Policy Enforcement on Digi Hunch

Istio External Authorization via OIDC

Fri, 25 Feb 2022 01:29:00 -0400

Istio service mesh allows application developers to offload non-core features to infrastructure layer. We explored authentication and authorization with Istio in a basic lab. In this post we continue to explore its capabilities with OIDC integration. This capability is made available thanks to the CUSTOM action in authorization policy, supported since the release of 1.9. It enables any workload on Istio to integrate with an external IAM solution.

The rest of this post, provides the step-by-step instruction to configure OIDC integration, based on Istio’s External Authorization use case. My work is influenced by two blog posts from jetstack and elastisys on similar topic, with my own additions, simplifications and clarifications. In this lab I use my own DNS hostname demo1.digihunch.com and to follow along you need to bring your own hostname as well. The prerequisites of this lab is summarized as below:

An identity provider: we use Google in the lab but it can be anything with OIDC capabilities (e.g. Azure AD, Facebook). Read my old post for more details on OIDC.
A public IP address for Cluster’s Ingress service: Since we use Google as identity provider on the Internet, we need our service to be exposed on the Internet because the OIDC integration involves a two-way HTTP redirect.
Public DNS resolution to the service’s public IP: during authentication, the identity provider will call back with JWT. So we need our hostname demo1.digihunch.com resolved to the service’s public IP address. This is also required in other scenarios. For example, cert-manager needs to automatically configure X.509 certificate using Let’s Encrypt.
An application workload: this is the tenant workload without its own IAM capability. We use a dummy workload base on a hello-world image, for simplicity.

The files required in this lab is in istio-oidc repo under the envoy directory. It is also important to understand the Authentication Code flow in OIDC authorization code flow to make sense of the integration between OIDC provider and Istio on our platform. For example, the client application needs a one-time registration with identity provider. Let’s start with this configuration.

Configure Identity Provider

We need a Google account with a GCP project and log on to APIs & Services to register our application. At the console, from the left side bar, click on “OAuth consent screen” and create an App. Provide the followings:

Step 1. OAuth consent screen. Under App Information, provide App name and User support email.
Step 1. OAuth consent screen. Under App domain, provide Application home page
Step 1. OAuth consent screen. Under Authorized domains, provide an Authorized domain.
Step 3. Test users. Add the emails for Test users.

The summary page looks like this:

OAuth Consent Screen

Then on the sidebar, click on Credentials. We create new credentials for OAuth 2.0 Client ID. Select Web application and add “https://demo1.digihunch.com/oauth2/callback” as an Authorized redirect URI. The credential created includes a Client ID and a Client secret, which can be downloaded as a JSON file. Here is my screen:

API Credential

Note that the redirect URIs cannot be localhost. Because the browser will consume the redirect URI and it has no clue where localhost is. It must resolve to the public IP of the ingress.

We need to keep the Client ID and Client secret for later use. According to Google’s documentation, there is a discovery document from which we can understand the keys in their Open ID configuration. We need the value of two keys (issuer and jwks_url) to use later in the configuration.

Configure Infrastructure Services

My real-quicK-cluster project has a few different ways to quickly bring up a cluster. For this lab, I use Azure CLI command to create a simple, three-node cluster:

$ az aks create \
 -g AutomationTest \
 -n orthCluster \
 --node-count 3 \
 --enable-addons monitoring \
 --generate-ssh-keys \
 --vm-set-type VirtualMachineScaleSets \
 --network-plugin azure \
 --network-policy calico \
 --tags Owner=MyOwner

Then we configure kubeconfig credential to connect to the cluster, so that kubectl and helm CLI commands to connect to the newly created cluster.

$ az aks get-credentials --resource-group AutomationTest --name orthCluster

Then we will install the infrastructure services and workload. We can start with loading environment variables for use later:

OIDC_DISCOVERY=$(curl "https://accounts.google.com/.well-known/openid-configuration")
OIDC_ISSUER_URL=$(echo $OIDC_DISCOVERY | jq -r .issuer)
OIDC_JWKS_URI=$(echo $OIDC_DISCOVERY | jq -r .jwks_uri) 
COOKIE_SECRET=$(openssl rand -base64 32 | tr -- '+/' '-_')
WEB_HOST="demo1.digihunch.com"
CLIENT_ID="ThisIsTheClientIDFromGoogle"
CLIENT_SECRET="ThisIsTheClientSecretFromGoogle"

For WEB_HOST, CLIENT_ID, CLIENT_SECRET, you need to bring your own values. The COOKIE_SECRET value is randomly generated. The variables OIDC_ISSUER_URL and OIDC_JWKS_URI are parsed from Google OpenID configuration and they should remain static. We can validate the value with echo commands. Also, to use helm v3 later, we need to add the repositories we need. These steps can be skipped if they have been performed in the client environment:

