1. 背景
随着微服务架构的流行，服务按照不同的维度进行拆分，一次请求往往需要涉及到多个服务。而诸多的服务可能分布在了几千台服务器，横跨多个不同的数据中心。为了快速定位和解决故障，应用性能进行分析，全链路监控组件就在这样的问题背景下产生了。最出名的是谷歌公开的论文提到的 Google Dapper。想要在这个上下文中理解分布式系统的行为，就需要监控那些横跨了不同的应用、不同的服务器之间的关联动作。
1.1 全链路原理
通过业务调用过程中添加并传递调用链ID，实现应用间生成链路数据，最终串联成一条完整的调用链。其中整个调用过程中每个请求都要透传TxId、SpanId和pSpanId。
1.2 Spring Cloud Gateway
作为Spring Cloud官方推出的第二代网关框架，Spring cloud gateway是基于Spring 5.0、Spring Boot2.0和Reactor等技术开发的网关，采用了NIO模型进行通信。
1.2.1 Spring WebFlux
Spring Boot 2.0 包括一个新的 spring-webflux 模块，名称中的 Flux 来源于 Reactor 中的类 Flux。该模块包含对响应式 HTTP 和 WebSocket 客户端的支持，以及对 REST、HTML 和 WebSocket 交互等程序的支持。 一般来说，Spring MVC 用于同步处理；Spring Webflux 用于异步处理。
1.2.2 Mono与Flux
Mono表示的是包含 0 或者 1 个元素的异步序列，即要么成功发布元素，要么错误
Flux 表示的是包含 0 到 N 个元素的异步序列 ，即要么成功发布 0 到 N 个元素，要么错误
Flux和Mono之间可以相互转换，比如对一个 Flux 序列进行计数操作，得到的结果是一个 Mono对象，或者把两个 Mono 序列合并在一起，得到的是一个 Flux 对象。
2. Spring Cloud Gateway不做监控
Spring Cloud Gateway作为入口网关，主要负责服务的路由转发。如果网关没有进行监控，则全链路会缺失网关节点，直接展示为用户访问后续应用；不能有效定位用户请求慢是网关问题还是后续节点。
3. Spring Cloud Gateway进行监控
由于Spring Cloud Gateway采用了Reactor框架非阻塞式调用，任务之间会跨线程执行，导致HTTP头信息所需的调用链ID不好传递。
- Gateway接收线程
- Gateway返回线程
- 路由转发线程
3.1 Spring Cloud Gateway流程
现对Spring Cloud Gateway的流程进行梳理，本篇由于只涉及到跨线程服务调度，不讨论路由过程。
3.1.1 Spring Cloud Gateway的请求入口
org.springframework.http.server.reactive.ReactorHttpHandlerAdapter#apply 先将接收到的HttpServerRequest或者最终需要返回的HttpServerResponse包装转换为ReactorServerHttpRequest和ReactorServerHttpResponse，再处理请求。
public Mono apply(HttpServerRequest request, HttpServerResponse response) {
NettyDataBufferFactory bufferFactory = new NettyDataBufferFactory(response.alloc());
ServerHttpRequest adaptedRequest;
ServerHttpResponse adaptedResponse;
try {
adaptedRequest = new ReactorServerHttpRequest(request, bufferFactory);
adaptedResponse = new ReactorServerHttpResponse(response, bufferFactory);
} catch (URISyntaxException ex) {
logger.error(“Invalid URL ” + ex.getMessage(), ex);
response.status(HttpResponseStatus.BAD_REQUEST);
return Mono.empty();
}
if (adaptedRequest.getMethod() == HttpMethod.HEAD) {
adaptedResponse = new HttpHeadResponseDecorator(adaptedResponse);
}
returnthis.httpHandler.handle(adaptedRequest, adaptedResponse)
.doOnError(ex -> logger.error(“Handling completed with error“, ex))
.doOnSuccess(aVoid -> logger.debug(“Handling completed with success“));
}
3.1.2 构造网关上下文
org.springframework.web.server.adapter.HttpWebHandlerAdapter#handle
- createExchange()构造网关上下文ServerWebExchange
- getDelegate()通过委托的方式获取一系列需要处理的WebHandler
public Mono handle(ServerHttpRequest request, ServerHttpResponse response) {
ServerWebExchange exchange = createExchange(request, response);
returngetDelegate().handle(exchange)
.onErrorResume(ex -> handleFailure(request, response, ex))
.then(Mono.defer(response::setComplete));
}
protected ServerWebExchange createExchange(ServerHttpRequest request, ServerHttpResponse response) {
returnnew DefaultServerWebExchange(request, response, this.sessionManager,
