Routing problem?

Frank Shute frank at esperance-linux.co.uk
Tue Nov 19 13:47:05 GMT 2002


On Tue, Nov 19, 2002 at 12:14:52PM +0000, Matthew Seaman wrote:
>
> On Tue, Nov 19, 2002 at 11:25:23AM +0000, Lou Kamenov wrote:
> > In some email I received from Matthew Seaman
> > <m.seaman at infracaninophile.co.uk> on Tue, 19 Nov 2002 09:23:04 +0000 :
> > > On Tue, Nov 19, 2002 at 08:09:38AM +0000, Frank Shute wrote:
> > > > 18  193.ATM6-0.GW6.TCO1.ALTER.NET (152.63.37.37)  215.447 ms 
> > > > 207.485 ms  209.401 ms 19 * * *
> > > 
> > > Looks like that router is kaput.
> > >  
> > 
> > Erm who knows, my route is different.
> > 
> > -- my traceroute --
> > 14  193.ATM7-0.GW6.TCO1.ALTER.NET (152.63.37.53)  98.324 ms  95.686 ms 
> > 97.938 ms
> > 15  pbs-gw.customer.ALTER.NET (157.130.32.130)  97.744 ms  98.016 ms 
> > 100.161 ms 16  * * *
> > -- end --
> > 
> > AFAICS {www.}pbs.org doesnt reply to any icmps, however we know how
> > traceroute works /icmp/udp/.
> 
> Since traceroute can send out whatever sort of packet you want (tcp,
> udp, gre (whatever that is), defaults to udp) and it's the
> interveining gateways that send back the ICMP TTL exceeded or
> destination unreachable responses, I tried a a few different options.
> None worked.  It's not unreasonable for them to filter out packets to
> random UDP or TCP high ports... Even so, seems they're filtering ICMP
> echo {request,reply} as well. Probably all ICMP types. Grrr... I hate
> sites that just blindly do that.
>  
> > > > ie. it seems to get stuck at alter.net's Godforsaken network.
> > > > Anybody got an idea what's going on here? 
> > I think that it`s only a filter... 
> 
> Looks like they've got at least two routers on some sort of load
> balance:
> 
>     193.ATM7-0.GW6.TCO1.ALTER.NET (152.63.37.53)
>     193.ATM6-0.GW6.TCO1.ALTER.NET (152.63.37.37)
> 
> > I can access www.pbs.org (149.48.192.134) with my browser..
> 
> Yeah --- I can access the site by IP number.  Slow, and it looks to me
> like my system is having to do some retries to retrieve all the images
> on the site: maybe one of those routers is sending stuff into
> never-never land.

I think the images are a problem because they've got things like:

<img src="http://www.pbs.org/images/features/discs/nova/venice_keahey.gif">

in their html instead of:

<img src="./images/features/discs/nova/venice_keahey.gif">

so I guess the lookup for a lot of the images is failing. If I was
really keen I'd dump the html and see if:

s/www.pbs.org/149.48.192.134/

does the trick.

> 
> They've got another problem though:
> 
>     happy-idiot-talk:~:% dig @8496-gslb.nyc3.aens.net www.pbs.org IN AAAA
>     
>     ; <<>> DiG 8.3 <<>> @8496-gslb.nyc3.aens.net www.pbs.org IN AAAA 
>     ; (1 server found)
>     ;; res options: init recurs defnam dnsrch
>     ;; res_nsend to server 8496-gslb.nyc3.aens.net  63.240.96.196: Operation timed out
> 
> Stupid nameservers that won't respond *at all* to AAAA requests, even
> if it's just to say "Huh?".  That will put the kybosh on any
> application that uses getaddrinfo(3), like Mozilla.

Thanks to all who replied. As long as I know it's not a problem with
my box, I can sleep easy :)

It's been down for some days now, so it looks like they've got quite a
problem.

-- 

 Frank 

*-*-*-*-*-*-*-*-*-*-*
   Boroughbridge.
 Tel: 01423 323019
     ---------
PGP keyID: 0xC0B341A3
*-*-*-*-*-*-*-*-*-*-*

http://www.esperance-linux.co.uk/






More information about the Ukfreebsd mailing list