$ echo $OIDC_ISSUER_URL
https://accounts.google.com
$ echo $OIDC_JWKS_URI
https://www.googleapis.com/oauth2/v3/certs
$ helm repo add jetstack https://charts.jetstack.io
$ helm repo add istio https://istio-release.storage.googleapis.com/charts
$ helm repo add oauth2-proxy https://oauth2-proxy.github.io/manifests
$ helm repo update

Now, with Helm ready, we can install Cert Manager, Istio CRD, Control Plane, Gateways as well as OAuth2-Proxy using Helm. We run each of the following commands from the envoy directory, where the required files are stored:

$ helm install cert-manager jetstack/cert-manager \
 --namespace cert-manager \
 --create-namespace \
 --version v1.7.1 \
 --set installCRDs=true
$ helm install -n istio-system istio-base istio/base --create-namespace
$ helm -n istio-system install istiod istio/istiod -f istiod-values.yaml --wait
$ helm -n istio-system install istio-ingress istio/gateway -f ingress-gateway-values.yaml
$ kubectl -n istio-system get po
$ kubectl create ns oauth2-proxy && kubectl label ns oauth2-proxy istio-injection=enabled
$ helm install -n oauth2-proxy \
 --version 6.0.1 \
 --values oauth2-proxy-values.yaml \
 --set config.clientID=$CLIENT_ID \
 --set config.clientSecret=$CLIENT_SECRET \
 --set config.cookieSecret=$COOKIE_SECRET \
 --set extraArgs.oidc-issuer-url=$OIDC_ISSUER_URL \
 --set extraArgs.whitelist-domain=$WEB_HOST \
 oauth2-proxy oauth2-proxy/oauth2-proxy
$ kubectl -n oauth2-proxy get pods -l app=oauth2-proxy

Note that the istiod installation step uses a value file that includes extensionProviders. This is where we tell Istio to connect to external authorization provider:

meshConfig:
 accessLogFile: /dev/stdout
 extensionProviders:
 - name: oauth2-proxy
 envoyExtAuthzHttp:
 service: oauth2-proxy.oauth2-proxy.svc.cluster.local
 port: 4180
 includeRequestHeadersInCheck:
 - cookie
 headersToUpstreamOnAllow:
 - authorization
 headersToDownstreamOnDeny:
 - set-cookie

When we install istio ingress gateway, we need to expose port 80 as well as 443. Port 80 is used in ACME protocol for certificate configuration. We also installed oauth2-proxy with some configurations from oauth2-proxy-values.yaml as well as some argument set imperatively.

Once configuration is successful, we should be able to confirm that oauth2-proxy service is running. We should also be able to tell the External IP address of the Ingress:

$ kubectl -n oauth2-proxy get svc
$ kubectl -n istio-system get service istio-ingressgateway -o jsonpath='{.status.loadBalancer.ingress[0].ip}'

Now we can go to our DNS configuration portal, to populate the DNS A-record for demo1.digihunch.com with this IP address:

Add DNS A-record

It may take several minutes to a couple of hours, for the A-record update/creation to take effect, depending on the TTL. Sit tight and query DNS with nslookup command, until it is set. In the next step, we will need successful DNS resolution for certificate configuration.

Install workload and configure Certificate

Now we install the workload, along with the necessary constructs such as Gateway CRD, Virtual Service, and certificates. We can use kubectl kustomize command to preview the changes, then apply the changes with -k:

$ kubectl kustomize demo
$ kubectl apply -k demo

When you have your own DNS hostname, modify the kustomization.yaml accordingly before applying. This command uses Kustomization to apply numerous YAML manifests in the demo directory, including the following activities:

Create a Namespace named demo
Create Deployment and Service
Set up PeerAuthentication for the mesh
Configure Ingress class, ClusterIssuer and Certificate using cert-manager
Configure Virtual Service and Gateway

This configures everything we need and we need to verify from several aspects. The demo Service is a ClusterIP service exposed within the cluster at port 80. Then we check the status of certificate:

$ kubectl -n istio-system get certificate
$ kubectl -n istio-system describe certificate demo

The column of READY should have a value of True. Check the logs from cert manager Pods if that’s not the case. Common reasons include:

The automatic ACME validation is just not ready
The automatic ACME validation is still waiting for DNS resolution
The Istio ingress gateway port 80 is not open for ACME validation
The let’s encrypt server applies rate limiting

Note that for the ACME server, we can use staging server’s URL or productions. The former provisions a certificate that may come off as insecure as the CA is not fully trusted by browser. The latter doesn’t have this issue, but the server applies rate limiting. For more details of how validation with ACME work, check out my previous post.

Once the certificate is applied, our website is open to any visitor with HTTPS access:

Then we will use Request Authentication and Authorization Policy to tighten up the access by requiring visiting user to log in with Google identity.

