AWS Lambda Monitoring Overview

• CloudWatch Metrics for Lambda:
  • Invocations: Number of times the function is called.
  • Duration: Time taken to process an event.
  • Errors: Number of invocations resulting in an error.
  • Throttles: Number of requests that were throttled due to concurrency limits.
  • DeadLetterErrors: Failures to send an event to the dead letter queue.
  • IteratorAge: Lag in reading from a stream-based event source mapping.
  • ConcurrentExecutions: Number of function instances processing events simultaneously.
• Lambda Metrics Dashboard:
  • Provides a visual representation of the above metrics for monitoring.
• CloudWatch Alarms:
  • Can be set up based on metrics to alert on certain conditions, such as:
    • No invocations (invocations == 0).
    • Any errors (errors > 0).
    • Throttling occurrences (throttles > 0).
• CloudWatch Logs:
  • Lambda functions send logs to CloudWatch Logs.
  • Requires the Lambda execution role to have permissions for log operations.
  • Each function instance has a separate log stream within a log group.
• CloudWatch Logs Insights:
  • Allows searching and analyzing Lambda logs.
  • Can run queries to extract information like:
    • Number of errors in the last 7 days.
    • Last 100 errors.
    • Percentage of cold starts.
    • Invocations using 100% of assigned memory.
    • Average memory usage.
    • Number of timeouts.
• Lambda Insights:
  • Collects and summarizes system-level metrics (CPU, memory, disk, network).
  • Provides diagnostic information (cold starts, worker shutdowns).
  • Data is sent to CloudWatch Logs in a Lambda Insights log group.
  • Implemented as a Lambda layer and extension.

Example CloudWatch Logs Insights Queries

• Last 100 Errors:

  fields @timestamp, @message
  | filter @message like /Error/
  | sort @timestamp desc
  | limit 100
  

• Percentage of Cold Starts:

  fields @timestamp, @message
  | stats count() as totalInvocations, sum(isColdStart) as coldStarts
  | filter @message like /START RequestId: [ID] Version: $LATEST/
  | eval isColdStart = @message like /Init Duration/
  | eval coldStartPercentage = coldStarts / totalInvocations * 100
  

• Average Memory Usage:

  fields @timestamp, @memorySize, @maxMemoryUsed
  | stats avg(@maxMemoryUsed / @memorySize * 100) as avgMemoryUsage
  

Setting Up Lambda Insights

1. Add Lambda Insights as a layer to your Lambda function.
2. Ensure the Lambda function has the necessary permissions.
3. Monitor the system-level metrics and diagnostic information through the CloudWatch Logs Lambda Insights log group.

Conclusion

• Monitoring AWS Lambda functions is crucial for understanding performance and identifying issues.
• CloudWatch provides a suite of tools for metrics, alarms, and logs.
• CloudWatch Logs Insights and Lambda Insights offer advanced monitoring capabilities.
• Proper setup and use of these tools can help quickly isolate and resolve issues with Lambda functions.