PrematureCloseException：连接过早关闭

问题描述 投票：0回答：2

我正在使用 Web-flux/Reactive 和 Webclient，在 Tomcat 和 spring-boot 上运行它。

一切正常。我读了很多相关内容。问题似乎是，每当你使用Webclient时，你都必须返回或使用响应，否则它会关闭连接而你还没有消费它，并且你会看到很多日志消息说

the connection close prematurely
，如果我有一个 404 状态代码是错误的场景，我可以使用 
OnStatus
并抛出异常，但我的场景是：当上游服务返回 404 时，我必须手动返回一个 mono 空。所以我不使用 Webclient 请求的响应，我只是使用 
ClientResponse
中的 
.exchange()
来检查状态并处理它。我最初的问题是日志消息，因为它只是“垃圾”，您不想在日志消息中看到很多内容。我在某处读到，如果发生这种情况，连接也无法重新使用，所以听起来真的很糟糕，但我不知道......而且我只是在找不到它时收到此消息，如果响应是 200 它返回对象并且不打印日志消息。

我尝试使用

clientResponse.BodyToMono(Void.Class)
，但它也不起作用。日志消息不断出现。

 @Bean
  public WebClient webClient(
      @Value("${http.client.connection-timeout-millis}") final int connectionTimeoutMillis,
      @Value("${http.client.socket-timeout-millis}") final int socketTimeoutMillis,
      @Value("${http.client.wire-tap-enabled}") final boolean wireTapEnabled,
      final ObjectMapper objectMapper) {

    Consumer<Connection> doOnConnectedConsumer = connection ->
        connection
            .addHandler(new ReadTimeoutHandler(socketTimeoutMillis, MILLISECONDS))
            .addHandler(new WriteTimeoutHandler(connectionTimeoutMillis, MILLISECONDS));

    TcpClient tcpClient = TcpClient.newConnection()
        .wiretap(wireTapEnabled)
        .option(CONNECT_TIMEOUT_MILLIS, connectionTimeoutMillis)
        .doOnConnected(doOnConnectedConsumer);

    return WebClient.builder()
        .clientConnector(new ReactorClientHttpConnector(HttpClient.from(tcpClient).compress(true)))
        .exchangeStrategies(customExchangeStrategies(objectMapper))
        .build();
  } 

// ..........

    MultiValueMap<String, String> params = getParams(t1, t2);

    return webClient.get()
        .uri(HttpUtils.buildUrl(serviceUrl, params, name))
        .exchange()
        .flatMap(this::handleClientResponse)
        .onErrorMap(Exception.class, ex -> handleUnexpectedEx(ex, name, params));
  }

日志条目

2019-07-08 11:56:51.972  WARN [-,,,] 1504 --- [ctor-http-nio-3] reactor.netty.channel.FluxReceive        : [id: 0x66c8568c, L:/127.0.0.1:62319 ! R:localhost/127.0.0.1:8990] An exception has been observed post termination

reactor.netty.http.client.PrematureCloseException: Connection prematurely closed DURING response

