Spring中resilience4j.circuitbreaker.configs与backends的区别及用法问询

Great question — let’s break down these three configuration properties clearly, since backends is a legacy concept that often causes confusion for newer Resilience4j users.

1. resilience4j.circuitbreaker.configs.<baseConfigName>: Reusable Base Configuration Templates #

Think of this as a "blueprint" for circuit breaker settings. These are global, reusable configurations that you can apply to multiple circuit breaker instances to avoid duplicating code.

For example, you might define a base config with common defaults:

# Define a shared base config
resilience4j.circuitbreaker.configs.default.sliding-window-size=100
resilience4j.circuitbreaker.configs.default.failure-rate-threshold=50
resilience4j.circuitbreaker.configs.default.wait-duration-in-open-state=10s

This default config isn’t a working circuit breaker on its own — it’s just a set of rules that other instances can reference and override as needed.

2. resilience4j.circuitbreaker.instances.<instanceName>: Actual Circuit Breaker Instances #

This is the modern, recommended way to define working circuit breaker instances. Each instance maps to a specific service, method, or external dependency you want to protect.

You can either configure an instance from scratch, or reference a base config to reuse settings and override specific values:

# Create an instance that uses the default base config
resilience4j.circuitbreaker.instances.paymentService.base-config=default
# Override one setting for this specific instance
resilience4j.circuitbreaker.instances.paymentService.wait-duration-in-open-state=15s

To use this instance in code, you’d reference its name in annotations or via the registry:

@CircuitBreaker(name = "paymentService")
public void processPayment() {
    // Payment processing logic
}

3. resilience4j.circuitbreaker.backends.<backendName>: Legacy Instance Naming #

This is an older naming convention for circuit breaker instances, used in Resilience4j versions 0.17.x and earlier. It’s functionally identical to instances — the only difference is the property name. It’s kept for backward compatibility, so existing code doesn’t break when upgrading.

How backends relate to configs #

Just like instances, you can link a backend to a base config using the base-config property:

# Legacy way to define a circuit breaker "backend" (same as an instance)
resilience4j.circuitbreaker.backends.notificationService.base-config=default
resilience4j.circuitbreaker.backends.notificationService.failure-rate-threshold=40

Using backends in code #

You reference backends the same way you’d reference instances — by name in annotations or registry calls:

@CircuitBreaker(name = "notificationService")
public void sendAlert() {
    // Notification logic
}

Key Takeaways #

• configs = Reusable, global setting templates (no active circuit breaker on their own)
• instances = Modern, recommended way to define active circuit breaker instances
• backends = Legacy alias for instances, maintained for backward compatibility
• Both instances and backends can reference configs via base-config to reuse and customize settings

内容的提问来源于stack exchange，提问作者Dirk Hoffmann