OpenFeign + Spring Cloud — Production Integration

Q: OpenFeign + Spring Cloud — Production Integration

**OpenFeign** is declarative; you define a Java interface annotated with `**@FeignClient**(name, url)` and Spring Cloud generates an HTTP proxy. Production-grade usage requires four pillars: **`RequestInterceptor`** to inject auth/tracing headers, **`ErrorDecoder`** to translate upstream errors into typed exceptions, **Resilience4j** `CircuitBreaker`/`Retry` via `@CircuitBreaker` on the Feign bean (or Spring Cloud CircuitBreaker auto-config), and explicit **`Feign.Builder` timeouts** (`connectTimeout`, `readTimeout`) because the defaults are 10s/60s — far too long. Use Spring Cloud LoadBalancer (`name = "payment-service"` with no `url`) to resolve service names via Eureka/Consul/Kubernetes.

Intermediate·3 min read·

NetflixSpring CloudStripeUberAdobeZalando

Fast recall

Quick revision

Core answer

OpenFeign is declarative; you define a Java interface annotated with **@FeignClient**(name, url) and Spring Cloud generates an HTTP proxy. Production-grade usage requires four pillars: `RequestInterceptor` to inject auth/tracing headers, `ErrorDecoder` to translate upstream errors into typed exceptions, Resilience4j CircuitBreaker/Retry via @CircuitBreaker on the Feign bean (or Spring Cloud CircuitBreaker auto-config), and explicit `Feign.Builder` timeouts (connectTimeout, readTimeout) because the defaults are 10s/60s — far too long. Use Spring Cloud LoadBalancer (name = "payment-service" with no url) to resolve service names via Eureka/Consul/Kubernetes.

Interview-ready guide

Interview answer

OpenFeign is still the right pick when you're already on Spring Cloud. It turns an HTTP call into an interface method, and then four pillars make it production-grade.

Declaration and discovery

@FeignClient(name = "payment-service", configuration = PaymentClientConfig.class) — no url, so Spring Cloud LoadBalancer resolves the host via Eureka, Consul, or Kubernetes
The per-client configuration class scopes interceptors, error decoders, and timeouts to this one remote

RequestInterceptor

Runs before every call; I inject the bearer token, tenant ID, and correlation headers here
Keep it scoped to the configuration class so internal tokens never leak to unrelated clients

ErrorDecoder

The default throws a generic FeignException — I always replace it with a decoder that maps 400/401/404/409/429/5xx to typed domain exceptions
For 5xx and I/O errors I return feign.RetryableException so the Retryer fires; everything else short-circuits

Timeouts

Feign defaults to 10s connect and 60s read, which is how one slow downstream saturates the caller's Tomcat pool
I set connectTimeout: 1000 and readTimeout: 2500 for internal calls, then compute the retry budget to fit inside the caller's own SLA

Circuit breaker + retry

I turn on spring.cloud.openfeign.circuitbreaker.enabled=true and define a Resilience4j CircuitBreaker per client name
Either Feign's Retryer or Resilience4j @Retry — never both, or attempts multiply
Fallback bean on @FeignClient(fallback = ...) for graceful degradation when the circuit opens

Observability

Micrometer auto-binds http.client.requests by client name; Micrometer Tracing injects traceparent into every call
loggerLevel: basic in production; never FULL, because it logs bodies and auth headers

When I wouldn't use Feign

Green-field Spring 6 service with no Spring Cloud — @HttpExchange is lighter
Reactive or streaming pipelines — use WebClient directly

03 · Deep dive

OpenFeign + Spring Cloud — Production Integration

About 4 min with examples

Direct Answer

OpenFeign is declarative; you define a Java interface annotated with **@FeignClient**(name, url) and Spring Cloud generates an HTTP proxy. Production-grade usage requires four pillars: RequestInterceptor to inject auth/tracing headers, ErrorDecoder to translate upstream errors into typed exceptions, Resilience4j CircuitBreaker/Retry via @CircuitBreaker on the Feign bean (or Spring Cloud CircuitBreaker auto-config), and explicit Feign.Builder timeouts (connectTimeout, readTimeout) because the defaults are 10s/60s — far too long. Use Spring Cloud LoadBalancer (name = "payment-service" with no url) to resolve service names via Eureka/Consul/Kubernetes.

