Monitoring and Observability

This document explains monitoring setup for serverless applications.

CloudWatch Logs

All Lambda functions automatically log to CloudWatch Logs.

View Logs

# Hello World function
aws logs tail /aws/lambda/devops-studio-dev-hello-world --follow

# API Handler function
aws logs tail /aws/lambda/devops-studio-dev-api-handler --follow

# Event Processor function
aws logs tail /aws/lambda/devops-studio-dev-event-processor --follow

CloudWatch Metrics

Lambda Metrics

Monitor:

• Invocations - Number of function invocations
• Duration - Execution time
• Errors - Number of errors
• Throttles - Number of throttled invocations

View Metrics

aws cloudwatch get-metric-statistics \
  --namespace AWS/Lambda \
  --metric-name Invocations \
  --dimensions Name=FunctionName,Value=devops-studio-dev-hello-world \
  --start-time $(date -u -d '1 hour ago' +%Y-%m-%dT%H:%M:%S) \
  --end-time $(date -u +%Y-%m-%dT%H:%M:%S) \
  --period 300 \
  --statistics Sum

CloudWatch Dashboards

Create dashboards to visualize:

• Lambda invocations
• Error rates
• Duration
• API Gateway requests
• DynamoDB operations

X-Ray Tracing

Enable X-Ray for distributed tracing:

from aws_xray_sdk.core import xray_recorder

@xray_recorder.capture('my_function')
def my_function():
    # Your code
    pass

Cost Monitoring

AWS Cost Explorer

Monitor costs by:

• Service (Lambda, API Gateway, DynamoDB)
• Function name
• Time period

Billing Alarms

Set up billing alarms to get notified of unexpected costs.

Best Practices

1. Enable Logging - Always enable CloudWatch Logs
2. Set Alarms - Monitor error rates and duration
3. Use Dashboards - Visualize metrics
4. Enable X-Ray - For distributed tracing
5. Monitor Costs - Track spending regularly