源码解读提高RGW并发数-fastcgi模式

用户1260683

发布于 2018-01-31 12:03:04

1.3K0

发布于 2018-01-31 12:03:04

以rgw服务的main()为入口，查看整个fastcgi的初始化过程，代码如下

#src/rgw/rgw_main.cc
int main(int argc, const char **argv)
    if (framework == "fastcgi" || framework == "fcgi") {
      RGWProcessEnv fcgi_pe = { store, &rest, olog, 0 };

      fe = new RGWFCGXFrontend(fcgi_pe, config);

    dout(0) << "starting handler: " << fiter->first << dendl;
    int r = fe->init(); #调用RGWFCGXFrontend的init()方法

再看init()方法构建了一个RGWFCGXProcess，并将rgw_thread_pool_size作为实参传递进去。

#src/rgw/rgw_frontend.h
class RGWFCGXFrontend : public RGWProcessFrontend {
public:
  RGWFCGXFrontend(RGWProcessEnv& pe, RGWFrontendConfig* _conf)
    : RGWProcessFrontend(pe, _conf) {}

  int init() {
    pprocess = new RGWFCGXProcess(g_ceph_context, &env,
                  g_conf->rgw_thread_pool_size, conf);
    return 0;
  }
};

默认rgw_thread_pool_size为100，代码定义如下

#src/common/config_opts.h
OPTION(rgw_thread_pool_size, OPT_INT, 100)

通过RGWFCGXProcess的构造函数发现max_connections=num_threads + (num_threads >> 3)，也就是说默认情况下max_connections=100+1=101，代码注释中也提到这是为了确保能够尽可能多的处理请求。

#src/rgw/rgw_process.h
class RGWFCGXProcess : public RGWProcess {
    int max_connections;
public:

  /* have a bit more connections than threads so that requests are
   * still accepted even if we're still processing older requests */
  RGWFCGXProcess(CephContext* cct, RGWProcessEnv* pe, int num_threads,
         RGWFrontendConfig* _conf)
    : RGWProcess(cct, pe, num_threads, _conf),
      max_connections(num_threads + (num_threads >> 3))
    {}

  void run();
  void handle_request(RGWRequest* req);
};

所以num_threads控制着max_connections的数量，如果你想提高单个rgw进程的最大并发数量，需要调高rgw_thread_pool_size。

本文参与腾讯云自媒体分享计划，分享自微信公众号。

原始发表：2017-06-23，如有侵权请联系 cloudcommunity@tencent.com 删除

fastcgi