From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mailman by lists.gnu.org with tmda-scanned (Exim 4.43) id 1LzvEV-0005VN-Dx for qemu-devel@nongnu.org; Fri, 01 May 2009 12:05:03 -0400 Received: from exim by lists.gnu.org with spam-scanned (Exim 4.43) id 1LzvEP-0005T9-Sx for qemu-devel@nongnu.org; Fri, 01 May 2009 12:05:02 -0400 Received: from [199.232.76.173] (port=34424 helo=monty-python.gnu.org) by lists.gnu.org with esmtp (Exim 4.43) id 1LzvEP-0005T3-P2 for qemu-devel@nongnu.org; Fri, 01 May 2009 12:04:57 -0400 Received: from mx20.gnu.org ([199.232.41.8]:14945) by monty-python.gnu.org with esmtps (TLS-1.0:RSA_AES_256_CBC_SHA1:32) (Exim 4.60) (envelope-from ) id 1LzvEP-0006nn-AJ for qemu-devel@nongnu.org; Fri, 01 May 2009 12:04:57 -0400 Received: from mail.codesourcery.com ([65.74.133.4]) by mx20.gnu.org with esmtp (Exim 4.60) (envelope-from ) id 1LzvEM-0005vj-CY for qemu-devel@nongnu.org; Fri, 01 May 2009 12:04:54 -0400 From: Paul Brook Subject: Re: [Qemu-devel] PATCH: enabling TCP keepalives - v3 Date: Fri, 1 May 2009 17:04:47 +0100 References: <49F9FEBA.6050901@gmail.com> <200905011616.13164.paul@codesourcery.com> <49FB1BE4.30207@us.ibm.com> In-Reply-To: <49FB1BE4.30207@us.ibm.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200905011704.48102.paul@codesourcery.com> List-Id: qemu-devel.nongnu.org List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-devel@nongnu.org Cc: Anthony Liguori , David Ahern > I don't think an option is very useful. If you lose your connection and > now have a bunch of orphan connections, it's too late to have the > foresight to have used an option on startup. In that case, it would be > much better to just be able to close existing connections. > > I think there's a very, very small number of people that would have the > foresight to always use keepalive=60 or whatever the option would be. In that case we should do nothing. Users with a flakey net connection can tweak their kernel (via /proc/sys) to use lower timeouts. AFAICS There's no way qemu can know what a "better" value is. As a concrete example my ADSL connection takes between 90 and 180 seconds to resync after a noise spike, so 60 seconds is absolutely the wrong timeout value here. Paul