RequestAuthentication and AuthorizationPolicy

We configure a CUSTOM action in the AuthorizationPolicy, and specified the provider by name oauth2-proxy. The YAML manifests for RequestAuthentication and AuthorizationPolicy looks like this:

apiVersion: security.istio.io/v1beta1
kind: RequestAuthentication
metadata:
 name: istio-ingressgateway
 namespace: istio-system
spec:
 jwtRules:
 - issuer: https://accounts.google.com
 jwksUri: https://www.googleapis.com/oauth2/v3/certs
 selector:
 matchLabels:
 app: istio-ingressgateway
---
apiVersion: security.istio.io/v1beta1
kind: AuthorizationPolicy
metadata:
 name: istio-ingressgateway
 namespace: istio-system
spec:
 action: CUSTOM
 provider:
 name: oauth2-proxy
 rules:
 - {}
 selector:
 matchLabels:
 app: istio-ingressgateway

Note that we cannot apply Request Authentication and Authorization Policy before we confirm that the Certificate has been created. This is because when letsencrypt use HTTP resolver to provision certificate with ACME protocol, the request from letsencrypt should get to the validating URI without being asked to provide a Google identity. So do not perform this step yet before confirming the previous step:

$ kubectl apply -f oidc-auth.yaml

Now launch your browser in incognito mode and browse to our URL (https://demo1.digihunch.com) and you will be re-redirected to log on to Google (accounts.google.com). Once logged on, you will have access to the site.

If you run into errors, check the oauth2-proxy log from Pod stdout, which should give a reason of 4xx errors. If oauth2-proxy log indicates no activity, confirm if the request has been forwarded to the proxy. Check the service object of the proxy and make sure it is exposed to the correct port, as indicated in the meshConfig. To check mesh config, examine the configmap named istio in the namespace of istio-system. To check if Istio’s authorization is unable to speak with oatuh2-proxy, inspect the log of istiod Pod.

To clean up the lab, remove the app registration from Google, and then delete the cluster from Azure:

$ az aks delete -g AutomationTest -n orthCluster

In this lab we used a very open AuthorizationPolicy. We can polish it up with more conditions such as:

apiVersion: security.istio.io/v1beta1
kind: AuthorizationPolicy
metadata:
 name: istio-ingressgateway
 namespace: istio-system
spec:
 action: CUSTOM
 provider:
 name: oauth2-proxy
 rules:
 - when:
 - key: request.auth.audiences
 values:
 - $CLIENT_ID
 selector:
 matchLabels:
 app: istio-ingressgateway

Under the rules section, we can explore the claims from the JWT and make creative use of it to achieve our specific goals for IAM.

Bottom line

In this lab, we registered our app with Google and obtained client ID and secret for it to identify our application. We configured oauth2-proxy as our IAM application, carrying the credential validated by Google. We also configured Istio to delegate authorization to oauth2-proxy as external authorization provide, making up an entire OIDC integration.

The third party service (IdP) we integrate with is GCP, which natively supports OIDC protocol. However, if the third party IdP does not support OIDC natively, such as Active Directory via LDAP, then we would need one more proxy to sit between OAuth2 and the external IdP service. One good choice is the Dex connector. Another more complex yet powerful alternative connector is Key Cloak. Consider Dex or KeyCloak as an OIDC proxy, when:

The external IdP does not support OIDC; or
We choose not to use the OIDC capability of the external IdP because we want to managed OIDC centrally in our own proxy
The external identity provider does support OIDC, but the configuration is managed by a different team (e.g. IAM team) other than the platform team

AuthorizationPolicy is the key piece in this integration, and it is executed at the http filter in envoy sidecar proxy. It can be done with alternatives to OAuth2-Proxy such as the authservice project. A slightly different approach to implement OIDC integration is to use leverage extensibility of WebAssembly (a sandboxing technology to extend Envoy), which is suggested in this example for WasmPlugin. However, Wasm is still considered experimental.

Istio Authentication and Authorization

Sat, 05 Feb 2022 23:04:00 -0400

Applications running on Kubernetes platform seeks to offload common non-business features to the platform. Istio helps Kubernetes bridge that gap. It can enforce mTLS communication, which is known as Peer Authentication. It can help with two other things with the use of JWT token: when a web request presents a JWT token, it can validate whether it is authentic. Then, it can use the claims in JWT token to drive authorization decision on whether the specific request is allowed or denied. Both will use Istio CRDs.

