Date: Wed, 21 Sep 2005 12:55:53 +0100 (BST) From: Robert Watson <rwatson@FreeBSD.org> To: Andre Oppermann <andre@freebsd.org> Cc: freebsd-net@freebsd.org, Sten Daniel =?iso-8859-1?Q?S=F8rsdal?= <lists@wm-access.no>, Milscvaer <millueradfa@yahoo.com> Subject: Re: UDP dont fragment bit Message-ID: <20050921125325.E34322@fledge.watson.org> In-Reply-To: <43314916.2451ED6A@freebsd.org> References: <20050918212110.61962.qmail@web54501.mail.yahoo.com> <20050920134408.Y34322@fledge.watson.org> <43313924.9050009@wm-access.no> <20050921114511.D34322@fledge.watson.org> <43314916.2451ED6A@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, 21 Sep 2005, Andre Oppermann wrote: > I can think of a couple of uses to say IP DF on a UDP socket. Will cook > up a patch to add such a sysctl in a few hours. The problem is that I think this solves only half of the likely problem. If what application developers really want is a way to do PMTU for UDP-based applications, the other half has to be done too: the ability to receive and process MTU data on the socket. Given that UDP sockets are often used unconnected, this means providing at least two additional pieces of information for ICMP MTU events: host/path information, and the MTU cap reported. And as UDP sockets are often used for quite a bit of traffic to different hosts at once, we might want to find a way to do this under load with an event stream rather than a condition queried using a simple socket option. So I think learning a bit more about the specific applications would be quite helpful -- in particular, what ICMP/MTU information they want, and how they will use it. Robert N M Watson
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20050921125325.E34322>