2019-07-08 11:56:52.013 DEBUG [-,,,] 1504 --- [ctor-http-nio-2] reactor.netty.ReactorNetty               : [id: 0xf50bdf8d, L:/127.0.0.1:62324 ! R:localhost/127.0.0.1:8990] Non Removed handler: ReadTimeoutHandler, context: ChannelHandlerContext(ReadTimeoutHandler, [id: 0xf50bdf8d, L:/127.0.0.1:62324 ! R:localhost/127.0.0.1:8990]), pipeline: DefaultChannelPipeline{(reactor.left.httpCodec = io.netty.handler.codec.http.HttpClientCodec), (reactor.left.decompressor = io.netty.handler.codec.http.HttpContentDecompressor), (WriteTimeoutHandler = io.netty.handler.timeout.WriteTimeoutHandler), (ReadTimeoutHandler = io.netty.handler.timeout.ReadTimeoutHandler), (reactor.right.reactiveBridge = reactor.netty.channel.ChannelOperationsHandler)}
2019-07-08 11:56:52.014 DEBUG [-,,,] 1504 --- [ctor-http-nio-2] reactor.netty.ReactorNetty               : [id: 0xf50bdf8d, L:/127.0.0.1:62324 ! R:localhost/127.0.0.1:8990] Non Removed handler: WriteTimeoutHandler, context: ChannelHandlerContext(WriteTimeoutHandler, [id: 0xf50bdf8d, L:/127.0.0.1:62324 ! R:localhost/127.0.0.1:8990]), pipeline: DefaultChannelPipeline{(reactor.left.httpCodec = io.netty.handler.codec.http.HttpClientCodec), (reactor.left.decompressor = io.netty.handler.codec.http.HttpContentDecompressor), (WriteTimeoutHandler = io.netty.handler.timeout.WriteTimeoutHandler), (ReadTimeoutHandler = io.netty.handler.timeout.ReadTimeoutHandler), (reactor.right.reactiveBridge = reactor.netty.channel.ChannelOperationsHandler)}
2019-07-08 11:56:52.014 DEBUG [-,,,] 1504 --- [ctor-http-nio-2] r.netty.resources.NewConnectionProvider  : [id: 0xf50bdf8d, L:/127.0.0.1:62324 ! R:localhost/127.0.0.1:8990] onStateChange([response_incomplete], GET{uri=/service/TWFDHF?T1=1.0.0&T2=1, connection=SimpleConnection{channel=[id: 0xf50bdf8d, L:/127.0.0.1:62324 ! R:localhost/127.0.0.1:8990]}})
2019-07-08 11:56:52.014  WARN [-,,,] 1504 --- [ctor-http-nio-2] reactor.netty.channel.FluxReceive        : [id: 0xf50bdf8d, L:/127.0.0.1:62324 ! R:localhost/127.0.0.1:8990] An exception has been observed post termination

