我正在用spring boot开发REST API。我需要记录所有的请求与输入参数(与方法,例如。GET, POST等),请求路径,查询字符串,此请求对应的类方法,以及此操作的响应,包括成功和错误。例如:

成功的要求:

http://example.com/api/users/1

Log应该是这样的:

{
   HttpStatus: 200,
   path: "api/users/1",
   method: "GET",
   clientIp: "0.0.0.0",
   accessToken: "XHGu6as5dajshdgau6i6asdjhgjhg",
   method: "UsersController.getUser",
   arguments: {
     id: 1 
   },
   response: {
      user: {
        id: 1,
        username: "user123",
        email: "user123@example.com"   
      }
   },
   exceptions: []       
}

或请求错误:

http://example.com/api/users/9999

Log应该是这样的:

{
   HttpStatus: 404,
   errorCode: 101,                 
   path: "api/users/9999",
   method: "GET",
   clientIp: "0.0.0.0",
   accessToken: "XHGu6as5dajshdgau6i6asdjhgjhg",
   method: "UsersController.getUser",
   arguments: {
     id: 9999 
   },
   returns: {            
   },
   exceptions: [
     {
       exception: "UserNotFoundException",
       message: "User with id 9999 not found",
       exceptionId: "adhaskldjaso98d7324kjh989",
       stacktrace: ...................    
   ]       
}

我希望Request/Response是一个单独的实体,在成功和错误的情况下都具有与该实体相关的自定义信息。

春季实现这一目标的最佳做法是什么,可能是使用过滤器吗?如果是,能否提供具体的例子?

我使用过@ControllerAdvice和@ExceptionHandler,但正如我提到的,我需要在一个地方(和单个日志)处理所有成功和错误请求。


当前回答

自从最初的问题发布以来,执行器HTTP跟踪是否有任何发展,即有一种方法来丰富它与响应体?

用来自MDC、Spring-Sleuth或Zipkin的自定义元数据(如traceId和spanId)来丰富它怎么样?

同样对我来说,执行器HTTP跟踪不工作Spring Boot 2.2.3,我在这里找到了修复:https://juplo.de/actuator-httptrace-does-not-work-with-spring-boot-2-2/

pom.xml

<dependency>
  <groupId>org.springframework.boot
  <artifactId>spring-boot-starter-actuator
</dependency>

application.properties

management.endpoints.web.exposure.include=httptrace

解决办法:

解决这个问题的简单方法是添加一个@Bean类型 InMemoryHttpTraceRepository到你的@Configuration-class:

@Bean
public HttpTraceRepository htttpTraceRepository()
{
  return new InMemoryHttpTraceRepository();
}

解释:

The cause of this problem is not a bug, but a legitimate change in the default configuration. Unfortunately, this change is not noted in the according section of the documentation. Instead it is burried in the Upgrade Notes for Spring Boot 2.2 The default-implementation stores the captured data in memory. Hence, it consumes much memory, without the user knowing, or even worse: needing it. This is especially undesirable in cluster environments, where memory is a precious good. And remember: Spring Boot was invented to simplify cluster deployments! That is, why this feature is now turned of by default and has to be turned on by the user explicitly, if needed.

其他回答

下面是我如何在春季数据休息 通过使用 org.springframework.web.util.ContentCachingRequestWrapper和 org.springframework.web.util.ContentCachingResponseWrapper

/**
 * Doogies very cool HTTP request logging
 *
 * There is also {@link org.springframework.web.filter.CommonsRequestLoggingFilter}  but it cannot log request method
 * And it cannot easily be extended.
 *
 * https://mdeinum.wordpress.com/2015/07/01/spring-framework-hidden-gems/
 * http://stackoverflow.com/questions/8933054/how-to-read-and-copy-the-http-servlet-response-output-stream-content-for-logging
 */
public class DoogiesRequestLogger extends OncePerRequestFilter {

  private boolean includeResponsePayload = true;
  private int maxPayloadLength = 1000;

  private String getContentAsString(byte[] buf, int maxLength, String charsetName) {
    if (buf == null || buf.length == 0) return "";
    int length = Math.min(buf.length, this.maxPayloadLength);
    try {
      return new String(buf, 0, length, charsetName);
    } catch (UnsupportedEncodingException ex) {
      return "Unsupported Encoding";
    }
  }

  /**
   * Log each request and respponse with full Request URI, content payload and duration of the request in ms.
   * @param request the request
   * @param response the response
   * @param filterChain chain of filters
   * @throws ServletException
   * @throws IOException
   */
  @Override
  protected void doFilterInternal(HttpServletRequest request, HttpServletResponse response, FilterChain filterChain) throws ServletException, IOException {

    long startTime = System.currentTimeMillis();
    StringBuffer reqInfo = new StringBuffer()
     .append("[")
     .append(startTime % 10000)  // request ID
     .append("] ")
     .append(request.getMethod())
     .append(" ")
     .append(request.getRequestURL());