Peer Authentication

Istio can enforce mTLS for TCP traffic between Pods. According to its documentation, enforcing mTLS at mesh level is as simple as applying a Peer Authentication resource to the root-level namespace:

apiVersion: security.istio.io/v1beta1
kind: PeerAuthentication
metadata:
 name: default
 namespace: istio-system
spec:
 mtls:
 mode: STRICT

The role of mTLS is so Pods can validates each other’s identity and then encrypt the TLS traffic in between. Each workload must first have an identity and Envoy proxy addressed this issue by adopting SPIFFE framework. It gives each workload an identity in the format of <TRUST_DOMAIN>/ns/<NAMESPACE>/sa/<SERVICE_ACCOUNT>. For example: spiffe://cluster.local/ns/myapp-dev/sa/default. It is also important to understand that only Pods with injected Envoy sidecar have SPIFFE workload identity and therefore is able to speak in mTLS. For new services, this is usually not an issue. For migrating workload without sidecar, a Pod without sidecar may connect with one in the mesh (with sidecar) if the mtls mode is PERMISSIVE in Peer Authentication. Otherwise, the connect is reset at layer 4 with the following error:

curl: (56) Recv failure: Connection reset by peer
command terminated with exit code 56

Therefore, it is advisable to start with PERMISSIVE mode for a precautionary migration of workload to mTLS. With mTLS all effective at the mesh level, there is no need to natively configure TLS between services.

The SPIFFE identity used in PeerAuthentication can also be used in Request Authorization as rule conditions. I will discuss request authentication before request authorization. To understand request authentication, let’s first warm up on JWT.

JSON Web Token

JSON Web Token (JWT, RFC 7519) is a format to carry JSON payload with optional signature and/or encryption. It can be thought of as a document (in JSON format) with signature for web servers to exchange information. The signature portion makes it friendly for document consumers to validate the authenticity. It is also URL-safe, and thereby adopted in web-browser SSO context, to pass identity of an authenticated user between and identity provider and a service provider.

JWT enables token-based authentication, a significant improvement from traditional session-based authentication. The traditional session-based authentication can be illustrated as below:

session-based authentication

This authentication model has major drawbacks. First, a mechanism to validate the authenticity of Cookie is missing. Second, the server has to keep the session information, making itself not stateless, unless a state store such as memcached is introduced.

token-based authentication

In token-based authentication such as using JWT, a token is issued. The authenticity of the token are validated before the server provides data, and it can be validated by any backend server.

The payload of JWT consists of claims, which are statements about an identity (such as name, role, email). There are custom claims as well as standard reserved claims, such as iss (issuer), sub (subject), aud (audience), iat (issued at time), exp (expiration time), and jti (JWT ID). When a program produces a JWT, it turns the raw payload into standardize payload by adding the required reserved claims and may sort the claims alphabetically.

The JWT consists of three parts with a period as delimiter:

The third part is a signature in the format of JWS (JSON Web Signature, RFC 7515) for the JWT consumer to validate its authenticity. The first and second parts, as you can tell, are the claims in the document. Their base64 encoding can be decoded with no effort and should therefore be considered exposed. Although JWT addresses the authenticity of information, it does not intend to address the confidentiality of the payload at HTTP layer. The payload should not carry sensitive information and should always be used with secure HTTPS port. To tackle this issue, there is JWE (JSON Web Encryption, RFC 7516) which is an implementation similar to JWT which also encrypts the payload.

Some IAM protocols are built on top of JWT. For example, the OpenID Connect specification also defines a set of standard claims that it uses while still allow custom claims.

Request Authentication

Istio can perform request authentication using its CRD. It is important to distinguish request authentication and user authentication. In user authentication, the identify provider typically looks up an identity store and compares password hash results to check whether the identity of the visiting user is authentic or not. This is outside of Istio’s capability but many off-the-shelf solution excels at it, such as Azure AD. Once the user’s identity is validated by identity provider, and a JWT is issued for downstream service providers to consume. Istio’s CRD can front the service provider and validate that the presented JWT is authentic. It authenticates the identity of a request (as truly issued by the trusted issuer without being tampered). This process does not involve checking user’s identity, even though user’s identity could be stored in the payload by the JWT issuer.

Istio uses the RequestAuthentication CRD to perform this function. The JWT issuer signs with its private key and stores the signature in the JWT. When it is presented to Istio, Istio’s RequestAuthentication CRD needs the public key of the issuer in order to validate the JWT. The public key usually comes in as a JWK (JSON Web Key, RFC7517), a format convertible to and from PEM format. The JWK can be provided either inline in the RequestAuthentication’s YAML manifest, or via a URI.

Below is an example of a basic RequestAuthentication declaration:

apiVersion: security.istio.io/v1beta1
kind: RequestAuthentication
metadata:
 name: httpbin
 namespace: foo
spec:
 selector:
 matchLabels:
 app: httpbin
 jwtRules:
 - issuer: "issuer-foo"
 jwksUri: https://example.com/.well-known/jwks.json

In this example (from the documentation), the jwtRule requires that the issuer be issuer-foo, and the JWK (containing public key) is provided by a given URI address. Istio will pass the authentication once the signature in the presented JWT is verified with the JWK.