启用窃听的日志条目

2019-07-10 14:51:19.295 DEBUG [-,,,] 2940 --- [ctor-http-nio-2] reactor.netty.tcp.TcpClient              : [id: 0x677da0d4, L:/127.0.0.1:62385 ! R:localhost/127.0.0.1:8990] UNREGISTERED
2019-07-10 14:51:19.541 DEBUG [-,,,] 2940 --- [ctor-http-nio-4] reactor.netty.tcp.TcpClient              : [id: 0xa7f41d1e, L:/127.0.0.1:62384 - R:localhost/127.0.0.1:8990] CLOSE
2019-07-10 14:51:19.542 DEBUG [-,,,] 2940 --- [ctor-http-nio-4] r.n.resources.PooledConnectionProvider   : [id: 0xa7f41d1e, L:/127.0.0.1:62384 ! R:localhost/127.0.0.1:8990] Channel cleaned, now 0 active connections and 9 inactive connections
2019-07-10 14:51:19.542 DEBUG [-,,,] 2940 --- [ctor-http-nio-4] r.n.resources.PooledConnectionProvider   : [id: 0xa7f41d1e, L:/127.0.0.1:62384 ! R:localhost/127.0.0.1:8990] Channel closed, now 0 active connections and 8 inactive connections
2019-07-10 14:51:19.542 DEBUG [-,,,] 2940 --- [ctor-http-nio-4] reactor.netty.ReactorNetty               : [id: 0xa7f41d1e, L:/127.0.0.1:62384 ! R:localhost/127.0.0.1:8990] Non Removed handler: ReadTimeoutHandler, context: ChannelHandlerContext(ReadTimeoutHandler, [id: 0xa7f41d1e, L:/127.0.0.1:62384 ! R:localhost/127.0.0.1:8990]), pipeline: DefaultChannelPipeline{(reactor.left.loggingHandler = io.netty.handler.logging.LoggingHandler), (reactor.left.httpCodec = io.netty.handler.codec.http.HttpClientCodec), (WriteTimeoutHandler = io.netty.handler.timeout.WriteTimeoutHandler), (ReadTimeoutHandler = io.netty.handler.timeout.ReadTimeoutHandler), (reactor.right.reactiveBridge = reactor.netty.channel.ChannelOperationsHandler)}
2019-07-10 14:51:19.543 DEBUG [-,,,] 2940 --- [ctor-http-nio-4] reactor.netty.ReactorNetty               : [id: 0xa7f41d1e, L:/127.0.0.1:62384 ! R:localhost/127.0.0.1:8990] Non Removed handler: WriteTimeoutHandler, context: ChannelHandlerContext(WriteTimeoutHandler, [id: 0xa7f41d1e, L:/127.0.0.1:62384 ! R:localhost/127.0.0.1:8990]), pipeline: DefaultChannelPipeline{(reactor.left.loggingHandler = io.netty.handler.logging.LoggingHandler), (reactor.left.httpCodec = io.netty.handler.codec.http.HttpClientCodec), (WriteTimeoutHandler = io.netty.handler.timeout.WriteTimeoutHandler), (ReadTimeoutHandler = io.netty.handler.timeout.ReadTimeoutHandler), (reactor.right.reactiveBridge = reactor.netty.channel.ChannelOperationsHandler)}
2019-07-10 14:51:19.543 DEBUG [-,,,] 2940 --- [ctor-http-nio-4] reactor.netty.tcp.TcpClient              : [id: 0xa7f41d1e, L:/127.0.0.1:62384 ! R:localhost/127.0.0.1:8990] INACTIVE
2019-07-10 14:51:19.544 DEBUG [-,,,] 2940 --- [ctor-http-nio-4] r.n.resources.PooledConnectionProvider   : [id: 0xa7f41d1e, L:/127.0.0.1:62384 ! R:localhost/127.0.0.1:8990] onStateChange(GET{uri=/dataviews/TWN_EMPLBENINFO_FIXED?version=1.0.0&sequence=1, connection=PooledConnection{channel=[id: 0xa7f41d1e, L:/127.0.0.1:62384 ! R:localhost/127.0.0.1:8990]}}, [response_incomplete])
2019-07-10 14:51:19.544  WARN [-,,,] 2940 --- [ctor-http-nio-4] reactor.netty.channel.FluxReceive        : [id: 0xa7f41d1e, L:/127.0.0.1:62384 ! R:localhost/127.0.0.1:8990] An exception has been observed post termination

和我的经纪人

protected Mono handleClientResponseError(final ClientResponse clientResponse) {
    clientResponse.bodyToMono(Void.class);
    Mono<ErrorResponse> errorResponse = clientResponse.body(BodyExtractors.toMono(ErrorResponse.class));

    return errorResponse.flatMap(err -> {
      log.debug("Received HttpStatusCodeException when calling {} Registry: {}", getGatewayName(),
          err.getErrorEnvelope().getMessage());

      return Mono.error(new UpStreamServiceHttpException(err, clientResponse.rawStatusCode()));
    }).switchIfEmpty(Mono.error(() -> {
      log.debug("Received HttpStatusCodeException when calling {} Registry: {}", getGatewayName());

      return new UpStreamServiceHttpException("Bad Gateway", clientResponse.rawStatusCode());
    }));
  }

我尝试使用

clientResponse.bodyToMono(Void.class);
来完成响应，但它仍然不起作用，我需要一种方法来完成Webclient的响应，然后能够重新使用连接池并摆脱那些日志消息。

我在 Github 上创建了一个问题，但它已关闭，所以我在这里创建这个问题。 https://github.com/spring-projects/spring-framework/issues/23249

spring netty spring-webflux reactive spring-webclient
2个回答
20
投票

所以我终于找到了一种方法来完成回复，我将在这里分享给遇到这个问题的人

在我拥有这个之前：

 private Mono<Optional<JsonNode>> handleHttpErrorStatus(final ClientResponse clientResponse) {
    if (clientResponse.statusCode().equals(HttpStatus.NOT_FOUND)) {
      clientResponse.bodyToMono(Void.class);
      return Mono.just(Optional.empty());
    } else {
      return handleClientResponseError(clientResponse);
    }
  }

