CloudWatch를 활용한 로깅과 모니터링

When your Lambda functions run in the cloud, you can't just attach a debugger. This is where logging and monitoring become incredibly important!

They help you understand what your function is doing, debug issues, and ensure it's performing well.

AWS CloudWatch is the central observability service for AWS. It collects and processes raw data from AWS services (like Lambda) into readable metrics and logs.

• CloudWatch Logs: Stores your function's text output.
• CloudWatch Metrics: Gathers performance data (invocations, errors, duration).
• CloudWatch Alarms: Notifies you when metrics cross defined thresholds.

Good news! AWS Lambda automatically integrates with CloudWatch Logs. Any output your function sends to stdout (standard output) or stderr (standard error) will be captured.

This means simple print() statements in Python, or console.log() in Node.js, will show up in CloudWatch Logs.

While print() works, for more robust logging in Python, you should use the built-in logging module. It allows you to:

• Set different log levels (DEBUG, INFO, WARNING, ERROR, CRITICAL).
• Include timestamps and other metadata automatically.
• Format your log messages consistently.

Try running this simple Python Lambda function. Notice how both print() and logger.info() messages are captured. In a real Lambda, these would appear in CloudWatch Logs.

When Lambda sends logs to CloudWatch, they are organized in a specific way:

• Log Group: A container for logs from a specific application or service. For Lambda, it's typically /aws/lambda/YOUR_FUNCTION_NAME.
• Log Stream: Within a Log Group, each instance or invocation of your Lambda function creates a new Log Stream to store its logs.

You can access your function's logs in the AWS Console:

1. Navigate to the Lambda service.
2. Select your function.
3. Go to the 'Monitor' tab.
4. Click 'View logs in CloudWatch' to see the Log Group and Streams.

Here you can filter, search, and analyze your log data to debug issues.

Beyond logs, CloudWatch automatically collects metrics for your Lambda functions, giving you insights into their performance and health without any extra code.

Key metrics include:

• Invocations: Total number of times your function was triggered.
• Errors: Count of failed invocations.
• Duration: Execution time of your function.
• Throttles: When Lambda denied an invocation due to concurrency limits.

CloudWatch Alarms allow you to set up notifications or actions based on metric thresholds. For example, you can create an alarm that:

• Triggers if the 'Errors' metric for your function is greater than 0 for 5 minutes.
• Sends a notification via Amazon SNS (Simple Notification Service) to your email or an alerting system.

This is crucial for proactive monitoring!

Great job! You've learned how critical logging and monitoring are for serverless applications, especially with AWS Lambda.

• Lambda seamlessly integrates with CloudWatch Logs for capturing output.
• The Python logging module provides robust logging.
• CloudWatch Metrics give you insights into function performance.
• CloudWatch Alarms enable proactive alerts based on these metrics.

These tools are essential for debugging, performance tuning, and maintaining healthy serverless backends.