getCodecConfigurer(), getLocaleContextResolver(), this.applicationContext);
}
3.1.3 进入Filter链
org.springframework.cloud.gateway.handler.FilteringWebHandler#handle 获得 GatewayFilter 数组，并根据获得的 GatewayFilter 数组创建DefaultGatewayFilterChain，过滤处理请求。
public Mono handle(ServerWebExchange exchange) {
Route route = exchange.getRequiredAttribute(GATEWAY_ROUTE_ATTR);
List gatewayFilters = route.getFilters();
List combined = new ArrayList(this.globalFilters);
combined.addAll(gatewayFilters);
AnnotationAwareOrderComparator.sort(combined);
logger.debug(“Sorted gatewayFilterFactories: “+ combined);
returnnew DefaultGatewayFilterChain(combined).filter(exchange);
}
3.1.4 执行Filter链
org.springframework.cloud.gateway.handler.FilteringWebHandler$DefaultGatewayFilterChain#filter 过滤器的链式调用
public Mono filter(ServerWebExchange exchange) {
return Mono.defer(() -> {
if (this.index < filters.size()) {
GatewayFilter filter = filters.get(this.index);
DefaultGatewayFilterChain chain = new DefaultGatewayFilterChain(this, this.index + 1);
returnfilter.filter(exchange, chain);
} else {
return Mono.empty(); // complete
}
});
}
3.1.5 Gateway Filter适配器
org.springframework.cloud.gateway.handler.FilteringWebHandler$GatewayFilterAdapter#filter
GatewayFilterAdapter是GlobalFilter过滤器的包装类，最终委托Global Filter进行执行。
privatestaticclass GatewayFilterAdapter implements GatewayFilter {
privatefinalGlobalFilter delegate;
public GatewayFilterAdapter(GlobalFilter delegate) {
this.delegate = delegate;
}
public Mono filter(ServerWebExchange exchange, GatewayFilterChain chain) {
returnthis.delegate.filter(exchange, chain);
}
}
3.1.6 Netty路由网关过滤器
org.springframework.cloud.gateway.filter.NettyRoutingFilter#filter
GlobalFilter实现有很多，此处只分析NettyRoutingFIlter和NettyWriteResponseFilter。而NettyRoutingFilter负责使用 Netty HttpClient 代理对下游的请求。
public Mono filter(ServerWebExchange exchange, GatewayFilterChain chain) {
// 获得 requestUrl
URI requestUrl = exchange.getRequiredAttribute(GATEWAY_REQUEST_URL_ATTR);
// 判断是否能够处理，http或https前缀
String scheme = requestUrl.getScheme();
if (isAlreadyRouted(exchange) || (!”http“.equals(scheme) && !”https“.equals(scheme))) {
return chain.filter(exchange);
}
// 设置已经路由
setAlreadyRouted(exchange);
ServerHttpRequest request = exchange.getRequest();
// 创建Netty Request Method对象
final HttpMethod method = HttpMethod.valueOf(request.getMethod().toString());
final String url = requestUrl.toString();
HttpHeaders filtered = filterRequest(this.headersFilters.getIfAvailable(), exchange);
final DefaultHttpHeaders httpHeaders = new DefaultHttpHeaders();
filtered.forEach(httpHeaders::set);
String transferEncoding = request.getHeaders().getFirst(HttpHeaders.TRANSFER_ENCODING);
boolean chunkedTransfer = “chunked“.equalsIgnoreCase(transferEncoding);
boolean preserveHost = exchange.getAttributeOrDefault(PRESERVE_HOST_HEADER_ATTRIBUTE, false);
// 请求后端服务
returnthis.httpClient.request(method, url, req -> {
final HttpClientRequest proxyRequest = req.options(NettyPipeline.SendOptions::flushOnEach)
.headers(httpHeaders)
.chunkedTransfer(chunkedTransfer)
.failOnServerError(false)
.failOnClientError(false);
if (preserveHost) {
String host = request.getHeaders().getFirst(HttpHeaders.HOST);
proxyRequest.header(HttpHeaders.HOST, host);
}
return proxyRequest.sendHeaders() //发送请求头
.send(request.getBody().map(dataBuffer -> // 发送请求Body
((NettyDataBuffer)dataBuffer).getNativeBuffer()));
}).doOnNext(res -> {
ServerHttpResponse response = exchange.getResponse();
// put headers and status so filters can modify the response
HttpHeaders headers = new HttpHeaders();
res.responseHeaders().forEach(entry -> headers.add(entry.getKey(), entry.getValue()));
exchange.getAttributes().put(“original_response_content_type“, headers.getContentType());
HttpHeaders filteredResponseHeaders = HttpHeadersFilter.filter(
this.headersFilters.getIfAvailable(), headers, exchange, Type.RESPONSE);
response.getHeaders().putAll(filteredResponseHeaders);
HttpStatus status = HttpStatus.resolve(res.status().code());
if (status != null) {
response.setStatusCode(status);
} elseif (response instanceof AbstractServerHttpResponse) {
// https://jira.spring.io/browse/SPR-16748
((AbstractServerHttpResponse) response).setStatusCodeValue(res.status().code());
} else {
thrownew IllegalStateException(“Unable to set status code on response: ” +res.status().code()+”, “+response.getClass());
}
// Defer committing the response until all route filters have run
// Put client response as ServerWebExchange attribute and write response later NettyWriteResponseFilter
exchange.getAttributes().put(CLIENT_RESPONSE_ATTR, res);
}).then(chain.filter(exchange));
}
3.1.7 Netty 回写响应网关过滤器
org.springframework.cloud.gateway.filter.NettyWriteResponseFilter#filter NettyWriteResponseFilter 与NettyRoutingFilter成对出现，负责将代理响应写回网关客户端响应。
public Mono filter(ServerWebExchange exchange, GatewayFilterChain chain) {
// then方法实现After Filter逻辑
return chain.filter(exchange).then(Mono.defer(() -> {
// 获得Netty Response
HttpClientResponse clientResponse = exchange.getAttribute(CLIENT_RESPONSE_ATTR);
if (clientResponse == null) {
return Mono.empty();
}
log.trace(“NettyWriteResponseFilter start“);
ServerHttpResponse response = exchange.getResponse();
// 将Netty Response回写给客户端
NettyDataBufferFactory factory = (NettyDataBufferFactory) response.bufferFactory();
//TODO: what if it’s not netty
final Flux body = clientResponse.receive()
.retain() // ByteBufFlux => ByteBufFlux
.map(factory::wrap); // ByteBufFlux => Flux
MediaType contentType = response.getHeaders().getContentType();
return (isStreamingMediaType(contentType) ?
response.writeAndFlushWith(body.map(Flux::just)) :
response.writeWith(body));
}));
}
4. Spring Cloud Gateway进行监控
当最终将网关也进行监控，可以全局的看到应用请求流转，网关的请求业务分流，各应用的调用负载情况。
点击查看原文
或扫描下方的微信公众号二维码查询
