docs/eventing/samples/k8s-events

Flow Example

A simple function which receives and logs Kubernetes Events. This examples uses all the different components of the Knative Eventing stack with a stub ClusterBus and the k8sevents EventSource. The function is deployed using a Knative serving service, so it scales automatically as event traffic increases.

Prerequisites

• A Kubernetes cluster with Knative installed. Follow the installation instructions if you need to create one.
• Docker installed and running on your local machine, and a Docker Hub account configured (you'll use it for a container registry).
• The core Knative eventing tools installed. You can install them with:

  kubectl apply --filename https://storage.googleapis.com/knative-releases/eventing/latest/release.yaml
  

Configuring Knative

To use this sample, you'll need to install the stub ClusterBus and the k8sevents EventSource.

kubectl apply --filename https://storage.googleapis.com/knative-releases/eventing/latest/release-clusterbus-stub.yaml
kubectl apply --filename https://storage.googleapis.com/knative-releases/eventing/latest/release-source-k8sevents.yaml

Granting permissions

Because the k8sevent EventSource needs to create a Deployment and call the Kubernetes watch API on events, you'll need to provision a special ServiceAccount with the necessary permissions.

The serviceaccount.yaml file provisions a service account, creates a role which can create Deployments in the default namespace and can view all Kubernetes resources. In a production environment, you might want to limit the access of this service account to only specific namespaces.

kubectl apply --filename serviceaccount.yaml

Build and deploy the sample

1. Use Docker to build the sample code into a container. To build and push with Docker Hub, run these commands replacing {username} with your Docker Hub username:

   # Build the container on your local machine
   # Note: The relative path points to the _root_ of the `knative/docs` repo
   docker build -t {username}/k8s-events --file Dockerfile ../../../
   
   # Push the container to docker registry
   docker push {username}/k8s-events
   

2. After the build has completed and the container is pushed to Docker Hub, you can deploy the function into your cluster. Ensure that the container image value in function.yaml matches the container you built in the previous step. Apply the configuration using kubectl:

   kubectl apply --filename function.yaml
   

3. Check that your service is running using:

   kubectl get services.serving.knative.dev --output "custom-columns=NAME:.metadata.name,READY:.status.conditions[2].status,REASON:.status.conditions[2].message"
   NAME              READY     REASON
   read-k8s-events   True      <none>
   

4. Create the flow sending Kubernetes Events to the service:

   kubectl apply --filename flow.yaml
   

5. If you have the full knative install, you can read the function logs using Kibana (see below if you are using a lite or no-mon install). To access Kibana, you need to run kubectl proxy to start a local proxy to access the logging stack:

   kubectl proxy
   Starting to serve on 127.0.0.1:8001
   

   Then visit the Kibana interface and search for the following string (or use the link above):

   kubernetes.container_name:user-container kubernetes.labels.serving_knative_dev\/configuration:read-k8s-events
   

   

   If you are running a lite or no-mon install, you can use kubectl logs to read the logs from the pod:

   kubectl get pods
   # Find a pod with read-k8s-events prefix
   kubectl logs read-k8s-events-....
   

Understanding what happened

When the flow is created, it provisions the following resources:

1. A Feed resource to connect the specified EventSource to a Channel on the bus:

   kubectl get --output yaml feed k8s-event-flow-....
   

   apiVersion: feeds.knative.dev/v1alpha1
   kind: Feed
   metadata:
     name: k8s-event-flow
     namespace: default
   spec:
     action:
       channelName: k8s-event-flow-channel.default.svc.cluster.local
     serviceAccountName: feed-sa
     trigger:
       eventType: dev.knative.k8s.event
       parameters:
         namespace: default
       resource: k8sevents/dev.knative.k8s.event
       service: k8sevents
   

2. The trigger parameter on the Feed references an eventType, and provides some parameters to that EventType:

   kubectl get --output yaml eventtype dev.knative.k8s.event
   

   apiVersion: feeds.knative.dev/v1alpha1
   kind: EventType
   metadata:
     name: dev.knative.k8s.event
     namespace: default
   spec:
     description: subscription for receiving k8s cluster events
     eventSource: k8sevents
   

3. In turn, the EventType references an eventSource. A single EventSource may expose several different EventTypes, each of which provides different types of events from the same source. At the moment, the k8sevents adapter only knows how to listen for corev1.Event changes, but it's likely that other sorts of object watches will be supported in the future.

   kubectl get --output yaml eventsource k8sevents
   

   apiVersion: feeds.knative.dev/v1alpha1
   kind: EventSource
   metadata:
     name: k8sevents
     namespace: default
   spec:
     image: gcr.io/knative-releases/github.com/knative/eventing/pkg/sources/k8sevents@...
     parameters:
       image: gcr.io/knative-releases/github.com/knative/eventing/pkg/sources/k8sevents/receive_adapter@...
     source: k8sevents
     type: k8sevents
   

4. In addition to the eventType parameter, the Flow also referenced a Channel by kubernetes service name (as action.channelName. You can find the channel object by examining the ownerReferences on the Service:

   kubectl get --output yaml svc k8s-event-flow-channel
   

   apiVersion: v1
   kind: Service
   metadata:
     labels:
       channel: k8s-event-flow
     name: k8s-event-flow-channel
     namespace: default
     ownerReferences:
     - apiVersion: channels.knative.dev/v1alpha1
       blockOwnerDeletion: true
       controller: true
       kind: Channel
       name: k8s-event-flow
       uid: 234b1269-8e5d-11e8-aae1-42010a8a001e
   spec:
     ...
   

5. A Channel is the network address of a message dispatch queue on a Bus. The Channel and Bus provide a one-to-many messaging layer with optional persistence. Each Channel is associated with either a Bus or a ClusterBus:

   kubectl get --output yaml channel k8s-event-flow
   

   apiVersion: channels.knative.dev/v1alpha1
   kind: Channel
   metadata:
     name: k8s-event-flow
     namespace: default
   spec:
     clusterBus: stub
   

6. In this case, you can see that the k8s-event-flow Channel references the stub ClusterBus. Knative eventing supports both ClusterBuses, provisioned by a cluster administrator, and namespace-local Buses, which are useful for development or resource isolation. Typically, many users may share a persistent ClusterBus backed by a broker such as Kafka or Google PubSub. In these cases, the cluster administrator would create and manage both the underlying Kafka cluster and the ClusterBus definition. The stub bus is a simple in-memory forwarder with no dependencies, which is great for demos but will not durably store messages if the connected endpoints are down.

   kubectl get --output yaml clusterbus stub
   

   apiVersion: channels.knative.dev/v1alpha1
   kind: ClusterBus
   metadata:
     name: stub
   spec:
     dispatcher:
       args:
       - -logtostderr
       - -stderrthreshold
       - INFO
       image: gcr.io/knative-releases/github.com/knative/eventing/pkg/buses/stub@...
       name: dispatcher
       resources: {}
   

7. So far, you've explored how the events are collected from Kubernetes by the Feed and forwarded to the Bus. The last piece of the puzzle is how the events are forwarded from the Bus to your function. This is handled by a Subscription:

   kubectl get --output yaml subscription k8s-event-flow
   

   - apiVersion: channels.knative.dev/v1alpha1
   kind: Subscription
   metadata:
     name: k8s-event-flow
     namespace: default
   spec:
     channel: k8s-event-flow
     subscriber: read-k8s-events.default.svc.cluster.local
   

   A Subscription links a channel to an interested recipient of the events; in this case, the k8s-event-flow Channel with our read-k8s-events Knative Service.