Rate-limit scaling doesn't works for mergable ingresses

[vepatel]

Describe the bug
When configuring rate-limit scaling at master ingress, the rate-limit calculation based of replica count doesn't works (works fine with standard ingress, VS and VSRs)

To Reproduce

• Install edge build of NIC
• Configure mergable ingress example with rate-limit and scaling
• check directives inside the pod.

~/nginx/kubernetes-ingress/charts/nginx-ingress on tests/rate-limit-scaling ● λ k get ing                                                   
NAME                         CLASS   HOSTS              ADDRESS         PORTS     AGE
cafe-ingress-coffee-minion   nginx   cafe.example.com   35.240.31.112   80        71s
cafe-ingress-master          nginx   cafe.example.com   35.240.31.112   80, 443   72s
cafe-ingress-tea-minion      nginx   cafe.example.com   35.240.31.112   80        71s


~/nginx/kubernetes-ingress/charts/nginx-ingress on tests/rate-limit-scaling ● λ k describe ing cafe-ingress-master                          
Name:             cafe-ingress-master
Labels:           <none>
Namespace:        default
Address:          35.240.31.112
Ingress Class:    nginx
Default backend:  <default>
TLS:
  cafe-secret terminates cafe.example.com
Rules:
  Host        Path  Backends
  ----        ----  --------
  *           *     <default>
Annotations:  nginx.org/limit-req-key: ${binary_remote_addr}
              nginx.org/limit-req-rate: 50r/s
              nginx.org/limit-req-scale: true
              nginx.org/limit-req-zone-size: 10M
              nginx.org/mergeable-ingress-type: master
              

~/nginx/kubernetes-ingress/charts/nginx-ingress on tests/rate-limit-scaling ● λ k get po                                                    
NAME                                                     READY   STATUS    RESTARTS   AGE
test-release-nginx-ingress-controller-79754fc5bd-9qnfr   1/1     Running   0          121m
test-release-nginx-ingress-controller-79754fc5bd-hrgp9   1/1     Running   0          121m
test-release-nginx-ingress-controller-79754fc5bd-wdhc5   1/1     Running   0          121m


limit_req_zone ${binary_remote_addr} zone=default/cafe-ingress-coffee-minion:10M rate=50r/s;
limit_req_zone ${binary_remote_addr} zone=default/cafe-ingress-tea-minion:10M rate=50r/s;

Expected behavior

• It should be divided based on number of replicas

limit_req_zone ${binary_remote_addr} zone=default/cafe-ingress:10M rate=16r/s;

Your environment

• Version of the Ingress Controller - edge
• Version of Kubernetes - 1.29
• Kubernetes platform (e.g. Mini-kube or GCP) - GKE
• Using NGINX or NGINX Plus - OSS

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

[github-actions]

Hi @vepatel thanks for reporting!

Be sure to check out the docs and the Contributing Guidelines while you wait for a human to take a look at this 🙂

Cheers!

[vepatel]

@dbaumgarten #5113

[dbaumgarten]

Hi,
damn thats embarassing. So much testing and this still slipt through.
I opened a PR to fix that.

[vepatel]

no worries at all, thanks for the quick fix 👍🏼

[vepatel]

apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: cafe-ingress-master
  annotations:
    nginx.org/mergeable-ingress-type: "master"
    nginx.org/limit-req-rate: 50r/s
    nginx.org/limit-req-key: ${binary_remote_addr}
    nginx.org/limit-req-zone-size: 10M
    nginx.org/limit-req-scale: "true"
spec:
  ingressClassName: nginx
  tls:
  - hosts:
    - cafe.example.com
    secretName: cafe-secret
  rules:
  - host: cafe.example.com

limit_req_zone ${binary_remote_addr} zone=default/cafe-ingress-coffee-minion:10M rate=16r/s;
limit_req_zone ${binary_remote_addr} zone=default/cafe-ingress-tea-minion:10M rate=16r/s;

Fixed! thanks @dbaumgarten

[vepatel]

hey @dbaumgarten , there seems to be an issue when i'm installing NIC using manifests https://docs.nginx.com/nginx-ingress-controller/installation/installing-nic/installation-with-manifests/
the scaling doesn't seems to work at all. Can you please try it on your end and see if its the same?
Policy:

apiVersion: k8s.nginx.org/v1
kind: Policy
metadata:
  name: rate-limit-policy
spec:
  rateLimit:
    rate: 50r/s
    key: ${binary_remote_addr}
    zoneSize: 10M
    scale: true

replicas:

~/nginx/kubernetes-ingress/deployments on tests/rate-limit-scaling ● λ k get po -n nginx-ingress 
NAME                             READY   STATUS    RESTARTS   AGE
nginx-ingress-5676dd465f-bzhkb   1/1     Running   0          9s
nginx-ingress-5676dd465f-kq829   1/1     Running   0          9s
nginx-ingress-5676dd465f-mjlz5   1/1     Running   0          9s

conf:

limit_req_zone ${binary_remote_addr} zone=pol_rl_default_rate-limit-policy_default_webapp:10M rate=50r/s;

[dbaumgarten]

Hi,
this feature relies on a service being present for the ingress-pods. Also the controller must be configured with the correct name for the service.

The default deployment does not seem to set "-external-service" by default: https://github.com/nginxinc/kubernetes-ingress/blob/main/deployments/deployment/nginx-ingress.yaml#L96

Once this setting is set (and the referenced service exists) the scaling-feature should work just fine.