On Sat, 19 Aug 2000, Steve & Ulli Bowman wrote:
> G'day Stephen Illingworth,
> On 18-Aug-00, on the subject of "[Miami] Re: Miamitraceroute", you spoke
> thus:
> > Hello Steve
> >
> > On 18-Aug-00, you wrote:
> >
> >> On occassion when I get a "Cannot connect to X" error in my browser,
> >> and I do a Miamitraceroute to said host, there appears to be a loop in
> >> the path, eg.:
> >>
> [...]
> > This could be one of two things.
> >
> > 1) Badly configured routers (unlikely) 2) The machine,
> > www.search.asic.gov.au, is completely offline or the connection to that
> > machine is poor or nonexistant (probable)
No, it is #1. If it were #2 there would be no response at all.
> > Either way, there isn't a lot you can do apart from contact the sysops
> > at either www.search.asic.gov.au, or one of the two routers involved in
> > the loop.
>
> Is it usual to see two routers looping like this in these situations? I
> would have expected to see "* * *" in MiamiTraceRoute or something else
> meaningful.
It's not uncommon for routing tables to get fouled up in a way that
creates loops. The algorithms involved in maintaining routing tables
where multiple paths are possible are quite complex, and certain glitches
can cause this behavior. Sometimes it comes from a confused attempt to
find a way around a dead router. If it weren't for the time-to-live field
in the packets (which is also exploited to make traceroute work), things
would be even worse due to an increasing number of packets looping
indefinitely over the same links.
Indeed a simple dead router with no alternative will produce "* * *" from
that point on in the trace.
Fred Wright
--
To unsubscribe send "unsubscribe miami-talk-ml" to
"[EMAIL PROTECTED]". For help on list commands send "help" to
"[EMAIL PROTECTED]".
