Using kubernetes apps with apm.k8s.elastic.co

Question

Pulumi · Accepted Answer

In order to use Kubernetes applications with `apm.k8s.elastic.co` (which refers to Application Performance Monitoring by Elastic on a Kubernetes platform), you typically need to deploy an Elastic APM Server and configure your applications to send monitoring data to it.

The APM Server collects performance metrics and errors from your running applications and stores this data in Elasticsearch, allowing you to monitor and analyze the performance of your applications in Kibana.

Here’s how you can achieve this with Pulumi in TypeScript:

1. Deploy an Elastic APM Server in your Kubernetes cluster.
2. Configure your applications to send performance data to the APM Server.

Below is a simplified example of how you can use Pulumi to deploy an Elastic APM Server using an existing Helm chart.

```typescript
import * as pulumi from "@pulumi/pulumi";
import * as kubernetes from "@pulumi/kubernetes";

// Create a Kubernetes cluster (this step assumes you already have a configured cluster,
// therefore it's commented out, but is included for visibility on how the process generally starts)
/*
const cluster = new eks.Cluster("app-cluster", {
    // Cluster configuration here
});
*/

// Create a new instance of the Kubernetes provider linked to the cluster
const k8sProvider = new kubernetes.Provider("k8s-provider", {
    // If you have a kubeconfig file, you can reference it:
    // kubeconfig: kubeconfig_file_content,

// Or if you created a cluster as part of your Pulumi program, you can link directly:
    // kubeconfig: cluster.kubeconfig.apply(JSON.stringify),
});

// Deploy Elastic APM Server using a Helm chart
const elasticApmRelease = new kubernetes.helm.v3.Release("elastic-apm", {
    chart: "apm-server",
    // This Helm chart is sourced from the Elastic Helm repository
    repositoryOpts: {
        repo: "https://helm.elastic.co",
    },
    version: "7.10.0", // Specify the chart version you want to deploy
    // Values can be configured to customize the behavior of the chart
    values: {
        "apmConfig": {
            "apm-server.yml": `
            apm-server:
              host: "0.0.0.0:8200"
            output:
              elasticsearch:
                hosts: ["http://elasticsearch:9200"]
            `,
        },
    },
}, { provider: k8sProvider });

// Export the APM Server URL to access it externally (if configured to be exposed)
export const apmServerUrl = elasticApmRelease.status.loadBalancer.ingress[0].hostname;

// Application configurations would now need to include a reference to this APM server
// so they can send their performance data to it.
// This setup is specific to each application and would be configured in the deployment of the application itself.

```

In this example:

- We start by importing the necessary Pulumi packages for Kubernetes and potentially Elastic Kubernetes Service (EKS), if you're deploying your cluster on AWS.
- This code assumes you already have access to a Kubernetes cluster; if not, you would need to create one using Pulumi. The cluster creation step is present as a commented-out code snippet.
- We create an instance of the Pulumi Kubernetes Provider, linking it to your cluster's `kubeconfig`.
- Next, we use Pulumi’s `kubernetes.helm.v3.Release` resource to deploy the Elastic APM Server using its Helm chart, specifying the chart arguments and configurations.
- After the deployment, we export the APM Server's endpoint so you can configure your applications accordingly.

This program would need to be modified to match the exact details of your environment and application. The values passed to the `values` parameter of the Helm chart would also need to adhere to any specific requirements you have for configuring the Elastic APM Server. The chart version mentioned (`7.10.0`) is purely illustrative; you should specify the version compatible with your requirements.

To configure applications to use the APM Server, you’ll typically adjust the application's deployment manifest to include the APM server details (usually the APM Server URL). The exact changes would depend on the specifics of the application and the APM agent it uses (agents exist for multiple programming languages and frameworks).

Remember that you may also need to adjust firewall and security group settings to allow traffic between your applications and the APM Server and set up persistent storage for the APM Server for production workloads.