    String queryString = request.getQueryString();
    if (queryString != null) {
      reqInfo.append("?").append(queryString);
    }

    if (request.getAuthType() != null) {
      reqInfo.append(", authType=")
        .append(request.getAuthType());
    }
    if (request.getUserPrincipal() != null) {
      reqInfo.append(", principalName=")
        .append(request.getUserPrincipal().getName());
    }

    this.logger.debug("=> " + reqInfo);

    // ========= Log request and response payload ("body") ========
    // We CANNOT simply read the request payload here, because then the InputStream would be consumed and cannot be read again by the actual processing/server.
    //    String reqBody = DoogiesUtil._stream2String(request.getInputStream());   // THIS WOULD NOT WORK!
    // So we need to apply some stronger magic here :-)
    ContentCachingRequestWrapper wrappedRequest = new ContentCachingRequestWrapper(request);
    ContentCachingResponseWrapper wrappedResponse = new ContentCachingResponseWrapper(response);

    filterChain.doFilter(wrappedRequest, wrappedResponse);     // ======== This performs the actual request!
    long duration = System.currentTimeMillis() - startTime;

    // I can only log the request's body AFTER the request has been made and ContentCachingRequestWrapper did its work.
    String requestBody = this.getContentAsString(wrappedRequest.getContentAsByteArray(), this.maxPayloadLength, request.getCharacterEncoding());
    if (requestBody.length() > 0) {
      this.logger.debug("   Request body:\n" +requestBody);
    }

    this.logger.debug("<= " + reqInfo + ": returned status=" + response.getStatus() + " in "+duration + "ms");
    if (includeResponsePayload) {
      byte[] buf = wrappedResponse.getContentAsByteArray();
      this.logger.debug("   Response body:\n"+getContentAsString(buf, this.maxPayloadLength, response.getCharacterEncoding()));
    }

    wrappedResponse.copyBodyToResponse();  // IMPORTANT: copy content of response back into original response

  }


}

Spring已经提供了一个过滤器来完成这项工作。将以下bean添加到配置中

@Bean
public CommonsRequestLoggingFilter requestLoggingFilter() {
    CommonsRequestLoggingFilter loggingFilter = new CommonsRequestLoggingFilter();
    loggingFilter.setIncludeClientInfo(true);
    loggingFilter.setIncludeQueryString(true);
    loggingFilter.setIncludePayload(true);
    loggingFilter.setMaxPayloadLength(64000);
    return loggingFilter;
}

不要忘记将org.springframework.web.filter.CommonsRequestLoggingFilter的日志级别更改为DEBUG。

After adding Actuators to the spring boot bassed application you have /trace endpoint available with latest requests informations. This endpoint is working based on TraceRepository and default implementation is InMemoryTraceRepository that saves last 100 calls. You can change this by implementing this interface by yourself and make it available as a Spring bean. For example to log all requests to log (and still use default implementation as a basic storage for serving info on /trace endpoint) I'm using this kind of implementation:

import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.springframework.boot.actuate.trace.InMemoryTraceRepository;
import org.springframework.boot.actuate.trace.Trace;
import org.springframework.boot.actuate.trace.TraceRepository;
import org.springframework.stereotype.Component;

import java.util.List;
import java.util.Map;


@Component
public class LoggingTraceRepository implements TraceRepository {

  private static final Logger LOG = LoggerFactory.getLogger(LoggingTraceRepository.class);
  private final TraceRepository delegate = new InMemoryTraceRepository();

  @Override
  public List<Trace> findAll() {
    return delegate.findAll();
  }

  @Override
  public void add(Map<String, Object> traceInfo) {
    LOG.info(traceInfo.toString());
    this.delegate.add(traceInfo);
  }
}

这个traceInfo映射以这种形式包含了请求和响应的基本信息: {method=GET, path=/api/hello/John, headers={request={host=localhost:8080, user-agent=curl/7.51.0, accept=*/*}, response={X-Application-Context=application, Content-Type=text/plain;charset=UTF-8, Content-Length=10, Date=Wed, 29 Mar 2017 20:41:21 GMT, status=200}}}。这里没有响应内容。

编辑!测井POST数据

你可以通过覆盖WebRequestTraceFilter来访问POST数据,但不认为这是一个好主意(例如,所有上传的文件内容都会去日志) 下面是示例代码,但不要使用它:

package info.fingo.nuntius.acuate.trace;

import org.apache.commons.io.IOUtils;
import org.springframework.boot.actuate.trace.TraceProperties;
import org.springframework.boot.actuate.trace.TraceRepository;
import org.springframework.boot.actuate.trace.WebRequestTraceFilter;
import org.springframework.stereotype.Component;

import javax.servlet.ServletException;
import javax.servlet.http.HttpServletRequest;
import java.io.IOException;
import java.nio.charset.Charset;
import java.util.LinkedHashMap;
import java.util.Map;

