Access Policy

Access policies express which applications and services you are able to communicate with, both inbound and outbound. The default policy is to deny all incoming and outgoing traffic for your application, meaning you must be conscious of which services/application you consume, and who your consumers are.

In NAIS, we use Network Policies to express access policies. Network policies are applied to the pod, and are enforced by the network itself on layer 3 and 4 (TCP/IP). This means that the network will drop any traffic that is not allowed by the policy.

graph LR
  accTitle: Example Access Policy
  accDescr: The diagram shows three applications, A, B and C. Application A is allowed to communicate with B and C, but B and C are not allowed to communicate with A.

  A--"✅"-->B

  A--"✅"-->C

  B--"❌"-->A

  C--"❌"-->A

  subgraph accesspolicy-a[Access Policy A]
    A[Application A]
  end

  subgraph accesspolicy-b[Access Policy B]
    B[Application B]
  end

  subgraph accesspolicy-c[Access Policy C]
    C[Application C]
  end

Limitations

The access policy currently have the following limits:

1. No support for UDP. This is a limitation of Kubernetes, and will be fixed in the future.
2. Only for pod-to-pod communication. The Access policies only apply when communicating internally within the cluster using service discovery, and not through ingress traffic.
3. Only available in GCP. Network policies are only applied in GCP clusters. However, inbound rules for authorization in the context of TokenX or Azure AD apply to all clusters.

Access Policy vs. Ingress

Access policies are only applied to pod-to-pod communication, and not to ingress traffic. This means that you can still use ingress to expose your application to the internet, and use access policies to control which applications can communicate with your application.

Outbound requests to ingresses are regarded as external hosts, even if these ingresses exist in the same cluster.

Inbound rules

Inbound rules specifies what other applications in the same cluster your application receives traffic from.

Receive requests from other app in the same namespace

For app app-a to be able to receive incoming requests from app-b in the same cluster and the same namespace, this specification is needed for app-a:

nais.yamlvisualization

apiVersion: "nais.io/v1alpha1"
kind: "Application"
metadata:
  name: app-a
...
spec:
  ...
  accessPolicy:
    inbound:
      rules:
        - application: app-b

graph LR
  accTitle: Receive requests from other app in the same namespace
  accDescr: The diagram shows two applications in the same namespace, A and B. Application A is allowed to receive requests from B.

  app-b--"✅"-->app-a

  subgraph mynamespace
    app-a
    app-b
  end

Receive requests from other app in the another namespace

For app app-a to be able to receive incoming requests from app-b in the same cluster but another namespace (othernamespace), this specification is needed for app-a:

nais.yamlvisualization

apiVersion: "nais.io/v1alpha1"
kind: "Application"
metadata:
  name: app-a
...
spec:
  ...
  accessPolicy:
    inbound:
      rules:
        - application: app-b
          namespace: othernamespace

graph LR
  accTitle: Receive requests from other app in the another namespace
  accDescr: The diagram shows two applications in different namespaces, A and B. Application A is allowed to receive requests from B.

  app-b--"✅"-->app-a

  subgraph mynamespace
    app-a
  end

  subgraph othernamespace
    app-b
  end

Outbound rules

spec.accessPolicy.outbound.rules specifies which applications in the same cluster you allow your application to send requests to. To open for external applications, use the field spec.accessPolicy.outbound.external.

Send requests to other app in the same namespace

For app app-a to be able to send requests to app-b in the same cluster and the same namespace, this specification is needed for app-a:

nais.yamlvisualization

apiVersion: "nais.io/v1alpha1"
kind: "Application"
metadata:
  name: app-a
...
spec:
  ...
  accessPolicy:
    outbound:
      rules:
        - application: app-b

graph LR
  accTitle: Send requests to other app in the same namespace
  accDescr: The diagram shows two applications in the same namespace, A and B. Application A is allowed to send requests to B.

