Re: Proposal for two minor improvements in prtraceroute
- Date: Fri, 7 Jan 1994 11:00:41 +0100 (CET)
> *
> Well it does in my version ????
> traceroute with AS and policy additions [Jan 6 15:08:55 UTC]
>
> from AS1104 ns.ripe.net (192.87.45.1)
> to AS1128 Amsterdam1.dante.net (193.172.12.30)
>
> 1 AS1104 hef-router.nikhef.nl (192.87.45.80) [I]
> 2 AS1103 Amsterdam1.router.surfnet.nl (192.16.183.112) [D1]
> 3 AS1103 Amsterdam2.router.surfnet.nl (145.41.9.130) [I]
> 4 AS2043 amsterdam4.empb.net (193.172.4.17) [ERROR]
> 5 AS2043 amsterdam5.empb.net (193.172.4.19) [I]
> 6 AS1128 Amsterdam1.dante.net (193.172.12.30) [?]
>
> AS Path followed: 1104 1103 2043 1128
>
> AS1104 = NIKHEF-H
> AS1103 = SURFnet IP
> AS2043 = European Multiprotocol Backbone
> AS1128 = DANTE Gateway in Amsterdam
>
> So this is a mystery - can you mail to pride-tools@localhost a copy of
> your prtraceroute ?
> Also, just in case there was some confusion I have remade the 1.0.1
> distribution which may mean you get a newer prtraceroute as well.
That is certainly a mystery - I will send you my copy in a separate letter. I
picked it up from ftp.ripe.net as the 1.0.1 version yesterday, so it should be
pretty recent. A possible difference between our systems could be the quality
of the underlying perl interpreter, using C library gethostby* calls; however,
I have just used the perl debugger to verify that my perl interpreter works
proper on this point.
By the way, why does your run show ERROR in line 4 and question mark in line 6?
As far as I can see, the route is legal according to the specified policies
of all four ASs.
> Well not really - the address is chosen implicitly by the user (i.e
> the IP address typed in or slected by DNS as you typed a hostname and
> that IP address being traced to will always be shown).
>
Yes, I know that the IP address being traced will be shown; my proposal was
meant to counter the natural error for a human reader to commit when reading
to AS1128 Amsterdam1.dante.net (193.172.12.30)
namely to focus on the most readable part, the DNS name, and use that
when (for instance) trying to reproduce a run. That is why I suggested
that you used a graphically different output format in the case where
an operator explicitly specifies an IP address.
But this is no big deal; if you prefer your current solution, you stick to
that.
Greetings -
--peder chr.