@Component
public class CustomWebTraceFilter extends WebRequestTraceFilter {

  public CustomWebTraceFilter(TraceRepository repository, TraceProperties properties) {
    super(repository, properties);
}

  @Override
  protected Map<String, Object> getTrace(HttpServletRequest request) {
    Map<String, Object> trace = super.getTrace(request);
    String multipartHeader = request.getHeader("content-type");
    if (multipartHeader != null && multipartHeader.startsWith("multipart/form-data")) {
        Map<String, Object> parts = new LinkedHashMap<>();
        try {
            request.getParts().forEach(
                    part -> {
                        try {
                            parts.put(part.getName(), IOUtils.toString(part.getInputStream(), Charset.forName("UTF-8")));
                        } catch (IOException e) {
                            e.printStackTrace();
                        }
                    }
            );
        } catch (IOException | ServletException e) {
            e.printStackTrace();
        }
        if (!parts.isEmpty()) {
            trace.put("multipart-content-map", parts);
        }
    }
    return trace;
  }
}

为了记录所有带有输入参数和主体的请求,我们可以使用过滤器和拦截器。但是在使用过滤器或拦截器时,我们不能多次打印请求体。 更好的方法是使用spring-AOP。通过使用这个,我们可以将日志机制从应用程序中分离出来。AOP可用于记录应用程序中每个方法的输入和输出。

我的解决方案是:

 import org.aspectj.lang.ProceedingJoinPoint;
 import org.aspectj.lang.annotation.Around;
 import org.aspectj.lang.annotation.Aspect;
 import org.aspectj.lang.annotation.Pointcut;
 import org.aspectj.lang.reflect.CodeSignature;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 import org.springframework.stereotype.Component;
 import com.fasterxml.jackson.databind.ObjectMapper;
 @Aspect
 @Component
public class LoggingAdvice {
private static final Logger logger = 
LoggerFactory.getLogger(LoggingAdvice.class);

//here we can provide any methodName, packageName, className 
@Pointcut(value = "execution(* com.package.name.*.*.*(..) )")
public void myPointcut() {

}

@Around("myPointcut()")
public Object applicationLogger(ProceedingJoinPoint pjt) throws Throwable {
    ObjectMapper mapper = new ObjectMapper();
    String methodName = pjt.getSignature().getName();
    String className = pjt.getTarget().getClass().toString();
    String inputParams = this.getInputArgs(pjt ,mapper);
    logger.info("method invoked from " + className + " : " + methodName + "--Request Payload::::"+inputParams);
    Object object = pjt.proceed();
    try {
        logger.info("Response Object---" + mapper.writeValueAsString(object));
    } catch (Exception e) {
    }
    return object;
}

private String getInputArgs(ProceedingJoinPoint pjt,ObjectMapper mapper) {
    Object[] array = pjt.getArgs();
    CodeSignature signature = (CodeSignature) pjt.getSignature();

    StringBuilder sb = new StringBuilder();
    sb.append("{");
    int i = 0;
    String[] parameterNames = signature.getParameterNames();
    int maxArgs = parameterNames.length;
    for (String name : signature.getParameterNames()) {
        sb.append("[").append(name).append(":");
        try {
            sb.append(mapper.writeValueAsString(array[i])).append("]");
            if(i != maxArgs -1 ) {
                sb.append(",");
            }
        } catch (Exception e) {
            sb.append("],");
        }
        i++;
    }
    return sb.append("}").toString();
}

}

自从最初的问题发布以来,执行器HTTP跟踪是否有任何发展,即有一种方法来丰富它与响应体?

用来自MDC、Spring-Sleuth或Zipkin的自定义元数据(如traceId和spanId)来丰富它怎么样?

同样对我来说,执行器HTTP跟踪不工作Spring Boot 2.2.3,我在这里找到了修复:https://juplo.de/actuator-httptrace-does-not-work-with-spring-boot-2-2/

pom.xml

<dependency>
  <groupId>org.springframework.boot
  <artifactId>spring-boot-starter-actuator
</dependency>

application.properties

management.endpoints.web.exposure.include=httptrace

解决办法:

解决这个问题的简单方法是添加一个@Bean类型 InMemoryHttpTraceRepository到你的@Configuration-class:

@Bean
public HttpTraceRepository htttpTraceRepository()
{
  return new InMemoryHttpTraceRepository();
}

解释:

The cause of this problem is not a bug, but a legitimate change in the default configuration. Unfortunately, this change is not noted in the according section of the documentation. Instead it is burried in the Upgrade Notes for Spring Boot 2.2 The default-implementation stores the captured data in memory. Hence, it consumes much memory, without the user knowing, or even worse: needing it. This is especially undesirable in cluster environments, where memory is a precious good. And remember: Spring Boot was invented to simplify cluster deployments! That is, why this feature is now turned of by default and has to be turned on by the user explicitly, if needed.