Code Example

text

Enable Feign scanning at app startup

// Enable Feign scanning at app startup
@SpringBootApplication
@EnableFeignClients(basePackages = "com.acme.clients")
public class Application { }

// Declarative client — no url => resolved via LoadBalancer/discovery
@FeignClient(
    name = "payment-service",
    configuration = PaymentClientConfig.class,
    fallback = PaymentClientFallback.class   // optional, used when CB opens
)
public interface PaymentClient {

    @GetMapping("/api/v1/charges/{id}")
    ChargeResponse getCharge(@PathVariable("id") String id);

    @PostMapping("/api/v1/charges")
    ChargeResponse createCharge(
        @RequestHeader("Idempotency-Key") String idemKey,
        @RequestBody ChargeRequest body
    );

    @GetMapping("/api/v1/charges")
    Page<ChargeResponse> listCharges(@SpringQueryMap ChargeFilter filter);
}

// Fallback used when the circuit breaker is open (Resilience4j)
@Component
class PaymentClientFallback implements PaymentClient {
    @Override public ChargeResponse getCharge(String id) {
        return ChargeResponse.unknown(id);
    }
    @Override public ChargeResponse createCharge(String k, ChargeRequest b) {
        throw new PaymentUnavailableException("circuit open");
    }
    @Override public Page<ChargeResponse> listCharges(ChargeFilter f) {
        return Page.empty();
    }
}

Common Mistakes & Pitfalls

Shipping @FeignClient with default 10s/60s timeouts and no circuit breaker, so one slow downstream saturates the caller's Tomcat thread pool within minutes. Also: swallowing the Response body inside ErrorDecoder so errors arrive at the caller with no context.

Check your understanding

Answer these without looking back at the article.

How do you integrate Feign with Resilience4j so retry, circuit breaker, and timeout compose correctly?
What goes wrong when you leave the default Feign timeouts (10s/60s) in production?
How does Spring Cloud LoadBalancer resolve a @FeignClient with `name = "payment-service"` and no `url`?
When would you pick @HttpExchange over @FeignClient in a Spring 6 codebase?

RestTemplate vs WebClient vs Feign vs HTTP Interface

Continue · Next question · Q 18/44

WebClient Production Tuning — Pool, Timeouts, Backpressure

Microservices & APIs

REST, GraphQL, gRPC, Spring Cloud, Kafka, RabbitMQ, sagas

Full prep

Want a guided curriculum order?

These modules are also sequenced in our flagship backend interview track — open it when you prefer an ordered path instead of browsing by topic.

Open the guided track Browse all prep categories

OpenFeign + Spring Cloud — Production Integration

Fast recall

Quick revision

Core answer

Interview-ready guide

Interview answer

OpenFeign is still the right pick when you're already on Spring Cloud. It turns an HTTP call into an interface method, and then four pillars make it production-grade.

Declaration and discovery

@FeignClient(name = "payment-service", configuration = PaymentClientConfig.class) — no url, so Spring Cloud LoadBalancer resolves the host via Eureka, Consul, or Kubernetes
The per-client configuration class scopes interceptors, error decoders, and timeouts to this one remote

RequestInterceptor

Runs before every call; I inject the bearer token, tenant ID, and correlation headers here
Keep it scoped to the configuration class so internal tokens never leak to unrelated clients

ErrorDecoder

The default throws a generic FeignException — I always replace it with a decoder that maps 400/401/404/409/429/5xx to typed domain exceptions
For 5xx and I/O errors I return feign.RetryableException so the Retryer fires; everything else short-circuits

Timeouts

Feign defaults to 10s connect and 60s read, which is how one slow downstream saturates the caller's Tomcat pool
I set connectTimeout: 1000 and readTimeout: 2500 for internal calls, then compute the retry budget to fit inside the caller's own SLA

