使用 Azure CNI 网络策略，仅允许从 Kube.netes pod 出口到特定的 FQDN/DNS

Question

How can egress from a Kube.netes pod be limited to only specific FQDN/DNS with Azure CNI Network Policies?

This is something that can be achieved with:

Istio

apiVersion: config.istio.io/v1alpha2
kind: EgressRule
metadata:
  name: googleapis
  namespace: default
spec:
  destination:
      service: "*.googleapis.com"
  ports:
      - port: 443
        protocol: https

Cilium

apiVersion: "cilium.io/v2"
kind: CiliumNetworkPolicy
metadata:
  name: "fqdn"
spec:
  endpointSelector:
    matchLabels:
      app: some-pod
  egress:
  - toFQDNs:
    - matchName: "api.twitter.com"  
  - toEndpoints:
    - matchLabels:
        "k8s:io.kubernetes.pod.namespace": kube-system
        "k8s:k8s-app": kube-dns
    toPorts:
    - ports:
      - port: "53"
        protocol: ANY
      rules:
        dns:
        - matchPattern: "*"

OpenShift

apiVersion: network.openshift.io/v1
kind: EgressNetworkPolicy
metadata:
  name: default-rules 
spec:
  egress: 
  - type: Allow
    to:
      dnsName: www.example.com
  - type: Deny
    to:
      cidrSelector: 0.0.0.0/0

How can something similar be done with Azure CNI Network Policies?

Answer 1

Apply K8s network policies

`

apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
  name: default-rules
spec:
  podSelector:
    matchLabels:
      role: pod_role
  policyTypes:
  - Egress
  egress:
  - to:
    - ipBlock:
        cidr: 0.0.0.0/0
    - host: www.example.com

`

reference https://kubernetes.io/docs/concepts/services-networking/network-policies/

Answer 2

ATM network policies with FQDN/DNS rules are not supported on AKS.

If you use Azure CNI & Azure Policy Plugin you get the default Kubernetes Network Policies.

If you use Azure CNI & Calico Policy Plugin you get advanced possibilities like Global Network Polices but not the FQDN/DNS one. This is a paid feature on Calico Cloud unfortunately.

Answer 3

In case someone is hitting this page from google:

I found a solution that works nicely on my cloud provider (OpenTelekomCloud) and probably will on many other.

There is a project called gke-fqd.networkpolicies-golang

By defining a custom resource

apiVersion: networking.gke.io/v1alpha3
kind: FQDNNetworkPolicy
metadata:
  name: allow-test
  namespace: test1
spec:
  podSelector: {}
  egress:
    - to:
      - fqdns:
        - heise.de
      ports:
      - port: 443
        protocol: TCP
      - port: 80
        protocol: TCP

it will resolve the FQDNs, produce the final NetworkPolicy and update the records every 30 seconds. This is what the final Policy will look like

apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
  name: allow-test
  namespace: test1
  annotations:
    fqdnnetworkpolicies.networking.gke.io/owned-by: allow-test
spec:
  podSelector: {}
  egress:
    - ports:
        - protocol: TCP
          port: 443
        - protocol: TCP
          port: 80
      to:
        - ipBlock:
            cidr: 128.65.210.8/32
  policyTypes:
    - Ingress
    - Egress

I had to append following permissions to the clusterRole fqd.networkpolicies-manager-role in the yaml (downloaded from the release page) to make it work outside GKE

apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: fqdnnetworkpolicies-manager-role
rules:

...

- apiGroups:
  - networking.k8s.io
  resources:
  - networkpolicies
  verbs:
  - create
  - delete
  - get
  - list
  - patch
  - update
  - watch
- apiGroups:
  - networking.k8s.io
  resources:
  - networkpolicies/status
  verbs:
  - get
  - patch
  - update