From: Joseph McDonald <joe@...>
Date: 2001-06-22T12:12:33+09:00
Subject: [ruby-talk:16739] Re: socket bug


Thursday, June 21, 2001, 7:41:05 PM, you wrote:

DT> You're running out of FD's in the server process. You need to close the
DT> new socket in the parent after you fork:

DT>  s = TCPServer.new("0.0.0.0",6666)
DT>  while(new_sock = s.accept)
DT>    pid = fork
DT>    if(pid == nil)
DT>      line = new_sock.gets
DT>      new_sock.print("got #{line}\n")
DT>      new_sock.shutdown
DT>      exit
DT>    end
DT>    new_sock.close        #<<<<<<<<<<<<<<
DT>  end
DT>  s.close

Thanks Dave!

I added that and also added a close to the client side, but I still
get this error:

server:
./socketbug.rb:6:in `wait': No child processes (Errno::ECHILD)
        from ./socketbug.rb:6
        from ./socketbug.rb:5:in `call'
        from ./socketbug.rb:11

It seems to happen randomly, but maybe every 500 or so requests?
Any idea what causes it?

And then the client:

It seems to really slow down from time to time (it could be
the server that is hanging of course...) The hang time is sometimes up
to 10 seconds between requests.  I also get the:
./client.rb:7:in `open': No buffer space available - "socket(2)" (Errno::ENOBUFS)
which I don't understand, is "w1" still hanging around after the
close?  Here is the new client code.  As you can see, I just added the
w1.close

require 'socket'

1.upto(100000) do |x|
  puts x
  w1 = TCPsocket.open("127.0.0.1", 6666)
  w1.print "hello server\n"
  ans1 = w1.readlines.to_s
  w1.shutdown  ### IS THIS NECESSARY?
  w1.close
end

I haven't seen any more segfaults out of it though, which is a GOOD
THING.

thanks,
-joe