Article delegate-en/239 of [1-5169] on the server localhost:119
  upper oldest olders older1 this newer1 newers latest
[Top/Up] [oldest] - [Older+chunk] - [Newer+chunk] - [newest + Check]
Newsgroups: mail-lists.delegate-en

[DeleGate-En] Re:
22 Jan 1999 22:00:32 GMT (Yutaka Sato)

In message <_A237@delegate-en.ML_> on 01/23/99(03:52:31)
you Hiroshi Ueno <> wrote:
 |> This appears to verify a problem with sockets being left in an state that
 |> prevents reuse without shutting the system down.  Please see if you can
 |> recreate this on your system with TCP/IP 4.1 at the latest level and
 |> correct this condition.  It is not acceptable to need to shut the server
 |> down any time DeleGate needs to be restarted.
 |I tried TCP/IP v4.1+DeleGate combination on my Warp4 machine but I
 |could not recreate the problem. That is mainly because, on my
 |machine, any socket was not beeing left in used state. Netstat -l
 |command listed no sockets after terminating delegated.exe.
 |However, my understanding is that delegated issues setsockopt()
 |for every socket to set SO_REUSEADDR in order for that socket to
 |reuse the local port addresses. Sato-san, is this correct?

Yes.  It is done in bind_inet() function in inets.c file, at line
#1803 as:


Although I don't understand what is happening on the OS/2 environment,
I suppose using SO_REUSEPORT instead of SO_REUSEADDR will change the
situation, if SO_REUSEPORT is supported in OS/2.
But of course the problem is why a socket (bound by bind()) is left
in used state (after SIGTERM?) in some OS/2 environmnet, and I have
no idea on it.

Yutaka Sato <>   @ @ 
Computer Science Division, Electrotechnical Laboratory      ( - )
1-1-4 Umezono, Tsukuba, Ibaraki, 305-8568 Japan            _<   >_

  admin search upper oldest olders older1 this newer1 newers latest
[Top/Up] [oldest] - [Older+chunk] - [Newer+chunk] - [newest + Check]