Application Real-Time Monitoring Service:Monitor AWS CloudWatch metrics in Managed Service for Prometheus

How it works

YACE runs as a pod in your Container Service for Kubernetes (ACK) cluster and collects CloudWatch metrics through a three-stage pipeline:

1. Metric scraping: YACE periodically pulls metrics such as CPU utilization and disk I/O from Amazon Elastic Compute Cloud (Amazon EC2) instances through AWS APIs.

2. Format conversion: YACE transforms CloudWatch metrics into Prometheus-compatible format.

3. Endpoint exposure: YACE exposes an HTTP /metrics endpoint on default port 5000. Managed Service for Prometheus automatically discovers and scrapes this endpoint.

Prerequisites

Before you begin, make sure that you have:

• A Container Service for Kubernetes (ACK) cluster

• The container monitoring service activated

Step 1: Deploy YACE in an ACK cluster

1. Download the values.yaml file from the Helm chart repository and configure the following sections. Prometheus scrape annotations AWS authentication Use an AccessKey pair (recommended), a Kubernetes secret, or an IAM role. CloudWatch metric collection Configure only the products, metrics, and statistics you need. Amazon CloudWatch charges per metric API request.

   Note: The EC2 metrics above are for illustration. For a full list of AWS services that publish metrics to CloudWatch, see AWS services that publish CloudWatch metrics. Configure other products similarly.

      podAnnotations: {
        prometheus.io/path: /metrics,
        prometheus.io/port: "5000",
        prometheus.io/scrape: "true"
      }

      aws:
        role:
   
        # Reference a Kubernetes secret that stores AWS credentials.
        # Fields: access_key (AccessKey ID), secret_key (AccessKey Secret),
        # and optionally security_token (session token).
        secret:
          name:
          includesSessionToken: false
   
        # Specify these directly only if you did not configure a role or secret.
        aws_access_key_id:
        aws_secret_access_key:

      config: |-
        apiVersion: v1alpha1
        sts-region: us-west-1  # Must match the region of the AWS resource
        discovery:
          jobs:
          - type: AWS/EC2  # Product namespace (see https://github.com/prometheus-community/yet-another-cloudwatch-exporter)
            regions:
              - us-west-1
            searchTags:  # Filter instances by tag
              - key: scrapedEnable
                value: ^(ali)$
            metrics:
              - name: CPUUtilization
                statistics:
                - Average
                - Minimum
                - Maximum
                period: 60
                length: 600
              - name: DiskReadOps
                statistics:
                - Sum
                - Average
                - Minimum
                - Maximum
                period: 60
                length: 600
              - name: DiskWriteOps
                statistics:
                - Sum
                - Average
                - Minimum
                - Maximum
                period: 60
                length: 600
              - name: DiskReadBytes
                statistics:
                - Sum
                - Average
                - Minimum
                - Maximum
                period: 60
                length: 600

2. Download the latest chart release and install it: Or install from the remote repository:

      helm install -f values.yaml yace-exporter-cw \
        ./prometheus-yet-another-cloudwatch-exporter-0.40.0.tgz \
        -n aws-cloudwatch

      helm repo add prometheus-community https://prometheus-community.github.io/helm-charts
      helm repo update
      helm install -f values.yaml yace-exporter-cw \
        prometheus-community/prometheus-yet-another-cloudwatch-exporter \
        -n aws-cloudwatch

Step 2: Enable automatic metric scraping

After YACE is running in your ACK cluster, enable default service discovery in Managed Service for Prometheus so it automatically scrapes the YACE /metrics endpoint.

1. Log on to the ARMS console. In the left-side navigation pane, click Integration Management.

2. Select the region where the ACK cluster is deployed. On the Container Service tab, click the target environment name.

3. Click the Metric Scraping tab. On the Default Service Discovery tab, click Enabled. The system now scrapes YACE metrics automatically.

   

4. Click the Self-Monitoring tab to confirm that a scraping task named kubernetes-pods has been created. An up state indicates that metric scraping is active.

   

Step 3: Validate metrics

Click the Metrics Explorer tab and verify that metrics prefixed with aws_ are available.

Configure alerts and dashboards

To set up alerts and dashboards for these metrics, follow steps 3 and 4 in Monitor an Alibaba Cloud service.