On Wed, 2004-06-16 at 00:57, Bernd Eckenfels wrote: > Hello, > > I started that discussion on linux-kernel, but I think it is better here: <SNIP> > There are application problems, if the interface has (only) the link local > prefix or corrupt global prefixes. > > If the application prefers ipv6 it will try to use them, and only on connect > it receives an error. Some do then ipv4 fallback (lynx) some dont (mozilla). Fix your routing as that is the problem. Whenever a connect() fails because there is no route or the connection is refused the getaddrinfo() loop inside both those applications will try the next address, may that be IPv4 or IPv6. > So qestions: > > - is the kernel interpretation auf autoconf=0 or the documntation right? If you don't want IPv6 "rmmod ipv6", simple. > - how should an application avoid connecting via an interface which has > only addresses for the wrong scope? An application can't protect itself from misconfigurations, or do you want every application to do 'verifications' of the setup? :) > - is falling back from v6 to v4 on connect errors somehwere recommended/described? See the RFC's and the many transition documents. For instance one of the newer ones: http://gsyc.escet.urjc.es/~eva/doc/draft-ietf-v6ops-application-transition-02.txt And of course Eva's and Itojun's various other documents about this. > > I can prepare a kernel patch to make autoconf stop configuring any ipv6 > address on upping an interface. > > Some of the discussion is archived on the debian bugtracker: > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=253590 Like Trent mentions there: Fix the network ;) Protecting against userstupidity is not an option. Greets, Jeroen
Attachment:
signature.asc
Description: This is a digitally signed message part
