort gets changed to the one I specified.

[hafe]

So I guess the problem you could consider is what happens when Gateway and NginxProxy are created in the same transaction.

Seems like it created a never healing state (wrong NodePort).

The consequence would be that no external traffic could ever reach the Gateway