Circuit breaker + retry

I turn on spring.cloud.openfeign.circuitbreaker.enabled=true and define a Resilience4j CircuitBreaker per client name
Either Feign's Retryer or Resilience4j @Retry — never both, or attempts multiply
Fallback bean on @FeignClient(fallback = ...) for graceful degradation when the circuit opens

Observability

Micrometer auto-binds http.client.requests by client name; Micrometer Tracing injects traceparent into every call
loggerLevel: basic in production; never FULL, because it logs bodies and auth headers

When I wouldn't use Feign

Green-field Spring 6 service with no Spring Cloud — @HttpExchange is lighter
Reactive or streaming pipelines — use WebClient directly

03 · Deep dive

OpenFeign + Spring Cloud — Production Integration

About 4 min with examples

Direct Answer

OpenFeign is declarative; you define a Java interface annotated with **@FeignClient**(name, url) and Spring Cloud generates an HTTP proxy. Production-grade usage requires four pillars: RequestInterceptor to inject auth/tracing headers, ErrorDecoder to translate upstream errors into typed exceptions, Resilience4j CircuitBreaker/Retry via @CircuitBreaker on the Feign bean (or Spring Cloud CircuitBreaker auto-config), and explicit Feign.Builder timeouts (connectTimeout, readTimeout) because the defaults are 10s/60s — far too long. Use Spring Cloud LoadBalancer (name = "payment-service" with no url) to resolve service names via Eureka/Consul/Kubernetes.

Code Example

text

Enable Feign scanning at app startup

// Enable Feign scanning at app startup
@SpringBootApplication
@EnableFeignClients(basePackages = "com.acme.clients")
public class Application { }

// Declarative client — no url => resolved via LoadBalancer/discovery
@FeignClient(
    name = "payment-service",
    configuration = PaymentClientConfig.class,
    fallback = PaymentClientFallback.class   // optional, used when CB opens
)
public interface PaymentClient {

    @GetMapping("/api/v1/charges/{id}")
    ChargeResponse getCharge(@PathVariable("id") String id);

    @PostMapping("/api/v1/charges")
    ChargeResponse createCharge(
        @RequestHeader("Idempotency-Key") String idemKey,
        @RequestBody ChargeRequest body
    );

    @GetMapping("/api/v1/charges")
    Page<ChargeResponse> listCharges(@SpringQueryMap ChargeFilter filter);
}

// Fallback used when the circuit breaker is open (Resilience4j)
@Component
class PaymentClientFallback implements PaymentClient {
    @Override public ChargeResponse getCharge(String id) {
        return ChargeResponse.unknown(id);
    }
    @Override public ChargeResponse createCharge(String k, ChargeRequest b) {
        throw new PaymentUnavailableException("circuit open");
    }
    @Override public Page<ChargeResponse> listCharges(ChargeFilter f) {
        return Page.empty();
    }
}

Common Mistakes & Pitfalls

Check your understanding

Answer these without looking back at the article.

How do you integrate Feign with Resilience4j so retry, circuit breaker, and timeout compose correctly?
What goes wrong when you leave the default Feign timeouts (10s/60s) in production?
How does Spring Cloud LoadBalancer resolve a @FeignClient with `name = "payment-service"` and no `url`?
When would you pick @HttpExchange over @FeignClient in a Spring 6 codebase?

Full prep

Want a guided curriculum order?

These modules are also sequenced in our flagship backend interview track — open it when you prefer an ordered path instead of browsing by topic.

Open the guided track Browse all prep categories

OpenFeign + Spring Cloud — Production Integration

Direct Answer

Code Example

Common Mistakes & Pitfalls

Check your understanding

Related interview prep categories

Want a guided curriculum order?

OpenFeign + Spring Cloud — Production Integration

Direct Answer

Code Example

Common Mistakes & Pitfalls

Check your understanding

Related interview prep categories

Want a guided curriculum order?