Production Debugging & Incident Response Playbook · レッスン

構造化ロギングのベストプラクティス

解析しやすい構造化ロギングを実装し、本番環境の問題をより迅速にデバッグできるようにします。

「構造化ロギングのベストプラクティス」はCoddyKit上の無料Production Debugging & Incident Response Playbookレッスンです。これはレッスン1/4です。下記で完全なレッスンを無料で読むことができます。その後、ブラウザ内の組み込みコードエディタと24時間対応のAIチューターでハンズオン演習できます。これはProduction Debugging & Incident Response Playbook学習パスの一部であり、ウェブとCoddyKitアプリ全体で進捗が同期されます。 Production Debugging & Incident Response Playbookコースには全4レッスンが含まれています。

このレッスンの一部はまだ翻訳されておらず、英語で表示されています。

What are Logs?

Logs are records of events that happen in your application or system. Think of them as a diary for your software!

They're crucial for understanding what your program is doing, especially when things go wrong in a live "production" environment.

The Messy Truth

Often, logs are just plain text strings. This is called unstructured logging. While easy to write, unstructured logs are hard for computers to read and analyze, making debugging a slow, manual process.

Consider this example:

import logging

logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)

def process_order(order_id, item_count):
    logger.info(f"Processing order {order_id} with {item_count} items.")
    if item_count > 10:
        logger.warning(f"Large order detected for {order_id}. Items: {item_count}.")
    logger.info(f"Order {order_id} processed successfully.")

if __name__ == "__main__":
    process_order("ORD-123", 5)
    process_order("ORD-456", 12)

What is Structured Logging?

Structured logging means your logs are formatted as machine-readable data, not just free-form text. The most common format is JSON.

Instead of a single string, each log entry is an object with key-value pairs. This makes them easy to search, filter, and analyze programmatically.

Why Structured Logging Rocks

Structured logs offer many advantages:

Faster Debugging: Quickly find relevant events.
Better Analysis: Easily query and aggregate data.
Automated Tools: Integrate with monitoring and alerting systems.
Consistency: Ensures all logs contain expected fields.

JSON is King

While other formats exist, JSON (JavaScript Object Notation) is the most popular choice for structured logging due to its simplicity and widespread support.

A JSON log entry is a self-contained object, making it incredibly versatile for storing varied data. Here's what a structured log might look like:

{
  "timestamp": "2023-10-27T10:30:00Z",
  "level": "INFO",
  "service": "order-processor",
  "message": "Order processed successfully",
  "order_id": "ORD-123",
  "item_count": 5
}

Code It Up!

Let's see how to implement structured logging. Many languages have libraries that make this easy. Here's a basic Python example using the standard logging module with a custom JSON formatter.

import logging
import json

class JsonFormatter(logging.Formatter):
    def format(self, record):
        log_entry = {
            "timestamp": self.formatTime(record, self.datefmt),
            "level": record.levelname,
            "name": record.name,
            "message": record.getMessage(),
            "file": record.filename,
            "line": record.lineno
        }
        if hasattr(record, 'order_id'):
            log_entry['order_id'] = record.order_id
        if hasattr(record, 'item_count'):
            log_entry['item_count'] = record.item_count
        return json.dumps(log_entry)

logger = logging.getLogger(__name__)
logger.setLevel(logging.INFO)

handler = logging.StreamHandler()
handler.setFormatter(JsonFormatter())
logger.addHandler(handler)

def process_order(order_id, item_count):
    extra_data = {'order_id': order_id, 'item_count': item_count}
    logger.info("Processing order", extra=extra_data)
    if item_count > 10:
        logger.warning("Large order detected", extra=extra_data)
    logger.info("Order processed successfully", extra=extra_data)

if __name__ == "__main__":
    process_order("ORD-123", 5)
    process_order("ORD-456", 12)

Must-Have Fields

Every structured log entry should include these core fields for effective analysis:

timestamp: When the event happened (ISO 8601 format).
level: Severity (INFO, WARN, ERROR, DEBUG).
service: Which service or application generated the log.
message: A human-readable summary of the event.
hostname/pod_name: Where the log originated.

