storytold/terraform-aws-observability-accelerator

Fork 0

mirror of https://github.com/storytold/terraform-aws-observability-accelerator.git synced 2026-10-09 00:09:43 +00:00

Files

T

Ela AWS c5e4c0c718 Fixing commadn issue (#177 )

2023-06-07 15:25:02 +02:00

10 KiB

Raw Blame History

Amazon EKS cluster metrics

This example demonstrates how to monitor your Amazon Elastic Kubernetes Service (Amazon EKS) cluster with the Observability Accelerator's EKS monitoring module.

Monitoring Amazon Elastic Kubernetes Service (Amazon EKS) for metrics has two categories: the control plane and the Amazon EKS nodes (with Kubernetes objects). The Amazon EKS control plane consists of control plane nodes that run the Kubernetes software, such as etcd and the Kubernetes API server. To read more on the components of an Amazon EKS cluster, please read the service documentation.

The Amazon EKS infrastructure Terraform modules focuses on metrics collection to Amazon Managed Service for Prometheus using the AWS Distro for OpenTelemetry Operator for Amazon EKS. It deploys the node exporter and kube-state-metrics in your cluster.

It provides default dashboards to get a comprehensible visibility on your nodes, namespaces, pods, and Kubelet operations health. Finally, you get curated Prometheus recording rules and alerts to operate your cluster.

Additionally, you can optionally collect custom Prometheus metrics from your applications running on your EKS cluster.

Prerequisites

!!! note Make sure to complete the prerequisites section before proceeding.

Setup

1. Download sources and initialize Terraform

git clone https://github.com/aws-observability/terraform-aws-observability-accelerator.git
cd examples/existing-cluster-with-base-and-infra
terraform init

2. AWS Region

Specify the AWS Region where the resources will be deployed:

export TF_VAR_aws_region=xxx

3. Amazon EKS Cluster

To run this example, you need to provide your EKS cluster name. If you don't have a cluster ready, visit this example first to create a new one.

Specify your cluster name:

export TF_VAR_eks_cluster_id=xxx

4. Amazon Managed Service for Prometheus workspace (optional)

By default, we create an Amazon Managed Service for Prometheus workspace for you. However, if you have an existing workspace you want to reuse, edit and run:

export TF_VAR_managed_prometheus_workspace_id=ws-xxx

To create a workspace outside of Terraform's state, simply run:

aws amp create-workspace --alias observability-accelerator --query '.workspaceId' --output text

5. Amazon Managed Grafana workspace

To visualize metrics collected, you need an Amazon Managed Grafana workspace. If you have an existing workspace, create an environment variable as described below. To create a new workspace, visit our supporting example for Grafana

!!! note For the URL https://g-xyz.grafana-workspace.eu-central-1.amazonaws.com, the workspace ID would be g-xyz

export TF_VAR_managed_grafana_workspace_id=g-xxx

6. Grafana API Key

Amazon Managed Grafana provides a control plane API for generating Grafana API keys. As a security best practice, we will provide to Terraform a short lived API key to run the apply or destroy command.

Ensure you have necessary IAM permissions (CreateWorkspaceApiKey, DeleteWorkspaceApiKey)

!!! note Starting version v2.5.x and above, we use Grafana Operator and External Secrets to manage Grafana contents. Your API Key will be stored securely on AWS Secrets Manager and the Grafana Operator will use it to sync dashboards, folders and data sources. Read more here.

export TF_VAR_grafana_api_key=`aws grafana create-workspace-api-key --key-name "observability-accelerator-$(date +%s)" --key-role ADMIN --seconds-to-live 7200 --workspace-id $TF_VAR_managed_grafana_workspace_id --query key --output text`

Deploy

Simply run this command to deploy the example

terraform apply

Visualization

1. Prometheus data source on Grafana

Make sure to open the link in the output. After a successful deployment, this will open the Prometheus data source configuration on Grafana. Click Save & test and you should see a notification confirming that the Amazon Managed Service for Prometheus workspace is ready to be used on Grafana.

terraform output grafana_prometheus_datasource_test

2. Grafana dashboards

Go to the Dashboards panel of your Grafana workspace. You should see a list of dashboards under the Observability Accelerator Dashboards

Open a specific dashboard and you should be able to view its visualization

3. Amazon Managed Service for Prometheus rules and alerts

Open the Amazon Managed Service for Prometheus console and view the details of your workspace. Under the Rules management tab, you should find new rules deployed.

!!! note To setup your alert receiver, with Amazon SNS, follow this documentation

Custom Prometheus metrics collection

In addition to the cluster metrics, if you are interested in collecting Prometheus metrics from your pods, you can use setup custom metrics collection. This will instruct the ADOT collector to scrape your applications metrics based on the configuration you provide. You can also exclude some of the metrics and save costs.

Using the example, you can edit examples/existing-cluster-with-base-and-infra/main.tf. In the module module "workloads_infra" { add the following config (make sure the values matches your use case):

enable_custom_metrics = true

custom_metrics_config = {
    # list of applications ports (example)
    ports = [8000, 8080]

    # list of series prefixes you want to discard from ingestion
    dropped_series_prefix = ["go_gcc"]
}

After applying Terraform, on Grafana, you can query Prometheus for your application metrics, create alerts and build on your own dashboards. On the explorer section of Grafana, the following query will give you the containers exposing metrics that matched the custom metrics collection, grouped by cluster and node.

sum(up{job="custom-metrics"}) by (container_name, cluster, nodename)

Troubleshooting

1. Grafana dashboards missing or Grafana API key expired

In case you don't see the grafana dashboards in your Amazon Managed Grafana console, check on the logs on your grafana operator pod using the below command :

kubectl get pods -n grafana-operator

Output:

NAME                                READY   STATUS    RESTARTS   AGE
grafana-operator-866d4446bb-nqq5c   1/1     Running   0          3h17m

kubectl logs grafana-operator-866d4446bb-nqq5c -n grafana-operator