[prev in list] [next in list] [prev in thread] [next in thread] 

List:       sylpheed
Subject:    [sylpheed:25634] Re: Version 2.1.1 "CLOSE_WAIT" socket state problem under Win32
From:       Hiroyuki Yamamoto <hiro-y () kcn ! ne ! jp>
Date:       2005-08-26 8:30:44
Message-ID: 20050826173044.b0f4d8a3.hiro-y () kcn ! ne ! jp
[Download RAW message or body]

Hello,

On Thu, 25 Aug 2005 17:52:02 -0400
Casco Bucci <Casco.Bucci@comcast.net> wrote:

> Version 2.1.1 has been working without incident on a WinXP Pro SP1
> box since it was built after its release a few days ago.
> 
> While checking tcp sockets with the most excellent TCPView utility
> (<http://www.sysinternals.com/Utilities/TcpView.html>) discovered
> hundreds of "CLOSE_WAIT" socket state entries for Sylpheed.
> 
> They all went away upon closing Sylpheed.  Unfortunately, Sylpheed
> does it all over again after it is restarted.
> 
> From what I understand, "CLOSE_WAIT" means that the connection is
> over and Windows is waiting for Sylpheed to close the socket.
> 
> Although the multiple "CLOSE_WAIT" socket states do not seem to be
> affecting system stability or performance, I can envision Sylpheed
> keeping thousands of sockets in the "CLOSE_WAIT" state which cannot
> be healthy for any system.

Fixed it in the svn trunk.

Note: Sylpheed hadn't closed sockets properly because it called close()
for socket fds. In WinSock, closesocket() must be used for sockets.
So I've modified sock_close() to use g_io_channel_shutdown() to close
sockets.

Thanks for pointing it out.

-- 
Hiroyuki Yamamoto <hiro-y@kcn.ne.jp>

[prev in list] [next in list] [prev in thread] [next in thread]