Resilience4j Retry 和 "java.net.http.HttpClient" 一起工作时遇到问题

Question

我正在尝试使用 Resilience4j Retry 获得一个基本的“httpclient”“httprequest”“httpresponse”。

逐字代码来自：https://resilience4j.readme.io/docs/retry

  RetryConfig config = RetryConfig.custom()
  .maxAttempts(5)
  .waitDuration(Duration.ofMillis(1000))
  .retryOnResult(response -> response.getStatus() == 500)
  .retryOnException(e -> e instanceof WebServiceException)
  .retryExceptions(IOException.class, TimeoutException.class)
  .ignoreExceptions(BusinessException.class, OtherBusinessException.class)
  .build();

// Create a RetryRegistry with a custom global configuration
RetryRegistry registry = RetryRegistry.of(config);

// Get or create a Retry from the registry - 
// Retry will be backed by the default config
Retry retryWithDefaultConfig = registry.retry("name1");

注意，他们上面的代码没有定义通用的“T”，像这样：

  RetryConfig config = RetryConfig.<MyConcrete>custom()

和逐字代码来自：https://resilience4j.readme.io/docs/examples

Supplier<String> supplierWithResultAndExceptionHandler = SupplierUtils
  .andThen(supplier, (result, exception) -> "Hello Recovery");

Supplier<HttpResponse> supplier = () -> httpClient.doRemoteCall();
Supplier<HttpResponse> supplierWithResultHandling = SupplierUtils.andThen(supplier, result -> {
    if (result.getStatusCode() == 400) {
       throw new ClientException();
    } else if (result.getStatusCode() == 500) {
       throw new ServerException();
    }
    return result;
});
HttpResponse httpResponse = circuitBreaker
  .executeSupplier(supplierWithResultHandling);

======

所以使用这 2 个“部分”，我想出了这个。

注意，我正在使用一些“真实的”java.net.http.HttpClient 和 java.net.http.HttpResponse（来自 JDK11）

import io.github.resilience4j.core.SupplierUtils;
import io.github.resilience4j.retry.Retry;
import io.github.resilience4j.retry.RetryConfig;
import io.github.resilience4j.retry.RetryRegistry;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.springframework.http.HttpStatus;

import javax.inject.Inject;
import java.io.IOException;
import java.net.http.HttpClient;
import java.net.http.HttpRequest;
import java.net.http.HttpResponse;
import java.time.Duration;
import java.util.concurrent.TimeoutException;
import java.util.function.Supplier;

public final class ResilientHttpClient /* implements IResilientHttpClient */ {

    private static Logger logger;

    private final HttpClient httpClient;

    @Inject
    public ResilientHttpClient(final HttpClient httpClient) {
        this(LoggerFactory
                .getLogger(ResilientHttpClient.class), httpClient);
    }

    /**
     * Constructor, which pre-populates the provider with one resource instance.
     */
    public ResilientHttpClient(final Logger lgr,
                               final HttpClient httpClient) {
        if (null == lgr) {
            throw new IllegalArgumentException("Logger is null");
        }
        this.logger = lgr;

        if (null == httpClient) {
            throw new IllegalArgumentException("HttpClient is null");
        }

        this.httpClient = httpClient;

    }

    public String executeHttpRequest(String circuitbreakerInstanceName, HttpRequest httpRequest) {

        try {

            /* circuitbreakerInstanceName  is future place holder for .yml configuration see : https://resilience4j.readme.io/docs/getting-started-3 */

        RetryConfig config = RetryConfig.<HttpResponse>custom()
                    .waitDuration(Duration.ofMillis(1000))
                    .retryOnResult(response -> response.statusCode() == 500)
                    .retryOnException(e -> e instanceof ArithmeticException)
                    .retryExceptions(IOException.class, TimeoutException.class)
                    //.ignoreExceptions(BusinessException.class, OtherBusinessException.class)
                    .build();

            // Create a RetryRegistry with a custom global configuration
            RetryRegistry registry = RetryRegistry.of(config);

            // Get or create a Retry from the registry -
            // Retry will be backed by the default config
            Retry retryWithDefaultConfig = registry.retry(circuitbreakerInstanceName);

            Supplier<HttpResponse> supplier = () -> this.httpClient.send(httpRequest, HttpResponse.BodyHandlers.ofString());

