Microservices Communication Patterns (Saga, Circuit Breaker) · レッスン

再試行ロジックを備えたサーキットブレーカー

最適なエラー処理と復旧を実現するため、サーキットブレーカーと再試行機構が互いに及ぼす影響を理解します。

レッスン 2/411 ステップ

「再試行ロジックを備えたサーキットブレーカー」はCoddyKit上の無料Microservices Communication Patterns (Saga, Circuit Breaker)レッスンです。これはレッスン2/4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Microservices Communication Patterns (Saga, Circuit Breaker)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。 Microservices Communication Patterns (Saga, Circuit Breaker)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。

Combining Circuit Breaker & Retry

In distributed systems, failures are inevitable. We've learned about the Retry Pattern for transient issues and the Circuit Breaker for persistent ones.

But how do these powerful patterns work together? Combining them effectively is key to building truly resilient microservices.

Recap: The Retry Pattern

The Retry Pattern automatically re-attempts an operation that has failed due to a temporary, transient error.

Use case: Network glitches, temporary service unavailability, database deadlocks.
Goal: Overcome momentary hiccups without user intervention.
Mechanism: Usually involves a delay between retries (e.g., exponential backoff).

Recap: The Circuit Breaker

A Circuit Breaker prevents an application from repeatedly invoking a service that is likely to fail. It "trips" the circuit to stop calls when too many errors occur.

Use case: Service is down, overloaded, or consistently returning errors.
Goal: Fail fast, prevent cascading failures, give the failing service time to recover.
States: Closed, Open, Half-Open.

Synergy: CB and Retry

Imagine a service experiencing a brief network blip. Retry can handle this gracefully. But what if the service is completely offline for an extended period?

Without a Circuit Breaker, retries would continuously hammer the unresponsive service, wasting resources and prolonging the problem. This is where their combined power shines!

Order of Operations

When combining these patterns, a critical design decision is: which one wraps the other?

Does the Retry Pattern wrap the Circuit Breaker, or does the Circuit Breaker wrap the Retry Pattern?

The order significantly impacts how your system responds to different types of failures.

Retry Wrapping Circuit Breaker

If the Retry Pattern wraps the Circuit Breaker:

Retry attempts the operation.
The Circuit Breaker is engaged.
If the CB opens, the first attempt fails, and retry might try again, hitting the already open CB.
This can lead to retries hitting a fast-failing CB, not allowing the CB to fully protect the system initially.

This setup is generally less effective.

Circuit Breaker Wrapping Retry

If the Circuit Breaker wraps the Retry Pattern:

The Circuit Breaker monitors the entire retry operation.
If the initial call fails, retry attempts again.
Only if all retries fail within the configured attempts, does the Circuit Breaker count it as a single failure.
If enough such "all-retry-failed" attempts occur, the CB opens.

This is the recommended approach.

CB Protecting Retry Logic

Here's a conceptual Java example showing how a Circuit Breaker would wrap an operation that includes retry logic. Notice the Circuit Breaker's decision to open or close is based on the final outcome of the retried call.

Try running this example:

import java.util.concurrent.atomic.AtomicInteger;

public class Main {
    // Simulate a dependency that sometimes fails
    private static AtomicInteger serviceCallCount = new AtomicInteger(0);

    public static boolean unreliableServiceCall() {
        System.out.println("  Attempting service call...");
        int currentCount = serviceCallCount.incrementAndGet();
        if (currentCount % 3 == 0) { // Fails every 3rd call
            System.out.println("  Service call FAILED temporarily.");
            return false;
        }
        System.out.println("  Service call SUCCESS.");
        return true;
    }

    public static boolean executeWithRetry() {
        int maxRetries = 2;
        long delayMillis = 100;
        for (int i = 0; i <= maxRetries; i++) {
            try {
                if (unreliableServiceCall()) {
                    return true; // Success after retry
                }
            } catch (Exception e) {
                // Log exception, continue retry
            }
            if (i < maxRetries) {
                System.out.println("  Retrying in " + delayMillis + "ms...");
                try { Thread.sleep(delayMillis); } catch (InterruptedException e) { Thread.currentThread().interrupt(); }
            }
        }
        return false; // All retries failed
    }

