Mailing-List: contact cygwin-developers-help AT sourceware DOT cygnus DOT com; run by ezmlm List-Subscribe: List-Archive: List-Post: List-Help: , Sender: cygwin-developers-owner AT sources DOT redhat DOT com Delivered-To: mailing list cygwin-developers AT sources DOT redhat DOT com Message-ID: <059601c16e21$f088d540$0200a8c0@lifelesswks> From: "Robert Collins" To: "Jonathan Kamens" Cc: References: <20011115213959 DOT 6226 DOT qmail AT lizard DOT curl DOT com> <052f01c16e1f$ffb61890$0200a8c0 AT lifelesswks> <20011115215733 DOT 6471 DOT qmail AT lizard DOT curl DOT com> Subject: Re: Fixing openssh to avoid occasional spurious connection failures Date: Fri, 16 Nov 2001 09:07:38 +1100 MIME-Version: 1.0 Content-Type: text/plain; charset="Windows-1252" Content-Transfer-Encoding: 7bit X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 6.00.2600.0000 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000 X-OriginalArrivalTime: 15 Nov 2001 22:05:47.0829 (UTC) FILETIME=[AE2DAE50:01C16E21] BTW: this is off topic for cygwin-apps. It's either a cygwin AT cygwin DOT com or a cygwin-dev discussion.(it's not about how to package openssh ). ----- Original Message ----- From: "Jonathan Kamens" To: Cc: Sent: Friday, November 16, 2001 8:57 AM Subject: Re: Fixing openssh to avoid occasional spurious connection failures > > From: "Robert Collins" > > Date: Fri, 16 Nov 2001 08:51:47 +1100 > > > > And what about patching cygwin! to automatically retry on EADDRINUSE - > > if we can detect that the winsock bug is the culprit? That would fix > > ssh, squid, telnet, ftp.... > > As I noted in my message to cygwin-developers, I think this is too > hard to patch inside Cygwin. The problem is that once the failure > occurs, the socket is bound to a particular local port and can't be > rebound to another port, so the socket with the failure is toast. > This means that you'd have to burn down the socket and reconstruct a > new one to retry the failed connection, without the caller ever > noticing. I don't think it's practical to try to do that. Which socket is bound. The cygwin behind-the-scenes win32 socket, or the openssh/l socket that cygwin manages? Anyway, I've no intention to try patching this myself, but it does seem like something that should really be fixed in cygwin, not in n different client applications. Rob