Services or capabilities described in AWS documentation might vary by Region. To see the differences applicable to the AWS European Sovereign Cloud Region, see the AWS European Sovereign Cloud User Guide.

Monitor service events

Service Events provides automated deep observability for services monitored with CloudWatch Application Signals. It captures error metrics, function-level performance data, incident snapshots (when requests exceed latency thresholds or throw exceptions), and deployment events — without additional code changes.

How service events works

Service Events collects the following types of signals from your instrumented services:

Service Events is automatically enabled when you enable CloudWatch Application Signals for your service. Error metrics and exception tracking are active immediately. Function-call metrics require additional configuration — you must configure packages to instrument before function-call data is collected (see Enable function instrumentation). Service Events can be disabled by setting OTEL_AWS_SERVICE_EVENTS_ENABLED=false. Data flows from the ADOT SDK to the CloudWatch agent. The agent publishes events to CloudWatch Logs (/aws/service-events/service-name log groups) and CloudWatch Metrics.

Supported languages: Java, Python, and Node.js.

Data storage

Service Events stores data in CloudWatch Logs. CloudWatch publishes service event data to a log group with the prefix /aws/application-signals/service-name, where service-name is the value of your OTEL_SERVICE_NAME environment variable. One log group is created per service.

You are billed for log ingestion and storage at standard CloudWatch Logs rates.

View errors in the console

In the CloudWatch console, navigate to Application Signals, choose your service, and then choose the Errors tab. This tab shows exception metrics for your service.

The tab displays:

Select an exception to drill down into details including the stack trace, exception message, and a link to the associated trace.

Errors are grouped by operation, exception type, and top stack frames. Only the most recent representative of each group is shown.

View service events in logs

Service Events data is stored in CloudWatch Logs under log groups with the prefix /aws/service-events/service-name. You can query this data directly using CloudWatch Logs Insights to build custom views, create dashboards, or investigate specific incidents.

To query service events:

Service events in the CloudWatch Application Signals MCP (Model Context Protocol) server

Service events data is accessible through the CloudWatch Application Signals MCP (Model Context Protocol) server, enabling AI coding assistants and agents to query your service's runtime behavior directly.

Troubleshooting

Performance improvement

For setup and usage instructions, see the Application Signals MCP server on the GitHub website.

Configure service events

Prerequisites

To use service events, ensure you have the minimum required versions of the following components:

If you use Amazon EKS, see Enable your applications on Amazon EKS clusters for add-on setup instructions.

Features enabled by default

If you use CloudWatch Application Signals, the following service events signals are enabled by default with no additional configuration required:

The following features are opt-in and require setting environment variables to produce data:

General settings

Enable function instrumentation

Function instrumentation is enabled by default, but it produces no metrics until you configure which packages to instrument. Provide a package allow list to start collecting per-function telemetry:

Endpoint filtering

Endpoint filtering controls which endpoints generate endpoint error metrics and incident snapshots. These settings do not affect function instrumentation.

Latency thresholds

Use the following environment variables to configure latency thresholds for incident snapshot triggers.

Rate limiting

Use the following environment variables to control the rate at which service events data is collected and reported.

Configure deployment events

Deployment events are always emitted at application startup and every 24 hours. Providing deployment metadata enriches these events so you can correlate incidents and performance changes with specific code deployments.

Set the following environment variables on your application containers or processes to provide deployment metadata:

Configure deployment events with GitHub Actions

In your GitHub Actions workflow, use the built-in environment variables to populate deployment metadata. Add the following to your deployment step or container environment:

env:
  OTEL_AWS_SERVICE_EVENTS_GIT_COMMIT_SHA: ${{ github.sha }}
  OTEL_AWS_SERVICE_EVENTS_GIT_REPO_URL: ${{ github.server_url }}/${{ github.repository }}
  OTEL_AWS_SERVICE_EVENTS_DEPLOYMENT_ID: ${{ github.run_id }}
  OTEL_AWS_SERVICE_EVENTS_DEPLOYMENT_TIMESTAMP: $(date -u +%Y-%m-%dT%H:%M:%SZ)
  OTEL_AWS_SERVICE_EVENTS_DEPLOYMENT_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}

If you deploy container images, pass these values as environment variables in your task definition or pod spec. You can bake them into the image at build time or inject them at deploy time through your deployment configuration.

Configure deployment events with GitLab CI/CD

In your GitLab CI/CD pipeline, use the predefined CI/CD variables to populate deployment metadata. Add the following to your deployment job:

deploy:
  variables:
    OTEL_AWS_SERVICE_EVENTS_GIT_COMMIT_SHA: $CI_COMMIT_SHA
    OTEL_AWS_SERVICE_EVENTS_GIT_REPO_URL: $CI_PROJECT_URL
    OTEL_AWS_SERVICE_EVENTS_DEPLOYMENT_ID: $CI_PIPELINE_ID
    OTEL_AWS_SERVICE_EVENTS_DEPLOYMENT_TIMESTAMP: $(date -u +%Y-%m-%dT%H:%M:%SZ)
    OTEL_AWS_SERVICE_EVENTS_DEPLOYMENT_URL: $CI_PIPELINE_URL

Pass these variables to your application containers at deploy time through your container orchestration platform (for example, as environment variables in your Amazon ECS task definition or Kubernetes deployment manifest).