    // Conceptual Circuit Breaker logic for demonstration
    private static boolean circuitOpen = false;
    private static int failureCount = 0;
    private static final int FAILURE_THRESHOLD = 2; // Open after 2 consecutive failures
    private static final long RESET_TIMEOUT_MILLIS = 500; // Try to close after 0.5s
    private static long lastFailureTime = 0;

    public static boolean executeWithCircuitBreakerAndRetry() {
        if (circuitOpen) {
            if (System.currentTimeMillis() - lastFailureTime > RESET_TIMEOUT_MILLIS) {
                System.out.println("Circuit Breaker: Attempting HALF-OPEN state...");
                circuitOpen = false; // Move to half-open (for demo, just close)
                failureCount = 0; // Reset count
            } else {
                System.out.println("Circuit Breaker: OPEN! Failing fast.");
                return false; // Fail fast if open
            }
        }

        boolean success = executeWithRetry(); // Execute the retry logic

        if (!success) {
            failureCount++;
            lastFailureTime = System.currentTimeMillis();
            if (failureCount >= FAILURE_THRESHOLD) {
                circuitOpen = true;
                System.out.println("Circuit Breaker: OPENED due to repeated failures!");
            } else {
                System.out.println("Circuit Breaker: Failure detected, count=" + failureCount);
            }
        } else {
            failureCount = 0; // Reset failure count on success
            System.out.println("Circuit Breaker: Success, failure count reset.");
        }
        return success;
    }

    public static void main(String[] args) {
        System.out.println("--- Scenario: CB wrapping Retry ---");
        for (int i = 0; i < 7; i++) { // Simulate multiple requests
            System.out.println("\nRequest " + (i + 1) + ":");
            boolean overallSuccess = executeWithCircuitBreakerAndRetry();
            System.out.println("Overall result for Request " + (i + 1) + ": " + (overallSuccess ? "SUCCESS" : "FAILURE"));
            try { Thread.sleep(100); } catch (InterruptedException e) { Thread.currentThread().interrupt(); }
        }
    }
}

Why This Order is Best

Placing the Circuit Breaker around the Retry Pattern offers several advantages:

Efficient Failure Detection: The CB only opens after a series of genuinely failed operations (i.e., all retries failed), distinguishing transient issues from persistent outages.
Reduced Load: Once the CB is open, it prevents any further retry attempts, protecting the downstream service from being overwhelmed during a prolonged failure.
Faster Failures: When the service is truly down, the CB opens quickly, allowing your application to fail fast instead of waiting for all retries to exhaust.

Check Your Understanding

Consider a microservice that experiences intermittent network glitches (transient failures) and occasionally goes completely offline for maintenance (persistent failures).

You are implementing both the Retry Pattern and the Circuit Breaker Pattern to handle these scenarios. Which setup is generally recommended for optimal resilience?

対話的なレッスンで答える

Lesson Summary

We've explored the powerful synergy between the Circuit Breaker and Retry Patterns. While both enhance resilience, their combined effectiveness hinges on their interaction.

Remember, the best practice is to have the Circuit Breaker wrap the Retry Pattern. This allows retries to handle transient faults, while the Circuit Breaker steps in to protect against persistent failures, preventing cascading issues and improving overall system stability.

Keep building robust systems!

無料で開始

AI チューターと学ぶ Microservices Communication Patterns (Saga, Circuit Breaker) — 無料

ブラウザでリアルコードを書いて実行し、24/7 の AI チューターから瞬時にサポートを受け、ウェブまたはアプリで続きから学習できます。

コース: 12
レッスン: 48

無料で学習を始めるコース全体を表示

よくある質問

「再試行ロジックを備えたサーキットブレーカー」レッスンは無料ですか？

はい。「再試行ロジックを備えたサーキットブレーカー」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）、Microservices Communication Patterns (Saga, Circuit Breaker)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。 Microservices Communication Patterns (Saga, Circuit Breaker)コースには全4レッスンが含まれています。

「再試行ロジックを備えたサーキットブレーカー」で何を学びますか？

最適なエラー処理と復旧を実現するため、サーキットブレーカーと再試行機構が互いに及ぼす影響を理解します。ブラウザで直接実行するハンズオンコードでMicroservices Communication Patterns (Saga, Circuit Breaker)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。

Microservices Communication Patterns (Saga, Circuit Breaker)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Microservices Communication Patterns (Saga, Circuit Breaker)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2/4です。

「再試行ロジックを備えたサーキットブレーカー」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Microservices Communication Patterns (Saga, Circuit Breaker)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Microservices Communication Patterns (Saga, Circuit Breaker)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

← Microservices Communication Patterns (Saga, Circuit Breaker)に戻る