Authorization Policy

Istio’s Authorization Policy by itself can operate at both TCP or HTTP layers and is enforced at the envoy proxy. The result is an ALLOW or DENY decision, based on a set of conditions at both levels. If the traffic is HTTP then you should consider use some HTTP level information as it provides a lot more flexibility. Even when operating at HTTP layer, AuthorizationPolicy does not have to work in conjunction with RequestAuthentication. The rules can use path, methods, etc to drive an authorization decision, for example:

apiVersion: security.istio.io/v1beta1
kind: AuthorizationPolicy
metadata:
 name: authz-policy-orthanc
spec:
 selector:
 matchLabels:
 app: orth
 action: ALLOW
 rules:
 - from:
 - source:
 namespaces: ["istio-system","orthweb"]
 to:
 - operation:
 methods: ["GET","POST","PUT","HEAD","DELETE"]
 ports: ["8042"]
 - from:
 - source:
 namespaces: ["istio-system","orthweb"]
 to:
 - operation:
 ports: ["4242"]

The claims in the JWT payload can also be used to drive authorization decision, as exemplified in the Istio documentation, by using a when keyword in a rule and specifying the claim as a key:

 when:
 - key: request.auth.claims[iss]
 values: ["https://accounts.google.com"]

The when clause requires that the iss claim in the JWT must carry a specific value in order to ALLOW the HTTP request. While the claims in JWT is just an additional factor to drive authorization decision, using authenticated information to drive authorization decision makes the overall workflow more secure, and should therefore be used when applicable.

When using AuthorizationPolicy CRD, keep in mind:

Use correct selectors so it only applies to workloads labelled as such. Without selector the policy takes effect in the entire namespace which is a recipe for issues
When multiple policies (each with multiple rules) are applied to the same workload, be aware of the policy precedence. Troubleshooting may get tricky.
For an AuthorizationPolicy to use HTTP fields in rules, it first needs to identify the traffic as HTTP. To tell Authorization Policy to treat the traffic as HTTP, we need to understand Protocol Selection. In a nutshell, we can name the port as http or http-* in the Service manifest of the workload. Otherwise all HTTP-based rules will be missed.

For troubleshooting, we can check authorization policies effective on a Pod with:

$ istioctl x authz check orthanc-6c9679d8c7-2ttlf -n dev-orthweb

This returns the effective policies but does not necessarily indicate which rule is matched when a request is denied or allowed. To find out further information, you will need to follow Istio FAQ to set RBAC logging to debug, and then monitor the log in the istio-proxy sidecar.

Apart from HTTP fields, path, authenticated claims in JWT, Istio Authorization can also integrate with an Open Policy Agent (OPA) to drive actions, in advanced use cases.

Bottom line

While Istio itself does not perform user authentication, its support of JWT in RequestAuthentication allows a workload to integrate with external identity provider. This capability, along with creative use of claims in JWT, also empowers authorization capability.

Traffic Segmentation on Kubernetes Platform

Thu, 27 Jan 2022 13:54:00 -0400

When operating Kubernetes as a platform for multiple tenants, one of the concerns is controlling the network traffic. This is sometimes referred to as traffic segmentation. This initiative involves a broad range of technical topics from networking to containerization. By no means I am an expert on each of those topics. I have however developed some best practices in how to break down this challenge and hence bringing the thought into this post.

Tenant Isolation

Kubernetes has the concept of namespace to logically separate resources allocated for each tenant. Each tenant only operates within their given namespaces. The isolation of computing resources such as CPU and memory can be managed via ResourceQuota objects, and they are enforced at the kernel level, leaving networking isolation the main discussion in the topic of tenant isolation. If the platform hosts a lot of stateful workload then we also needs to address tenant isolation at the storage layer. In this post we focus on the network aspect of resource isolation, aka traffic segmentation.

Controlling network traffic can require a significant amount of efforts depending on the goal. That is why we need to first assess the multi-tenancy models:

Soft multi-tenancy: usually a platform is shared by multiple teams within the same organization. Tenants are incentivized to be good neighbours.
Hard multi-tenancy: usually a platform shared by multiple customers from different organizations. There is no trust between different tenants, or between tenant and platform operator.

Reality may sits somewhere in between, but we often have to come back to this model when making a technical decision, because it determines the degree of tenant isolation, or the amount of effort we are willing to put in on tenant isolation. At the tough end, is zero-trust network, which usually have the following requirement:

Requirement 1: All network connections are subject to enforcement (not just those that cross zone boundaries).
Requirement 2: Establishing the identity of a remote endpoint is always based on multiple criteria including strong cryptographic proofs of identity. In particular, network-level identifiers like IP address and port are not sufficient on their own as they can be spoofed by a hostile network.
Requirement 3: All expected and allowed network flows are explicitly allowed. Any connection not explicitly allowed is denied.
Requirement 4: Compromised workloads must not be able to circumvent policy enforcement.
Requirement 5: Many Zero Trust Networks also rely on encryption of network traffic to prevent disclosure of sensitive data to hostile entities snooping network traffic. This is not an absolute requirement if private data are not exchanged over the network, but to fit the criteria of a Zero Trust Network, encryption must be used on every network connection if it is required at all. A Zero Trust Network does not distinguish between trusted and untrusted network links or paths. Also note that even when not using encryption for data privacy, cryptographic proofs of authenticity are still used to establish identity.

