Skip to content

我在preread_by_lua_file 阶段进行 sock 的 peek 操作,我想优雅的获取缓冲区的所有数据该怎么实现 #365

Open

Description

@wjsw1781
local peek_num=1
  while true do
      local chunk, err =sock:peek(peek_num)
      ngx.log(ngx.ERR,"peek ..." ,chunk, ' == ',peek_num ,err )
      -- ngx.log(ngx.ERR,"peek ..." ,#chunk, ' == ',peek_num ,err,bytes_to_hex(chunk) )
      peek_num=peek_num+1
  end
 ngx.log(ngx.ERR,"退出循环 ..."  )

这样会卡死在最后一个缓冲区的下一个

但是当我添加了 preread_timeout 1s;


stream {
    preread_timeout 1s;

    server {
        listen 0.0.0.0:3889;  #监听本机地址和端口,当使用keeplived的情况下使用keeplived VIP
        preread_by_lua_file /root/socks_ss_gfw_ss_socks/nginx_redis_lua/test_socks.lua;
        proxy_pass backend; #
    }
    upstream backend{  
        server 127.0.0.2:1101;  # 爱写啥写啥  反正下面的代码也给你改了
        balancer_by_lua_file /root/socks_ss_gfw_ss_socks/nginx_redis_lua/balancer_by_lua_file.lua;
    } 

}
此时

        local chunk, err =pcall(function() return sock:peek(peek_num) end) 


这里的 pcall 还是捕获不到这种错误

我不信想使用 recive 这种消耗配合字节逐个解析的方法 可能还得配合超时来获取字节总是很浪费性能 难道就没有一个方法可以从缓冲区获取一次 tcp 交流的所有数据的方法?

Activity

  1. aifeiasdf commented on Mar 24, 2025

    @aifeiasdf
    local peek_num=1
      while true do
          local chunk, err =sock:peek(peek_num)
          ngx.log(ngx.ERR,"peek ..." ,chunk, ' == ',peek_num ,err )
          -- ngx.log(ngx.ERR,"peek ..." ,#chunk, ' == ',peek_num ,err,bytes_to_hex(chunk) )
          peek_num=peek_num+1
      end
     ngx.log(ngx.ERR,"退出循环 ..."  )
    

    这样会卡死在最后一个缓冲区的下一个

    但是当我添加了 preread_timeout 1s;

    
    stream {
        preread_timeout 1s;
    
        server {
            listen 0.0.0.0:3889;  #监听本机地址和端口,当使用keeplived的情况下使用keeplived VIP
            preread_by_lua_file /root/socks_ss_gfw_ss_socks/nginx_redis_lua/test_socks.lua;
            proxy_pass backend; #
        }
        upstream backend{  
            server 127.0.0.2:1101;  # 爱写啥写啥  反正下面的代码也给你改了
            balancer_by_lua_file /root/socks_ss_gfw_ss_socks/nginx_redis_lua/balancer_by_lua_file.lua;
        } 
    
    }
    此时
    
            local chunk, err =pcall(function() return sock:peek(peek_num) end) 
    

    这里的 pcall 还是捕获不到这种错误

    我不信想使用 recive 这种消耗配合字节逐个解析的方法 可能还得配合超时来获取字节总是很浪费性能 难道就没有一个方法可以从缓冲区获取一次 tcp 交流的所有数据的方法?

    "所有"这个概念不是很清晰,read函数不知道你缓冲区有多少数据,如果你想实现的是读多少处理多少最好的办法就是在write_filter_module的前面加一个c模块

  2. ZigHuang commented on Mar 26, 2025

    @ZigHuang

    I've encountered the same issue as you. After looking through online resources, I found that the njs module can be used to detecte HTTP request headers on stream module. And it also solved my issue.

    NJS-example

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions