本系列代码地址:https://github.com/JoJoTec/sp…
咱们这一节在后面实现的带有链路信息的 Publisher 的工厂的根底上,实现公共日志记录的 GlobalFilter。回顾下咱们的需要:
咱们须要在网关记录每个申请的:
-
HTTP 相干元素:
- URL 相干信息
- 申请信息,例如 HTTP HEADER,申请工夫等等
- 某些类型的申请体
- 响应信息,例如响应码
- 某些类型响应的响应体
- 链路信息
记录申请与响应的 Body 须要留神的中央
后面的章节咱们提到过,对于申请与响应的 body 解决,如果用其后果放入主链路的话,会造成 Spring Cloud Sleuth 的链路信息失落。还有两个要留神的中央是:
- TCP 粘包拆包导致一个申请体宰割成好几份或者一个包蕴含几个申请
- 读取后要开释本来的申请 body 读取进去的 DataBuffer
为何要开释本来的申请 body 读取进去的 DataBuffer?因为读取进去后占用的 DataBuffer 如果手动不开释那么底层的计数始终不归零会造成内存透露。能够参考框架代码看出,这里的 DataBuffer 是 须要手动开释 的,参考源码:
ByteBufferDecoder.java
@Override
public ByteBuffer decode(DataBuffer dataBuffer, ResolvableType elementType,
@Nullable MimeType mimeType, @Nullable Map<String, Object> hints) {int byteCount = dataBuffer.readableByteCount();
ByteBuffer copy = ByteBuffer.allocate(byteCount);
copy.put(dataBuffer.asByteBuffer());
copy.flip();
DataBufferUtils.release(dataBuffer);
if (logger.isDebugEnabled()) {logger.debug(Hints.getLogPrefix(hints) + "Read" + byteCount + "bytes");
}
return copy;
}
咱们是想把能够输入到日志的 body 转换成字符串进行输入,为了代码简洁避免出错,咱们应用一个工具类来实现将 DataBuffer 读取成字符串并开释的操作:
package com.github.jojotech.spring.cloud.apigateway.common;
import com.google.common.base.Charsets;
import org.springframework.core.io.buffer.DataBuffer;
import org.springframework.core.io.buffer.DataBufferUtils;
public class BufferUtil {public static String dataBufferToString(DataBuffer dataBuffer) {byte[] content = new byte[dataBuffer.readableByteCount()];
dataBuffer.read(content);
DataBufferUtils.release(dataBuffer);
return new String(content, Charsets.UTF_8);
}
}
编写实现公共日志记录 GlobalFilter
后面铺垫了那么多,咱们终于能够着手开始写这个日志 GlobalFilter 了:
package com.github.jojotech.spring.cloud.apigateway.filter;
import java.net.URI;
import java.util.Set;
import com.alibaba.fastjson.JSON;
import com.github.jojotech.spring.cloud.apigateway.common.BufferUtil;
import com.github.jojotech.spring.cloud.apigateway.common.TracedPublisherFactory;
import lombok.extern.log4j.Log4j2;
import org.reactivestreams.Publisher;
import reactor.core.publisher.Flux;
import reactor.core.publisher.Mono;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.cloud.gateway.filter.GatewayFilterChain;
import org.springframework.cloud.gateway.filter.GlobalFilter;
import org.springframework.core.Ordered;
import org.springframework.core.io.buffer.DataBuffer;
import org.springframework.core.io.buffer.DataBufferFactory;
import org.springframework.http.HttpHeaders;
import org.springframework.http.HttpMethod;
import org.springframework.http.MediaType;
import org.springframework.http.server.reactive.ServerHttpRequest;
import org.springframework.http.server.reactive.ServerHttpRequestDecorator;
import org.springframework.http.server.reactive.ServerHttpResponse;
import org.springframework.http.server.reactive.ServerHttpResponseDecorator;
import org.springframework.stereotype.Component;
import org.springframework.web.server.ServerWebExchange;
@Log4j2
@Component
public class CommonLogFilter implements GlobalFilter, Ordered {
// 能够输入的 body 格局
public static final Set<MediaType> legalLogMediaTypes = Set.of(
MediaType.TEXT_XML,
MediaType.TEXT_PLAIN,
MediaType.APPLICATION_XML,
MediaType.APPLICATION_JSON
);
@Autowired
private TracedPublisherFactory tracedPublisherFactory;
@Override
public Mono<Void> filter(ServerWebExchange exchange, GatewayFilterChain chain) {long startTime = System.currentTimeMillis();
ServerHttpRequest request = exchange.getRequest();
ServerHttpResponse response = exchange.getResponse();
// 获取用于拆包解决聚合读取申请和响应 body 的 buffer 的 factory
DataBufferFactory dataBufferFactory = response.bufferFactory();
// 申请 http 头
HttpHeaders requestHeaders = request.getHeaders();
// 申请 body 类型
MediaType requestContentType = requestHeaders.getContentType();
// 申请 uri
String uri = request.getURI().toString();
// 申请 http 办法
HttpMethod method = request.getMethod();
log.info("{} -> {}: header: {}", method, uri, JSON.toJSONString(requestHeaders));
Flux<DataBuffer> dataBufferFlux = tracedPublisherFactory.getTracedFlux(request.getBody(), exchange)
// 应用 buffer 在这里将所有 body 读取完防止拆包影响
.buffer()
.map(dataBuffers -> {
// 将所有 buffer 粘合在一起
DataBuffer dataBuffer = dataBufferFactory.join(dataBuffers);
// 只有在 debug 开启的时候,才会输入 body
if (log.isDebugEnabled()) {
// 只有特定的 body 类型才会输入具体的
if (legalLogMediaTypes.contains(requestContentType)) {
try {
// 将 body 转化为 String 进行输入,同时留神,原始的 buffer 须要被开释,因为 body 流曾经被读取进去,然而没有中央回收
// 参考
String s = BufferUtil.dataBufferToString(dataBuffer);
log.debug("body: {}", s);
dataBuffer = dataBufferFactory.wrap(s.getBytes());
}
catch (Exception e) {log.error("error read request body: {}", e.getMessage(), e);
}
}
else {log.debug("body: {}", request);
}
}
return dataBuffer;
});
return chain.filter(exchange.mutate().request(new ServerHttpRequestDecorator(request) {
@Override
public Flux<DataBuffer> getBody() {return dataBufferFlux;}
}).response(new ServerHttpResponseDecorator(response) {
@Override
public Mono<Void> writeWith(Publisher<? extends DataBuffer> body) {HttpHeaders responseHeaders = super.getHeaders();
// 调用这里的是写响应回客户端的 HttpClientConnect 的回写,曾经跳出了 Spring Cloud Sleuth 的链路 Span,所以没有链路追踪信息
// 然而咱们在 CommonTraceFilter 咱们将链路信息放入了响应 Header 中,所以这里咱们就不必手动减少链路信息了
log.info("response: {} -> {} {} header: {}, time: {}ms", method, uri, getStatusCode(), JSON.toJSONString(responseHeaders), System.currentTimeMillis() - startTime);
final MediaType contentType = responseHeaders.getContentType();
if (contentType != null && body instanceof Flux && legalLogMediaTypes.contains(contentType) && log.isDebugEnabled()) {
// 有 TCP 粘包拆包问题,这个 body 是屡次写入的,一次调用拿不到残缺的 body,所以这里转换成 fluxBody 利用其中的 buffer 来承受残缺的 body
Flux<? extends DataBuffer> fluxBody = tracedPublisherFactory.getTracedFlux(Flux.from(body), exchange);
return super.writeWith(fluxBody.buffer().map(buffers -> {DataBuffer buffer = dataBufferFactory.join(buffers);
try {String s = BufferUtil.dataBufferToString(buffer);
log.debug("response: body: {}", s);
return dataBufferFactory.wrap(s.getBytes());
} catch (Exception e) {log.error("error read response body: {}", e.getMessage(), e);
}
return buffer;
}));
}
// if body is not a flux. never got there.
