> --- Tarragon Allen <[EMAIL PROTECTED]> wrote: > > On Wednesday 24 December 2003 08:23, Matthew Whitworth wrote: > > > I have a host (eth0 = 10.0.30.251) to which I recently added a virtual > > > interface (eth0:0 = 10.0.30.249). Now all traffic that originates > > > from this host has the source address of eth0:0, which is causing some > > > problems with a firewall that I don't control. > > > > Annoying isn't it? > > > > Try setting the second IP address as an alias on the *loopback* adapter, > > ie: lo:0. Use a bitmask of 255.255.255.255 otherwise strange things may > > happen. This should force the system to only use the "real" address for > > outgoing requests, but it will still accept connections to the aliases > > address.
On Wednesday 24 December 2003 10:28, Mike Mestnik wrote: > Has this problem been reported to the linux kernel ppl? Has it been fixed? > It might also help to mention what versions are affected. > > mike Yes, it's known about. I think the general response would be that it's up to the userspace program, not the kernel, to decide what IP to bind to. If the userspace doesn't care, then the kernel will pick one, usually the same subnet IP if the destination is directly accessible, or the closest IP to the gateway. It's not a problem per se. t -- GPG: http://n12turbo.com/tarragon/public.key