As you can see there’s a lot of efforts involved in building a zero-trust network. The cost of building a zero-trust network is worth it only when we determines that the overall business requirement demands it.

Pod Networking

It is important to understand Pod networking before developing a traffic segmentation strategy. Pod networking has to do with the CNI driver used for the cluster. There are in general two categories:

Overlay network: Pods are placed on a VXLAN configuration. This is mostly seen in basic Kubenet mode or CNI drives such as Flannel. NAT is required for Pods to communicate across nodes, which might introduce performance issues when deployed at scale. Pods do not use IP address from the host network.
Regular network: In this mode Pods are on the same network as the nodes are. For example, Azure CNI assigns Pods with IP address from a given V-Net. The AWS-VPC CNI integrates VPC networking with Pods. Since Pods are on a corporate network, the traffic control must also consider measures at the whole network level.

The main benefit of the first approach, is that IP exhaustion is less likely due to the introduction of a VxLAN. The other benefit from a networking perspective is that the Pod networking is born separated from the corporate network. In the second approach, by assigning Pods with a corporate IP address (which brings the risk of IP exhaustion), Pods are also potentially exposed to all corporate traffic at layer 3. To tackle this additional risk, network security group should be used in the V-Net for Azure AKS, or security groups for Pods should be considered with AWS EKS. Although we will discuss Network Policy in the rest of this essay, Network Policy mostly addresses the traffic segmentation issue within a Kubernetes cluster. A Pod placed on the corporate network needs traffic segmentation strategies from the perspective of the whole network.

Another network-level traffic segmentation strategy is on the corporate firewall. For example, with AKS you can specify outbound type as user-defined routes (UDR) to direct all outbound traffic through a corporate firewall where traffic will be inspected. There are firewall products dedicated for managing highly dynamic pod traffic from Kubernetes. This strategy can be used in conjunction with network security groups.

Network Policy

Kubernetes’s default behaviour is to allow traffic between any two pods in the cluster network. This is undesirable. NetworkPolicy is the native Kubernetes construct for platform operators and application developer to control network traffic at layer 3/4. It uses namespace and pod selectors, and is defined based on allow rules, which is good for general use.

Further to the native Network Policy, you can adopt third party policies for advanced features. For example, Azure has Azure Network policy (works for Azure CNI only) and Calico Network policy (works for Calico CNI, Azure CNI or Kubenet). The third party network policies usually provides advanced features such as:

Deny rules
multiple types of endpoints in addition to Pods, for example, VMs, network interfaces which can be useful in network-level traffic control
ordering and priority of rules
Flexible matching rules

Calico network has a page that summarizes its features and how it extends the Kubernetes NetworkPolicy. Below is an example of a Calico’s network policy:

apiVersion: projectcalico.org/v3 
kind: NetworkPolicy 
metadata: 
 name: allow-tcp-6379 
 namespace: production 
spec: 
 selector: color == 'red'
 ingress: 
 - action: Allow 
 protocol: TCP 
 source: 
 selector: color == 'blue' 
 namespaceSelector: shape == 'circle'
 destination: 
 ports: 
 - 6379

It is as self-explanatory as Kubernetes Network Policy. No matter which kind of network policy, this approach takes effect at layer 3/4. The rules are eventually implemented in the kernel on the node (Iptables). The management of this layer is usually by the platform team and they need to have some application knowledge.

Authorization at Application Layer

Traffic above layer 4 is considered application layer traffic. At application layer, the decision to allow or deny a request is by definition an authorization decision. Another layer of protection can be placed at layer 4 is mTLS which ensures that each request to have an identity. The authorization can be built in the application, but it is also very common to offload these functions to the service mesh layer. For example, Istio has constructs such as PeerAuthentication, Request Authentication and Authorization Policy. We will those in more details in a few coming blog posts. Below is a simple example of Istio’s Authorization Policy:

apiVersion: security.istio.io/v1beta1 
kind: AuthorizationPolicy 
metadata: 
 name: "details-viewer" 
 namespace: default 
spec: 
 selector: 
 matchLabels: 
 app: details 
 action: ALLOW 
 rules: 
 - from: 
 - source: 
 principals: ["cluster.local/ns/default/sa/bookinfo-productpage"]
 to: 
 - operation: 
 methods: ["GET"]

The rule is also fairly self-explanatory. Compared to Network Policy, the point of enforcement of these Authorization policies are at the envoy proxy. The management of policies at this layer can be debatable if department boundaries are not clear, but it should in general be owned by personnels with good application knowledge.

