Readme.md

Application Insights Kubernetes Example (Worker)

This is an example of how to use ApplicationInsights.Kubernetes inside a worker. The project uses .NET 6, it should work in .NET Core 3.1 with minimal tweaks.

Goal

This is what it looks like when ApplicationInsights.Kubernetes is turned on for your worker application:

And in this example, we will walk you through the key steps.

Prerequisite

To finish the walk-through, there are some prerequisites.

• .NET 6 SDK. Refer to https://dot.net for more details.
  • This should work in .NET Core 3.1 with some simple tweaks.
• Ability to build a docker image for sanity check from a dockerfile.
  • Pick your way to build the docker image.
  • We used DockerDesktop in this example.
• A kubernetes cluster to deploy the image to.
  • We also used the one provided by DockerDesktop but it should work in any Kubernetes cluster.
• A container registry works with the Kubernetes cluster for deployment.
  • We used docker hub in this example.
• An application insights resource for testing the result.
• A clone/fork this repo.
  • Starting by navigate to example folder of: examples/WorkerExample.

How is the application built

• Refer to WorkerExample.csproj for NuGet packages needed. Here's a copy:

  <PackageReference Include="Microsoft.ApplicationInsights.Kubernetes" Version="2.*" />
  <PackageReference Include="Microsoft.ApplicationInsights.WorkerService" Version="2.20.0" />
  <PackageReference Include="Microsoft.Extensions.DependencyInjection" Version="6.0.0" />
  <PackageReference Include="Microsoft.Extensions.Hosting" Version="6.0.1" />

  Notes: these are current packages by the time of this example. As a common practice, it is recommended you always checking for latest stable ones to use.

• Add a Worker class. Notes:

  • It inherits from BackgroundService.
  • It injects TelemetryClient.
  • It sends one request operation and one custom event to application insights every 5 seconds.

• Refer to Program.cs for setup the worker as a generic host. Here are the critical services to register:

  services.AddApplicationInsightsTelemetryWorkerService();    // Enable Application Insights for workers.
  services.AddApplicationInsightsKubernetesEnricher();        // Enable Application Insights Kubernetes to enhance telemetries.
  services.AddHostedService<Worker>();                        // Register the background service of Worker.

• Making sure appsettings.json will be copied to output folder on build, pay attention to the build task in the project file like this:

  <ItemGroup>
      <!-- Build action is `None`, and the option for copy to output directory could be `Always` or `PreserveNewest` -->
      <None Include="./appsettings.json" CopyToOutputDirectory="PreserveNewest" />
  </ItemGroup>

  • Remember to update the connection string in appsettings.json when you try it on your own. You can get the connection string on the Overview blade of your application insights resource. Refer Connection Strings for more details.
  • Using of instrumentation key instead of connection string should work too. You will need to setup InstrumentationKey than ConnectionString in appsettings.json.

Build docker image

• Build the image

  docker build -t workerapp .

• Optionally run it locally

  docker run -d --name myapp workerapp
  docker logs myapp

  You will see logs like this:

  info: Microsoft.Hosting.Lifetime[0]
  Application started. Press Ctrl+C to shut down.
  info: Microsoft.Hosting.Lifetime[0]
      Hosting environment: Production
  info: Microsoft.Hosting.Lifetime[0]
      Content root path: /app

• Optionally, check out the Live Metrics in your application insights resource, you shall see request / custom events.

  • This is primary used to verify that the application is properly instrumented with Application Insights for worker.

  • Since the application is not yet hosted in a Kubernetes cluster, there's no Kubernetes related property yet.

    

• Delete the container when the test is done

  docker rm myapp -f

Push the image when it is ready

Notes: we are using dockerhub here as an example. you could use any container registry that deploys to your Kubernetes cluster.

• Tag the image, bump up the version accordingly, for example:

  docker tag workerapp dockerhub_account_name/ai-k8s-worker-example:1.0.0

• Push the image:

  docker push dockerhub_account_name/ai-k8s-worker-example:1.0.0

Deploy the container to K8s cluster

• You could use default namespace, but it is recommended to put the test application in a dedicated namespace, for example 'ai-k8s-demo'. To deploy a namespace, use content in k8s-namespace.yaml:

  kubectl create -f k8s-namespace.yaml
  

• Build a deployment yaml file, refer to k8s.yml.

  • Update the namespace and the image to pointing to the one that you intend to use.

• Setup proper role binding for RBAC enabled clusters

  If you have RBAC enabled for your cluster, permissions need to be granted to the service account to access the cluster info for telemetries. Refer to Configure RBAC permissions for details.

  For example, deploy a role assignment in namespace of ai-k8s-demo by calling:

  kubectl create -f ..\..\docs\sa-role.yaml

  The output looks like this:

  clusterrole.rbac.authorization.k8s.io/appinsights-k8s-property-reader created
  clusterrolebinding.rbac.authorization.k8s.io/appinsights-k8s-property-reader-binding created

• Deploy it:

  Making sure the image is up to date, including the version in k8s.yml, and then:

  kubectl create -f .\k8s.yml

  Now, check the live metrics again, you shall see enhanced telemetries:

  

• Clean up the deployment:

  kubectl delete -f .\k8s.yml

If you have ideas, feature requests or run into problems, feel free to file an issue.

References

• Application Insights for Worker
• .NET Generic Host
• Dockerize an ASP.NET Core application
• Create An application insights resource
• Use Connection Strings in Application Insights