Nginx 中的超时和“实体太大”

Question

我有一个使用 Express 的 NodeJS 应用程序。此应用程序无需通过 Nginx 访问即可完美运行。此应用程序需要上传约 100Gb 的文件。如果我通过 Nginx 访问此应用程序，则会发生以下情况：

413 Request Entity Too Large

查看日志时：

2016/01/26 11:09:07 [error] 21508#0: *1 upstream timed out (110: Connection timed out) while sending request to upstream, client: 10.9.56.33, server: 161.116.xxx.xx, request: "POST /metagenomics/pipeline/ HTTP/1.1", upstream: "http://127.0.0.1:8080/metagenomics/pipeline/", host: "10.4.2.11", referrer: "http://10.4.2.11/metagenomics/"
2016/01/26 11:09:07 [error] 21508#0: *1 client intended to send too large body: 32421677970 bytes, client: 10.9.56.33, server: 161.116.xxx.xx, request: "POST /metagenomics/pipeline/ HTTP/1.1", upstream: "http://127.0.0.1:8080/metagenomics/pipeline/", host: "10.4.2.11", referrer: "http://10.4.2.11/metagenomics/"

因此，我在上传大文件时遇到超时和问题。我的 nginx 配置是：

upstream metagenomics {
  server localhost:8080;
}

# HTTP server
server {

  listen                  80;
  server_name             161.116.xxx.xx;
  server_name_in_redirect off;

  access_log          /var/log/nginx/metagenomics-80.access.log;
  error_log           /var/log/nginx/metagenomics-80.error.log;

  # Use utf-8
  charset             utf-8;
  source_charset      utf-8;

  # output compression saves bandwidth
  gzip            on;
  gzip_http_version 1.0;
  gzip_comp_level 2;
  gzip_proxied any;
  gzip_types      text/plain text/html text/css application/x-javascript text/xml application/xml application/xml+rss text/javascript;

  # Default root
  root                        /usr/share/nginx/html/;

  # Default index
  index                       index.html;

  error_page 404 /404.html;
    location = /40x.html {
      root                    /usr/share/nginx/html/;
  }

  error_page 500 502 503 504 /50x.html;
    location = /50x.html {
      root                    /usr/share/nginx/html/;
  }
 }

  location /metagenomics/ {
    proxy_pass http://localhost:8080/metagenomics/;
    root /Synology/server_Metagenomics/server/;
    index index.html;
  }

  location /metagenomics/pipeline/ {
    proxy_pass http://localhost:8080/metagenomics/pipeline/;

    ##managing file issues
    client_max_body_size 100G;
    client_body_temp_path /Synology/server_Metagenomics/server/files/;
    #proxy_read_timeout 5s;
    #proxy_send_timeout 5s;
    #proxy_next_upstream_timeout 5s;
  }

}

这是以接受大文件请求的方式配置的。发生了什么？