Consistency between Policies

In-cluster traffic can be controlled with both Network Policy (Calico or Kubernetes) operating at layer 3-4, and Authorization Policy (Istio) at layer 4-7. This brings another challenge of maintaining consistency between the two types of policies. This is especially challenging when they are managed by different teams in a corporate and therefore many operators for soft multi-tenant platform choose not to implement Network Policy or only implements a baseline.

Some network solution providers builds a solution for this. For example, Calico has the capability to enforce network policy for Istio. This integration requires some configuration, but the enhanced GlobalNetworkPolicy supports HTTP methods, eliminating the need to define a separate Authorization Policy in Istio and worry about its consistency with NetworkPolicy. The platform build however, still needs to determine who owns this policy construct. Below is an example from Calico documentation:

apiVersion: projectcalico.org/v3
kind: GlobalNetworkPolicy
metadata:
 name: customer
spec:
 selector: app == 'customer'
 ingress:
 - action: Allow
 http:
 methods: ["GET"]
 egress:
 - action: Allow

One of the benefits of using this integration is a unified policy language based on GlobalNetworkPolicy CRD. In the mean time, organization should also develop strategy to ensure that, once Calico is integrated with Istio, then there is no need to separately build authorization policies, which may come in conflict with Global network policy.

Summary

Controlling network traffic is difficult on Kubernetes platform. In this article I proposed a few angles to approach this issue for enterprise clients.

Kubernetes Admission Control

Fri, 07 Jan 2022 22:21:00 -0400

This post discusses admission control, and its implementation – the OPA Gatekeeper. I also discuss Azure Policy as a different Gatekeeper implementation.

Admission Webhooks

Admission controller intercepts requests to the Kubernetes API server after the request has been authenticated and authorized, and prior to persistence of the object into etcd store. There are many compiled-in controllers, which can be turned on and off on the node with the arguments of kube-apiserver process. For example, the ImagePolicyWebhook can be enabled with value ImagePolicyWebhook added to the –enable-admission-plugins switch. Its configuration can be provided via the –admission-control-config-file switch.

In addition to the compiled-in admission plugins (which must be configured for kube-apiserver process on the node), admission plugins can be developed as extensions and run as webhooks configured at runtime. This allows users to configure webhooks via API access, dynamically without having to restart kube-apiserver process on the Node, which is usually hard to do with managed Kubernetes platforms. They are therefore called Dynamic Admission Control.

You can define two types of admission webhooks in dynamic admission control: validating admission webhook, and mutating admission webhook. Their interaction with API server can be illustrated in the diagram below:

The mutating admission hook takes action to change the API request, whereas the validating admission hook accepts or denies the request.

A good example of mutating webhook is Istio’s sidecar injector. We can view the configuration with this command:

$ kubectl get MutatingWebhookConfiguration istio-sidecar-injector -o yaml | less

From the manifest returned, we can see that in this configuration the request is forwarded to istiod service on port 443, at path /inject for processing. We can also see some matching rules to find the target Pod creation API request.

Validating webhook can be display with the following call:

$ kubectl get ValidatingWebhookConfiguration

The output of validating webhook is a yes or no decision. We usually use validating webhook in conjunction with a policy engine to decide whether the request should be accepted or denied.

Open Policy Agent

Open Policy Agent (OPA) is an open-source general-purpose policy engine that applies policies written in Rego language to ingested JSON document and returns a result. It is usually integrated with system which requires a policy engine. For example, Kyverno is a policy engine designed for Kubernetes. Styra (one of the OPA contributors) develops policy engines to integrate with Istio’s authorization policy. They have online courses on OPA policy authoring and microservice authorization with their product.

OPA is build to be a general-purpose, unified way of solving policy and authorization problem. With microservice authorization, the activities includes decision making (determine action based on input, aka Policy Decision Point, PDP), and decision enforcement (issue 400 code or 200 code depending on decision, aka Policy Enforcement Point, PEP). OPA is introduced to decouple these two activities. OPA’s input is a JSON payload and it uses Policy in Rego language to come to decision.

The team that developers Open Policy Agent also created their controller (with OPA as the core component) to run validating web hook and mutating web hook. The original version is OPA-Kubernetes that uses kube-mgmt. This original version is also dubbed Gatekeeper v1.0. When OPA starts, the kube-mgmt sidecar container will load Kubernetes Namespace and Ingress objects into OPA. You can configure the sidecar to load any kind of Kubernetes object into OPA. The sidecar establishes watches on the Kubernetes API server so that OPA has access to an eventually consistent cache of Kubernetes objects. It has gone through a couple of major version changes as summarized in this section. As of today, when we deploy Gatekeeper we should use version 3.