return super.writeWith(body);
}
}).build());
}
@Override
public int getOrder() {
// 指定程序,在 CommonTraceFilter(这个 Filter 是读取链路信息,最好在所有 Filter 之前)之后
return new CommonTraceFilter().getOrder() + 1;
}
}
须要留神的点都在正文当中明确标出了,请大家参考。
查看日志
咱们通过退出上面的日志配置,关上 body 的日志,这样日志就全了:
<AsyncLogger name="com.github.jojotech.spring.cloud.apigateway.filter.CommonLogFilter" level="debug" additivity="false" includeLocation="true">
<appender-ref ref="console" />
</AsyncLogger>
发送一个 POST 带 body 的申请,从日志中就能看到:
2021-11-29 14:08:42,231 INFO [sports,8481ce2786b686fa,8481ce2786b686fa] [24916] [reactor-http-nio-2][com.github.jojotech.spring.cloud.apigateway.filter.CommonLogFilter:59]:POST -> http://127.0.0.1:8181/test-ss/anything?test=1: header: {"Content-Type":["text/plain"],"User-Agent":["PostmanRuntime/7.28.4"],"Accept":["*/*"],"Postman-Token":["666b17c9-0789-46e6-b515-9a4538803308"],"Host":["127.0.0.1:8181"],"Accept-Encoding":["gzip, deflate, br"],"Connection":["keep-alive"],"content-length":["8"]}
2021-11-29 14:08:42,233 DEBUG [sports,8481ce2786b686fa,8481ce2786b686fa] [24916] [reactor-http-nio-2][com.github.jojotech.spring.cloud.apigateway.filter.CommonLogFilter:74]:body: ifasdasd
2021-11-29 14:08:42,463 INFO [sports,,] [24916] [reactor-http-nio-2][com.github.jojotech.spring.cloud.apigateway.filter.CommonLogFilter$1:96]:response: POST -> http://127.0.0.1:8181/test-ss/anything?test=1 200 OK header: {"traceId":["8481ce2786b686fa"],"spanId":["8481ce2786b686fa"],"Date":["Mon, 29 Nov 2021 14:08:43 GMT"],"Content-Type":["application/json"],"Server":["gunicorn/19.9.0"],"Access-Control-Allow-Origin":["*"],"Access-Control-Allow-Credentials":["true"],"content-length":["886"]}, time: 232ms
2021-11-29 14:08:42,466 DEBUG [sports,8481ce2786b686fa,8481ce2786b686fa] [24916] [reactor-http-nio-2][com.github.jojotech.spring.cloud.apigateway.filter.CommonLogFilter$1:105]:response: body: {
"args": {"test": "1"},
"data": "ifasdasd",
"files": {},
"form": {},
"headers": {
"Accept": "*/*",
"Accept-Encoding": "gzip, deflate, br",
"Content-Length": "8",
"Content-Type": "text/plain",
"Forwarded": "proto=http;host=\"127.0.0.1:8181\";for=\"127.0.0.1:57526\"","Host":"httpbin.org","Postman-Token":"666b17c9-0789-46e6-b515-9a4538803308","User-Agent":"PostmanRuntime/7.28.4","X-Amzn-Trace-Id":"Root=1-61a4deeb-3d016ff729306d862edcca0b","X-B3-Parentspanid":"8481ce2786b686fa","X-B3-Sampled":"0","X-B3-Spanid":"5def545b28a7a842","X-B3-Traceid":"8481ce2786b686fa","X-Forwarded-Host":"127.0.0.1:8181","X-Forwarded-Prefix":"/test-ss"},"json": null,"method":"POST","origin":"127.0.0.1, 61.244.202.46","url":"http://127.0.0.1:8181/anything?test=1"
}
2021-11-29 14:08:42,474 INFO [sports,,] [24916] [reactor-http-nio-2][reactor.util.Loggers$Slf4JLogger:269]:8481ce2786b686fa,8481ce2786b686fa -> 127.0.0.1:57526 - - [2021-11-29T14:08:42.230008Z[Etc/GMT]] "POST /test-ss/anything?test=1 HTTP/1.1" 200 886 243 ms
微信搜寻“我的编程喵”关注公众号,每日一刷,轻松晋升技术,斩获各种 offer: