FeignClient timeout setting in OpenFegin practice

Offer arrives, dig friends to pick up! I am participating in the 2022 Spring Recruitment Check-In Event, click to view the event details .

1. Problem description:

If we have some synchronous request interfaces on the line such as:

1. Bulk email

2. Execute tens of thousands of data processing at a time

3. Synchronously call the payment interface

If we have the above operation, the following exception may occur

Actually there are two solutions here:

1. Since the processing time of the callee is relatively long, the caller can choose to increase the waiting time .
2. Modify the processing logic of the callee and optimize the design.
3. Modify the processing logic of the call, modify the direct call to an implicit call.

The problem is not divergent, I just want to make a salted fish, to ensure that the service does not report errors, we can directly change the waiting time of feign

Second, the solution:

In the solution part, we still post some of the existing code and logic to respond to the current scene.

part of the code

1. FeignThe interface is defined as follows:

@FeignClient(name = "payment-service", path = "/payment")
public interface PaymentFeign {
​
    @PostMapping("/create")
    PaymentVo create(@RequestBody @Validated PaymentDto paymentDto);
}
复制代码

2. The calling code is as follows:

    @Autowired
    private PaymentFeign paymentFeign;
​
    @GetMapping("/create")
    public OrderVo create(@Validated OrderDto orderDto) {
        log.info("uri:/order/create lang:{}", LocaleContextHolder.getLocale().getLanguage());
        // 忽略逻辑代码 N 行
        paymentFeign.create(paymentDto);
        log.info("uri:/order/create paymentFegin.create cost: {} ms", System.currentTimeMillis() - start);
        return orderVo;
    }
复制代码

3. feign timeout configuration

feign:
  client:
    config:
      default:
        # 日志级别
        loggerLevel: full
        # 超时设置 1.5 秒超时
        connectTimeout: 1500
        readTimeout: 1500
  # 断路器
  circuitbreaker:
    enabled: true
复制代码

Summarize the above code in one sentence: I called it in the xxx/createinterface , /payment/createand I set the timeout of feign to 1.5s. However, TimeOut is returned due to a timeout during the /payment/createcall .

solution

Option One:

Increase the default timeout of feign to 5s

feign:
  client:
    config:
      default:
        # 日志级别
        loggerLevel: full
        # 超时设置 5 秒超时
        connectTimeout: 5000
        readTimeout: 5000
  # 断路器
  circuitbreaker:
    enabled: true
复制代码

Option II:

Specify the contextId of the feign interface to set the current feign timeout to 5s.

feign:
  client:
    config:
      default:
        # 日志级别
        loggerLevel: full
        # 超时设置
        connectTimeout: 1500
        readTimeout: 1500
      payment-core:
        connectTimeout: 5000
        readTimeout: 5000
  # 断路器
  circuitbreaker:
    enabled: true
复制代码

PaymentFeignSet the contextId property on for

@FeignClient(name = "payment-service", contextId = "payment-core",  path = "/payment")
public interface PaymentFeign {
​
    @PostMapping("/create")
    PaymentVo create(@RequestBody @Validated PaymentDto paymentDto);
}
复制代码

In doing so, we can reduce the impact and reduce the impact downstream.

3. Summary:

Reference address information: spring-cloud-openfeign