Enrich Your Logs

Beyond essential fields, add contextual data specific to the event. This is key for tracing requests across distributed systems, helping you connect the dots when debugging complex issues.

request_id: To track a single user request.
user_id: To identify the user involved.
transaction_id: For specific business transactions.

import logging
import json
import uuid

# Reusing the JsonFormatter from previous scene
class JsonFormatter(logging.Formatter):
    def format(self, record):
        log_entry = {
            "timestamp": self.formatTime(record, self.datefmt),
            "level": record.levelname,
            "message": record.getMessage()
        }
        for key, value in record.__dict__.items():
            if not key.startswith('_') and key not in ['name', 'levelname', 'pathname', 'filename', 'module', 'exc_info', 'exc_text', 'stack_info', 'lineno', 'funcName', 'created', 'msecs', 'relativeCreated', 'thread', 'threadName', 'processName', 'process', 'args', 'msg', 'asctime']:
                log_entry[key] = value
        return json.dumps(log_entry)

logger = logging.getLogger(__name__)
logger.setLevel(logging.INFO)
handler = logging.StreamHandler()
handler.setFormatter(JsonFormatter())
logger.addHandler(handler)

def handle_web_request(user_id):
    request_id = str(uuid.uuid4())[:8]
    extra_data = {'request_id': request_id, 'user_id': user_id, 'service': 'api-gateway'}
    logger.info("Received web request", extra=extra_data)
    
    if user_id == "user-vip":
        logger.info("VIP user request detected", extra=extra_data)
    else:
        logger.debug("Standard user request", extra=extra_data)
    
    logger.info("Request processed", extra=extra_data)

if __name__ == "__main__":
    handle_web_request("user-123")
    handle_web_request("user-vip")

Log Levels

Log levels help categorize the severity and importance of a log message. Common levels include:

DEBUG: Detailed info, only useful when diagnosing problems.
INFO: Confirmation that things are working as expected.
WARN: An unexpected event, but the application is still running.
ERROR: An error that prevents some functionality from working.
CRITICAL: A severe error, application might be unable to continue.

Quick Check

Structured logging is a powerful technique for improving observability. Let's test your understanding of its key advantages.

Structured Logging Recap

You've learned about the power of structured logging! By formatting your logs as machine-readable data (like JSON), you unlock faster debugging, better analysis, and seamless integration with monitoring tools.

Remember to include essential fields and contextual data to make your logs truly useful for diagnosing issues in production.

よくある質問

「構造化ロギングのベストプラクティス」レッスンは無料ですか？

はい。「構造化ロギングのベストプラクティス」の完全なテキストはこのウェブで無料で読めます。インタラクティブに演習し（組み込みコードエディタと24時間対応のAIチューター）、Production Debugging & Incident Response Playbookコースの残りをアンロックするには、CoddyKit PROにアップグレードしてください。 Production Debugging & Incident Response Playbookコースには全4レッスンが含まれています。

「構造化ロギングのベストプラクティス」で何を学びますか？

解析しやすい構造化ロギングを実装し、本番環境の問題をより迅速にデバッグできるようにします。ブラウザで直接実行するハンズオンコードでProduction Debugging & Incident Response Playbookを演習し、24時間対応のAIチューターがレッスンを進める中での質問に答えます。

Production Debugging & Incident Response Playbookを始めるのに経験は必要ですか？

事前経験は必要ありません。CoddyKitのProduction Debugging & Incident Response Playbookは初級者から上級者向けに構成されているため、ここから始めるか最初から始めて、自分のペースで進むことができます。これはレッスン1/4です。

「構造化ロギングのベストプラクティス」レッスンにはどのくらい時間がかかりますか？

ほとんどのCoddyKitレッスンは約5～10分かかります。各レッスンはコンパクトでインタラクティブなので、着実に進歩し、ウェブとアプリ全体で正確に前回の場所から再開できます。

このProduction Debugging & Incident Response Playbookレッスンでコードを書いて実行できますか？

はい。すべてのProduction Debugging & Incident Response Playbookレッスンに組み込みコードエディタが含まれているため、ブラウザでリアルコードを書いて実行し、即座のAIフィードバックを取得できます。ローカル設定は不要です。

このコースのすべてのレッスン

← Production Debugging & Incident Response Playbookに戻る