但它不起作用，显然是因为我无论如何都没有归还它，所以在挖掘之后我尝试了这个：

private Mono<Optional<JsonNode>> handleHttpErrorStatus(final ClientResponse clientResponse) {
    if (clientResponse.statusCode().equals(HttpStatus.NOT_FOUND)) {
      return clientResponse.bodyToMono(Void.class).thenReturn((Optional.empty()));
    } else {
      return handleClientResponseError(clientResponse);
    }
  }

它工作正常，我检查了窃听日志，现在一切看起来都很好。

2019-07-10 16:44:39.096 DEBUG [-,,,] 16260 --- [ctor-http-nio-2] r.n.http.client.HttpClientOperations     : [id: 0x461d8170, L:/127.0.0.1:64298 - R:localhost/127.0.0.1:8990] Received response (auto-read:false) : [Content-Type=application/json;charset=UTF-8, Transfer-Encoding=chunked, Date=Wed, 10 Jul 2019 15:44:38 GMT]
2019-07-10 16:44:39.096 DEBUG [-,,,] 16260 --- [ctor-http-nio-2] r.n.resources.PooledConnectionProvider   : [id: 0x461d8170, L:/127.0.0.1:64298 - R:localhost/127.0.0.1:8990] onStateChange(GET{uri=/dataviews/TWN_EMPLPIM_AVRO?version=1.0.0&sequence=1, connection=PooledConnection{channel=[id: 0x461d8170, L:/127.0.0.1:64298 - R:localhost/127.0.0.1:8990]}}, [response_received])
2019-07-10 16:44:39.098 DEBUG [-,bdd6433efa276f10,bdd6433efa276f10,false] 16260 --- [ctor-http-nio-2] reactor.netty.channel.FluxReceive        : [id: 0x461d8170, L:/127.0.0.1:64298 - R:localhost/127.0.0.1:8990] Subscribing inbound receiver [pending: 0, cancelled:false, inboundDone: false]
2019-07-10 16:44:39.099 DEBUG [-,bdd6433efa276f10,bdd6433efa276f10,false] 16260 --- [ctor-http-nio-2] reactor.netty.tcp.TcpClient              : [id: 0x461d8170, L:/127.0.0.1:64298 - R:localhost/127.0.0.1:8990] CLOSE
2019-07-10 16:44:39.099 DEBUG [-,bdd6433efa276f10,bdd6433efa276f10,false] 16260 --- [ctor-http-nio-2] r.n.resources.PooledConnectionProvider   : [id: 0x461d8170, L:/127.0.0.1:64298 ! R:localhost/127.0.0.1:8990] Channel cleaned, now 1 active connections and 1 inactive connections
2019-07-10 16:44:39.099 DEBUG [-,bdd6433efa276f10,bdd6433efa276f10,false] 16260 --- [ctor-http-nio-2] r.n.resources.PooledConnectionProvider   : [id: 0x461d8170, L:/127.0.0.1:64298 ! R:localhost/127.0.0.1:8990] Channel closed, now 1 active connections and 0 inactive connections
2019-07-10 16:44:39.099 DEBUG [-,bdd6433efa276f10,bdd6433efa276f10,false] 16260 --- [ctor-http-nio-2] reactor.netty.ReactorNetty               : [id: 0x461d8170, L:/127.0.0.1:64298 ! R:localhost/127.0.0.1:8990] Non Removed handler: ReadTimeoutHandler, context: ChannelHandlerContext(ReadTimeoutHandler, [id: 0x461d8170, L:/127.0.0.1:64298 ! R:localhost/127.0.0.1:8990]), pipeline: DefaultChannelPipeline{(reactor.left.loggingHandler = io.netty.handler.logging.LoggingHandler), (reactor.left.httpCodec = io.netty.handler.codec.http.HttpClientCodec), (WriteTimeoutHandler = io.netty.handler.timeout.WriteTimeoutHandler), (ReadTimeoutHandler = io.netty.handler.timeout.ReadTimeoutHandler), (reactor.right.reactiveBridge = reactor.netty.channel.ChannelOperationsHandler)}
2019-07-10 16:44:39.099 DEBUG [-,bdd6433efa276f10,bdd6433efa276f10,false] 16260 --- [ctor-http-nio-2] reactor.netty.ReactorNetty               : [id: 0x461d8170, L:/127.0.0.1:64298 ! R:localhost/127.0.0.1:8990] Non Removed handler: WriteTimeoutHandler, context: ChannelHandlerContext(WriteTimeoutHandler, [id: 0x461d8170, L:/127.0.0.1:64298 ! R:localhost/127.0.0.1:8990]), pipeline: DefaultChannelPipeline{(reactor.left.loggingHandler = io.netty.handler.logging.LoggingHandler), (reactor.left.httpCodec = io.netty.handler.codec.http.HttpClientCodec), (WriteTimeoutHandler = io.netty.handler.timeout.WriteTimeoutHandler), (ReadTimeoutHandler = io.netty.handler.timeout.ReadTimeoutHandler), (reactor.right.reactiveBridge = reactor.netty.channel.ChannelOperationsHandler)}
2019-07-10 16:44:39.100 DEBUG [-,,,] 16260 --- [ctor-http-nio-2] reactor.netty.tcp.TcpClient              : [id: 0x461d8170, L:/127.0.0.1:64298 ! R:localhost/127.0.0.1:8990] READ COMPLETE
2019-07-10 16:44:39.102 DEBUG [-,,,] 16260 --- [ctor-http-nio-2] reactor.netty.tcp.TcpClient              : [id: 0x461d8170, L:/127.0.0.1:64298 ! R:localhost/127.0.0.1:8990] INACTIVE
2019-07-10 16:44:39.102 DEBUG [-,,,] 16260 --- [ctor-http-nio-2] reactor.netty.tcp.TcpClient              : [id: 0x461d8170, L:/127.0.0.1:64298 ! R:localhost/127.0.0.1:8990] UNREGISTERED
2019-07-10 16:44:39.122 DEBUG [-,,,] 16260 --- [ctor-http-nio-2] reactor.netty.tcp.TcpClient              : [id: 0xbe2cb147, L:/127.0.0.1:64295 - R:localhost/127.0.0.1:8990] READ: 398B
1
投票

