Low concurrent volume of HLS streaming

[liyoung1992]

Description'

Please ensure that the markdown structure is maintained.

The concurrent number of HLS streaming is low, reaching around 1000, while FLV can reach around 5000.

1. SRS version: SRS/3.0.169(OuXuli)'

Please ensure that the markdown structure is maintained.

1. The configuration of SRS is as follows (Config):

Please ensure that the markdown structure is maintained.

master:

listen              1935;
max_connections     1000;
pid                 ./objs/master.pid
srs_log_tank        file;
srs_log_file        ./objs/master.log;
daemon              on;
http_api {
    enabled         on;
    listen          1985;
}
http_server {
    enabled         on;
    listen          8180;
    dir             ./objs/nginx/html;
}

vhost __defaultVhost__ {
> # Low latency configuration------------------

Please ensure that the markdown structure is maintained.
   # tcp_nodelay on;
   # min_latency on;
    #play {
    #    gop_cache       off;
    #    queue_length    10;
    #    mw_latency      100;
    #}
    #publish {
    #    mr off;
    #}
    #------------------------
    forward {
        enabled on;
        destination 10.206.0.*:1935  10.206.0.*:1936 10.206.0.*:1937;
    }
> # HTTP callback.

Please ensure that the markdown structure is maintained.
    http_hooks {
      enabled off;
     #on_publish  http://172.31.160.*:9999/v1/srs_callback;
    }
    hls {
        enabled         on;
>         hls_path        ./objs/nginx/html;# Slice address, can use srs or nginx as proxy
        hls_fragment    2;# Size of each fragment
        hls_window      20;# HLS window

Please ensure that the markdown structure is maintained.
        hls_wait_keyframe  on;
    }
}

slave:

listen              1935;
max_connections     3000;
pid                 ./objs/slave.pid;
#pid                 objs/edge.pid;
srs_log_file        ./objs/slave.log;
http_api {
    enabled         on;
    listen          1997;
}

http_server {
    enabled         on;
    listen          8080;
    dir             ./objs/nginx/html;
}

stats {
    network         0;
    disk            sda sdb xvda xvdb;
}

vhost __defaultVhost__ {
    #tcp_nodelay on;
    #min_latency on;
   # play {
       # gop_cache       off;
      #  queue_length    10;
     #   mw_latency      100;
        #time_jitter     off;
    #} 
    #publish {
    #    mr off;
    #}
    http_remux{
        enabled     on;
        mount       [vhost]/[app]/[stream].flv;
        hstrs       on;
    }

    hls{
        enabled         on;
        hls_path        ./objs/nginx/html;
        hls_fragment    10;
        hls_window      60;
        hls_wait_keyframe  on;
        }
}

Expected behavior (Expect)

Please ensure that the markdown structure is maintained.

> HLS concurrency is greatly affected by IO. Are there any good methods to solve this problem?

Please ensure that the markdown structure is maintained.

TRANS_BY_GPT3

[winlinvip]

Currently, the performance of HTTP has not been optimized. It is mainly used for demos and can be hosted on Nginx or a web server.

In the future, the capabilities of the HTTP server will be enhanced, referring to #1657.

TRANS_BY_GPT3