  app-a--"✅"-->app-b

  subgraph mynamespace
    app-a
    app-b
  end

Send requests to other app in the another namespace

For app app-a to be able to send requests requests to app-b in the same cluster but in another namespace (othernamespace), this specification is needed for app-a:

nais.yamlvisualization

apiVersion: "nais.io/v1alpha1"
kind: "Application"
metadata:
  name: app-a
...
spec:
  ...
  accessPolicy:
    outbound:
      rules:
        - application: app-b
          namespace: othernamespace

graph LR
  accTitle: Send requests to other app in the another namespace
  accDescr: The diagram shows two applications in different namespaces, A and B. Application A is allowed to send requests to B.

  app-a--"✅"-->app-b

  subgraph mynamespace
    app-a
  end

  subgraph othernamespace
    app-b
  end

External services

External services are services that are not running in the same cluster, but are reachable from the cluster. This could be services running in other clusters, or services running in the same cluster but outside the cluster network.

Since this is not a native feature of Kubernetes Network Policies, we are leveraging Linkerd's Service Profiles to achieve this.

In order to send requests to services outside of the cluster, external.host configuration is needed:

nais.yamlvisualization

apiVersion: "nais.io/v1alpha1"
kind: "Application"
metadata:
  name: app-a
...
spec:
  ...
  accessPolicy:
    outbound:
      external:
        - host: www.external-application.com

graph LR
  accTitle: External services
  accDescr: The diagram shows an application, A, that is allowed to send requests to an external service.

  app-a--"✅"-->www.external-application.com

  subgraph cluster
    subgraph mynamespace
      app-a
    end
  end

Global Service Entries

There are some services that are automatically added to the mesh in dev-gcp and prod-gcp (search for global_serviceentries).

Advanced: Resources created by Naiserator

The previous application manifest examples will create Kubernetes Network Policies.

Kubernetes Network Policy

Default policy

Every app created will have this default network policy that allows traffic to Linkerd and kube-dns. It also allows incoming traffic from the Linkerd control plane and from tap and prometheus in the linkerd-viz namespace. This is what enables monitoring via the linkerd dashboard. These policies will be created for every app, also those who don't have any access policies specified.

apiVersion: extensions/v1beta1
kind: NetworkPolicy
metadata:
  labels:
    app: appname
    team: teamname
  name: appname
  namespace: teamname
spec:
  egress:
    - to:
        - namespaceSelector:
            matchLabels:
              linkerd.io/is-control-plane: "true"
        - namespaceSelector: {}
          podSelector:
            matchLabels:
              k8s-app: kube-dns
        - ipBlock:
            cidr: 0.0.0.0/0
            except:
              - 10.6.0.0/15
              - 172.16.0.0/12
              - 192.168.0.0/16
  ingress:
    - from:
        - namespaceSelector:
            matchLabels:
              linkerd.io/is-control-plane: "true"
    - from:
        - namespaceSelector:
            matchLabels:
              linkerd.io/extension: viz
          podSelector:
            matchLabels:
              component: tap
    - from:
        - namespaceSelector:
            matchLabels:
              linkerd.io/extension: viz
          podSelector:
            matchLabels:
              component: prometheus
  podSelector:
    matchLabels:
      app: appname
  policyTypes:
    - Ingress
    - Egress

Kubernetes network policies

The applications specified in spec.accessPolicy.inbound.rules and spec.accessPolicy.outbound.rules will append these fields to the default Network Policy:

apiVersion: extensions/v1beta1
kind: NetworkPolicy
...
spec:
  egress:
  - to:
    ...
    - namespaceSelector:
        matchLabels:
          name: othernamespace
      podSelector:
        matchLabels:
          app: app-b
    - podSelector:
        matchLabels:
          app: app-b
  - from:
    - namespaceSelector:
        matchLabels:
          name: othernamespace
      podSelector:
        matchLabels:
          app: app-b
    - podSelector:
        matchLabels:
          app: app-b
  podSelector:
    matchLabels:
      app: appname
  policyTypes:
  - Egress
  - Ingress

Info

Note that for namespace match labels to work, the namespaces must be labeled with name: namespacename.

kube-system should be labeled accordingly for the default rule that allows traffic to kube-dns, but in GCP, the label is removed by some job in regular intervals...

If you are working directly with Kubernetes Network Policies, we are recommending the Cilium Policy Editor which can be found at editor.cilium.io.

---

Last update: 2022-10-27
Created: 2019-10-04