ClusterInterceptors

Tekton Triggers ships with the ClusterInterceptor Custom Resource Definition (CRD), which allows you to implement a custom cluster-scoped Webhook-style Interceptor.

A ClusterInterceptor specifies an external Kubernetes v1 Service running custom business logic that receives the event payload from the EventListener via an HTTP request and returns a processed version of the payload along with an HTTP 200 response. The ClusterInterceptor can also halt processing if the event payload does not meet criteria you have configured as well as add extra fields that are accessible in the EventListener's top-level extensions field to other Interceptors and ClusterInterceptors chained with it and the associated TriggerBinding.

Structure of a ClusterInterceptor

A ClusterInterceptor definition consists of the following fields:

  • Required:
    • apiVersion - specifies the target API version, for example triggers.tekton.dev/v1alpha1
    • kind - specifies that this Kubernetes resource is a ClusterInterceptor object
    • metadata - specifies data that uniquely identifies this ClusterInterceptor object, for example a name
    • spec - specifies the configuration information for this ClusterInterceptor object, including:
      • [clientConfig] - specifies how a client, such as an EventListener communicates with this ClusterInterceptor object

Configuring the client of the ClusterInterceptor

The clientConfig field specifies the client, such as an EventListener and how it communicates with the ClusterInterceptor to exchange event payload and other data. You can configure this field in one of the following ways:

  • Specify the url field and as its value a URL at which the corresponding Kubernetes service listens for incoming requests from this ClusterInterceptor
  • Specify the service field and within it reference the corresponding Kubernetes service that’s listening for incoming requests from this ClusterInterceptor

For example:

spec:
  clientConfig:
    url: "http://interceptor-svc.default.svc/"
---
spec:
  clientConfig:
    service:
      name: "my-interceptor-svc"
      namespace: "default"
      path: "/optional-path" # optional
      port: 8081 # defaults to 80

Configuring a Kubernetes Service for the ClusterInterceptor

The Kubernetes object running the custom business logic for your ClusterInterceptor must meet the following criteria:

  • Fronted by a regular Kubernetes v1 Service listening on an HTTP port (default port is 80)
  • Accepts an HTTP POST request that contains an InterceptorRequest as a JSON body
  • Returns an HTTP 200 OK response that contains an InterceptorResponse as a JSON body. If the trigger processing should continue, the interceptor should set the continue field in the response to true. If the processing should be stopped, the interceptor should set the continue field to false and also provide additional information detailing the error in the status field.
  • Returns a response other than HTTP 200 OK only if payload processing halts due to a catastrophic failure.