serverless-workflow-github-showcase

DEPRECATION NOTICE

This example has a few old concepts from architecture and integration perspective. We are working on an updated version of this example. Until there, please try our other examples in this directory. In case you still need to run it, latest Kogito version which worked was KOGITO 1.29.0.Final.

Serverless Workflow GitHub Showcase

In this example we will deploy a GitHub "bot" application that will react upon a new PR being opened in a given GitHub project. The "bot" is implemented via service and event orchestration approach with Kogito implementation of the Serverless Workflow specification.

The image below illustrates an overview of the architecture:

The implementation relies on three services included in this repository:

1. Pull Request Checker Workflow: it will coordinate the changes in the PR opened in a given GitHub repository. Depending on the files changed in the project, different labels and reviewers will be automatically added in the PR. In the end of the workflow, the service will broadcast a "PR Verified" event to the cluster.
2. GitHub API Wrapper Service: calling the GitHub API as a GitHub Application requires a token to be generated on a small timeframe. This service generates a valid token each 5 minutes to make calls to the API. Besides token management, it wraps the API and simplifies its interface just for the sake of this example.
3. Notification Service: a simple Camel service to interact with the Slack API to notify a given channel.

Prerequisites

IMPORTANT! Before proceeding, please make sure you have everything listed in this section ready.

You may use CRC or Minikube if you don't have a cluster available with cluster admin rights. Or you can ask an administrator to install the prerequisites for you.

To deploy this example in your Kubernetes/OpenShift cluster, you will need:

1. A Quay.io account
2. A Kubernetes/OpenShift namespace to deploy the example: kubectl create ns kogito-github or `oc new-project kogito-github
3. Istio installed because it's required by Knative platform. You can follow the Knative documentation for a very basic and simple installation.
4. Knative Serving and Eventing components installed. We recommend installing the Knative Operator and install the rest of the components through it as described in their documentation.
5. Kogito Operator installed in the namespace kogito-github. Download the latest release, and run: NAMESPACE=kogito-github ./hack/install.sh. Alternatively, you can also install it via OperatorHub.

In your local machine you will need:

1. To clone this repository and go to serverless-workflow-github-showcase directory (git clone https://github.com/apache/incubator-kie-kogito-examples.git && cd serverless-workflow-github-showcase)
2. Java 17 SDK
3. Maven 3.8.1+
4. Podman or Docker to build the images
5. kubectl or oc client

Deploying the examples

Follow the instructions for each service to try them locally as standalone services and deploy them in your Kubernetes or OpenShift cluster:

1. GitHub Service
2. Notification Service
3. PR Checker SW Service

Knative and Kogito will bind them together. ❤️

In case of any problems, please file an issue or reach out to us at the Kogito Dev Mailing list.

Cleaning up

You can easily clean up the demo by deleting the namespace:

kubectl delete ns kogito-github

To clean up your /etc/hosts file, run the following script:

./scripts/cleanup-hosts-file.sh

It will remove all the lines added by the expose-on-minikube.sh script while deploying the services.