我遇到了同样的问题，我花了 5 天的时间一直在纠结这个问题，想知道在部署新功能后我到底做错了什么。我终于有了尤里卡时刻。

所以在我的 Dockerfile 中，我暴露了默认端口 8080。

这意味着我的应用程序映像中的 tomcat 服务器将始终在该端口上可用。我的容器还必须在端口 8080 上运行（在我的 .yml 文件内）

在应用端口映射指令将 Linux 主机上的外部端口 9003 映射到容器中的 8080 之前。

我错过的是，在使用 docker-compose 文件部署服务时，我忘记将 application.yml 文件中的 server.port 更改为 8080，以便进行端口映射。我将其保留为 9003 （因为我在本地计算机上使用 9003 来测试应用程序） 这意味着每当我使用 docker-compose 启动容器时，我的服务端口都会映射为

0.0.0.0:9003 -> 9003/tcp 而不是 0.0.0.0:9003 -> 8080/tcp。

这就是 api 网关不断抛出

的原因

PrematureCloseException：连接过早关闭

因为在我的微服务的容器实例中实际上没有找到可供 Linux 主机上的 9003 映射到的内部端口 9003。

我只需将 application.yml 中的 9003 更改为 8080 即可。

最新问题
© www.soinside.com 2019 - 2024. All rights reserved.