            Supplier<String> supplierWithResultAndExceptionHandler = SupplierUtils
                    .andThen(supplier, (result, exception) -> "Hello Recovery");

            Supplier<HttpResponse> supplierWithResultHandling = SupplierUtils.andThen(supplier, result -> {
                if (result.statusCode() == HttpStatus.BAD_REQUEST.value()) {
                    throw new RuntimeException("400");
                } else if (result.statusCode() == HttpStatus.INTERNAL_SERVER_ERROR.value()) {
                    throw new RuntimeException("500");
                }
                return result;
            });

            HttpResponse<String> response = retryWithDefaultConfig.executeSupplier(supplierWithResultHandling);

            String responseBody = response.body();

            return responseBody;

        } catch (Exception ex) {
            throw new RuntimeException((ex));
        }
    }

}

我遇到的问题是：

行：

Supplier<HttpResponse> supplier = () - > this.httpClient.send(httpRequest, HttpResponse.BodyHandlers.ofString());

给出“未处理的异常”错误（在 intelliJ 中）“IOException，InterruptedException”

所以修改方法为：

 public String executeHttpRequest(String circuitbreakerInstanceName, HttpRequest httpRequest) throws IOException, InterruptedException {

“感觉不对”。但即使我尝试了......它也没有解决任何问题。 :(

这可能是一些 lamda 检查异常巫术。

但更重要的是：

所以我不知道我将两个部分放在一起的方式是否正确。全工作区样本有点少

感谢您的帮助。让基本的 httpclient“重试”几次应该不会太难。但是我的头撞到了墙上。

我的 gradle 依赖项。

dependencies {

    implementation group: 'javax.inject', name: 'javax.inject', version: javaxInjectVersion
    implementation group: 'org.slf4j', name: 'slf4j-api', version: slf4jVersion

    implementation group: 'org.springframework', name: 'spring-web', version: springWebVersion

    implementation "io.github.resilience4j:resilience4j-circuitbreaker:${resilience4jVersion}"
    implementation "io.github.resilience4j:resilience4j-ratelimiter:${resilience4jVersion}"
    implementation "io.github.resilience4j:resilience4j-retry:${resilience4jVersion}"
    implementation "io.github.resilience4j:resilience4j-bulkhead:${resilience4jVersion}"
    implementation "io.github.resilience4j:resilience4j-cache:${resilience4jVersion}"
    implementation "io.github.resilience4j:resilience4j-timelimiter:${resilience4jVersion}"


    testCompile group: 'junit', name: 'junit', version: junitVersion
}

和

   resilience4jVersion = '1.5.0'
    slf4jVersion = "1.7.30"
    javaxInjectVersion = "1"
 springWebVersion = '5.2.8.RELEASE'
    junitVersion = "4.12"

Answer 1

只是出于兴趣：

您使用的是哪个 Java 版本？ Java11?
为什么不能使用 Spring 引导？ Resilience4j Spring Boot starter 大大简化了配置。

如果配置 retryOnResult(response -> response.getStatus() == 500)，则不必再使用 SupplierUtils 将具有特定状态代码的 HttpResponse 映射到运行时异常。

RetryConfig config = RetryConfig.<HttpResponse<String>>custom()
            .waitDuration(Duration.ofMillis(1000))
            .retryOnResult(response -> response.statusCode() == 500)
            .retryExceptions(IOException.class, TimeoutException.class)
            .build();

请不要在 executeHttpRequest 中创建注册表和配置，而是将它们注入您的构造函数。

您可以像这样创建一个静态方法：

public static <T> HttpResponse<T> executeHttpRequest(Callable<HttpResponse<T>> callable, Retry retry, CircuitBreaker circuitBreaker) throws Exception {
        return Decorators.ofCallable(callable)
            .withRetry(retry)
            .withCircuitBreaker(circuitBreaker)
            .call();
}

并按如下方式调用方法：

HttpResponse<String> response = executeHttpRequest(
    () -> httpClient.send(request, HttpResponse.BodyHandlers.ofString()), 
    retry, 
    circuitBreaker);

Resilience4j Retry 和 "java.net.http.HttpClient" 一起工作时遇到问题

Trouble with Resilience4j Retry and "java.net.http.HttpClient" working together

resilience4j