Ingress Requests Drop Unexpectedly in RKE2 Cluster with Cilium (v1.30.6 +rke2r1)

[Halilmehmet]

Description of the Issue

In an RKE2 cluster, requests to applications suddenly stop being processed by the rke2-ingress-nginx-controller.

Restarting the rke2-ingress-nginx-controller pod temporarily resolves the issue, but it reoccurs after some time. I suspect the issue is related to Cilium. However, I cannot capture any meaningful error logs.

---

Steps to Reproduce

1. Deploy an RKE2 cluster with the following configurations:
   • Cilium is enabled with kubeProxyReplacement: true.
   • disable-kube-proxy: true in the RKE2 configuration.
2. Send requests to applications through the ingress (e.g., HTTP service).
3. After some time, requests stop being processed.
4. Restarting the rke2-ingress-nginx-controller pod temporarily resolves the issue.

---

Environment Information

• RKE2 Version: v1.30.6 +rke2r1
• Operating System: Ubuntu 24.04
• Cluster Nodes: 3 server, 4 agent
• CNI: Cilium

---

#/etc/rancher/rke2/config.yaml
tls-san:
  - 10.10.10.10
token: token
node-taint:
  - "CriticalAddonsOnly=true:NoExecute"
cni:
  - cilium
disable-kube-proxy: "true"

#/var/lib/rancher/rke2/server/manifests/rke2-cilium-config.yaml
apiVersion: helm.cattle.io/v1
kind: HelmChartConfig
metadata:
  name: rke2-cilium
  namespace: kube-system
spec:
  valuesContent: |-
    kubeProxyReplacement: true
    k8sServiceHost: 127.0.0.1
    k8sServicePort: 6443
    eni:
      enabled: true
    hubble:
      enabled: true
      relay:
        enabled: true
      ui:
        enabled: true
        frontend:
          image:
            tag: "v0.12.3"
        backend:
          image:
            tag: "v0.12.3"

---

I would appreciate your assistance in diagnosing and resolving this issue. Thank you!

[brandond]

Duplicate of #7435