*
* I've hacked up a round-trip-route version of this that works by
* sending a loose-source-routed udp datagram through the destination
* back to yourself. Unfortunately, SO many gateways botch source
* routing, the thing is almost worthless. Maybe one day...
*
* -- Van Jacobson (van@ee.lbl.gov)
* Tue Dec 20 03:50:13 PST 1988
almost exactly ten years ago, to the week. sometimes it really seems like
nothing has happened in networking in the last ten years.... ;-)
neal
On Mon, 14 Dec 1998, Stefan Savage wrote:
> Cool! Two possible reasons why this didn't work.
>
> 1) traceroute/kernel shortcuts packets destined for self (I doubt this)
>
> 2) our campus doesn't forward LSR packets (true).
>
> Someone with an ADSL box want to try this?
>
> Run:
>
> traceroute -g www.net.cmu.edu my-home-machine
>
> and compare it to the output of plain:
>
> traceroute www.net.cmu.edu
>
> - Stefan
>
> -----Original Message-----
> From: acollins@hotate [mailto:acollins@hotate]
> Sent: Monday, December 14, 1998 9:48 PM
> To: savage@cs.washington.edu
> Cc: syn@cs.washington.edu
> Subject: Re: forward vs reverse path measurements: a neat use for lsping
>
>
>
> This sort of behavior is an option in some versions of traceroute with
> the -g option (seems to exist in the BSD version, but not the linux
> version on hotate). Stevens talks briefly about this in volume 1,
> starting at the bottom of page 105.
>
> It should be possible to get the whole round trip path with the
> following:
>
> host1: traceroute -g host2 host1
>
> which traceroutes to the originator, but insists on the target as a
> loose source route. None of this seems to work quite right on the
> loom cluster however, but I didn't try very hard.
>
> Andy
>
>
> > Ok, here's the crazy idea of the day (abusing IP is much more fun than
> > making slides):
> >
> > Modifying Kenichi's lsping to differentiate forward from reverse
> > paths.
> >
> > Ok, so first, the target host has to forward LSR IP packets. I think
> > Kenichi found that about 10% of the servers he surveyed would do this
> > (surprising, but apparently true). Ok, now traceroute from the source
> > to the target, and report the routers traversed and the final hop
> count
> > as finalTtl.
> >
> > Now, set ttl to FinalTtl + 1, and send the packet from the source
> "back
> > to the source", but loose source routed via the target. It will get
> to
> > the target and then will return back towards the source, but will only
> > go one hop before ttl is decremented to 0 and an ICMP error will be
> > returned.... to the source from the offending router. In this way,
> one
> > can implement "reverse traceroute" that gets the list of hosts
> traversed
> > in the "reverse" direction.
> >
> > Anyone (Kenichi?) interesting in trying to get this to work?
> >
> > - Stefan
> >
> >
> >
>
>
> --
> //----------------------------------------------------------------------
> ---
> // Andy Collins -- KC6YEY
> acollins@cs.washington.edu
> // Graduate Student, University of
> Washington
> // Ford Pinto Rule:
> // Never buy a car that has a wick.
> // (collection last updated by Don
> Woods)
> //----------------------------------------------------------------------
> ---
>