Gatekeeper v3

Currently, Gatekeeper v3 is the most popular choice for Kubernetes Policy Controller. The diagram bellow illustrate how Gatekeeper integrates with Kubernetes API server.

Gatekeeper and Kubernetes

We can follow this guide to install Gatekeeper but the key step is as simple as to apply the correct version of manifest. Alternatively it can be installed using Helm. After the installation, we should see a Service named gatekeeper-webhook-service in the gatekeeper-system namespace. We can also inspect the newly created validationg web hook configuration

k get validatingwebhookconfiguration gatekeeper-validating-webhook-configuration -o yaml | less

The result indicates that the configuration forwards incoming manifests to the gatekeeper-webhook-service web service at the path /v1/admin for validation, and then at /v1/admitlabel for labelling. The configuration also stores rules as matching criteria.

We can smoke test Gatekeeper v3, with the basic example in its directory. Apply the template, constraint and then the manifests in resources. The pod creation will fail with an error like:

Error from server ([pod-must-have-gk] you must provide labels: {"gatekeeper"}): error when creating "resources/bad_pod_namespaceselector.yaml": admission webhook "validation.gatekeeper.sh" denied the request: [pod-must-have-gk] you must provide labels: {"gatekeeper"}

The gatekeeper document also covers the details of using ConstraintTemplate and Constraints. However, Writing your own a policy in Rego still takes time and we want to piggyback on the community for commonly used policies. OPA‘s gatekeeper-library projects keeps a handful of those in its library directory. We can test the privileged container example:

$ cd gatekeeper-library/library/pod-security-policy/privileged-containers
$ kustomize build . | kubectl apply -f -
constrainttemplate.templates.gatekeeper.sh/k8spspprivilegedcontainer created
$ kubectl apply -f samples/psp-privileged-container/example_disallowed.yaml
pod/nginx-privileged-disallowed created
$ kubectl delete -f samples/psp-privileged-container/example_disallowed.yaml
pod "nginx-privileged-disallowed" deleted
$ kubectl apply -f samples/psp-privileged-container/constraint.yaml
k8spspprivilegedcontainer.constraints.gatekeeper.sh/psp-privileged-container created
$ kubectl apply -f samples/psp-privileged-container/example_disallowed.yaml
Error from server ([psp-privileged-container] Privileged container is not allowed: nginx, securityContext: {"privileged": true}): error when creating "samples/psp-privileged-container/example_disallowed.yaml": admission webhook "validation.gatekeeper.sh" denied the request: [psp-privileged-container] Privileged container is not allowed: nginx, securityContext: {"privileged": true}

Currently the library directory contains two sub-directories, general and pod-scurity-policy. The latter is to regulate Pod creation, while the former includes more common usecases such as disable node port, enforce https, and enforce probes. This is the place I start with when building a policy.

The policy constraints take effect cluster wide. When we have multiple clusters, we would like a unified place to manage policies.

Azure Policy with AKS

We take Azure Policy with AKS as an example to illustrate how public cloud platform can simplify policy management. When building AKS cluster, an addon profile for Azure Policy can be installed. This allows Azure Policy to connect to the AKS cluster. Azure Policy contains many built-in policies definitions (as well as initiative definitions which are groups of related policies). We can simply search by Kubernetes keyword and look for the built-in policies. For example, there is a built-in policy definition “Kubernetes clusters should not allow container privilege escalation. The definitions (policy or initiative) can be assigned to a resource group with enforcement action set to denied, and with excluded namespaces, as shown in the screenshot below

The assignment can take as long as 10 minutes to push down to the cluster. Then we should be able to confirm by checking the constraint CRDs. We can see this This setup brings a centralized policy management system that can be easily hooked up to multiple clusters.

Other benefits of this architecture includes the ability to report compliance. As per CIS report for Azure AKS recommendation 4.3:

Azure Policy extends Gatekeeper v3, an admission controller webhook for Open Policy Agent (OPA), to apply at-scale enforcements and safeguards on your clusters in a centralized, consistent manner. It covers many basic resource types but does not cover any well-known CRDs. Azure Policy makes it possible to manage and report on the compliance state of your Kubernetes clusters from one place.

Checks with Azure Policy service for policy assignments to the cluster.
Deploys policy definitions into the cluster as constraint template and constraint custom resources.
Reports auditing and compliance details back to Azure Policy service.

As of February 2022, AWS EKS doesn’t seem to have the equivalent of this capability to integrate with a policy management. The only option would be to install Gatekeeper v3 yourself on the cluster, or host it separately.

Bottom line

Admission control should be a standard setup in Kubernetes deployment. When building gatekeeper system on your own, it can be set up separately on a different cluster. When Kubernetes is provided as a platform, it is very helpful for platform operator to manage their tenants. If the tenant is application development team, it also makes sense for them to develop their own policies for the developers in their team.