On Mon, 14 Jul 2003, Jeroen Massar wrote:
> Mauro Tortonesi [mailto:[EMAIL PROTECTED] wrote:
>
> > are you sure? i can trace your host from ds6:
> >
> > [EMAIL PROTECTED] mauro]$ /usr/sbin/traceroute6 -s
> > 2001:760:204:10:10:a7ff:fe16:27f4
> > 3ffe:8114:2000:240:290:27ff:fe24:c19f
> > traceroute to 3ffe:8114:2000:240:290:27ff:fe24:c19f
> > (3ffe:8114:2000:240:290:27ff:fe24:c19f) from
> > 2001:760:204:10:10:a7ff:fe16:27f4, 30 hops max, 16 byte packets
> > 1 2001:760:204:10:202:16ff:febc:1fc1
> (2001:760:204:10:202:16ff:febc:1fc1) 1.366 ms * 0.791 ms
> > 2 3ffe:830f::a (3ffe:830f::a) 76.262 ms * 67.166 ms
> > 3 3ffe:8100:102::1:6 (3ffe:8100:102::1:6) 175.268 ms 179.469 ms
> 187.643 ms
> > 4 3ffe:8120::19:2 (3ffe:8120::19:2) 208.773 ms 232.596 ms 202.754 ms
> > 5 ipng.nl (2001:6e0::250:4ff:fe4a:7708) 190.97 ms 181.706 ms 182.735
> ms
> > 6 3ffe:8114:2000:240:290:27ff:fe24:c19f
> > (3ffe:8114:2000:240:290:27ff:fe24:c19f) 219.356 ms 228.461 ms 215.707
> ms
>
> You have to realize that the internet is a dynamic thing.
of course.
> And my previous message was some days ago. The routing is thus
> also a lot different now. Hop 4 above matches hop 2 below.
> Hop 5 above matches hop 1 below, hop 6 is my endpoint.
> So far so good. But then there is renater in between...
>
> traceroute to 2001:760:204:10:10:a7ff:fe16:27f4
> (2001:760:204:10:10:a7ff:fe16:27f4) from
> 3ffe:8114:2000:240:290:27ff:fe24:c19f, 30 hops max, 16 byte packets
> 1 gw-20.ams-02.nl.sixxs.net (3ffe:8114:1000::26) 19.703 ms 19.312 ms
> 19.369 ms
> 2 Amsterdam.core.ipv6.intouch.net (2001:6e0::2) 19 ms 19.669 ms 19.917
> ms
> 3 3ffe:8120::19:1 (3ffe:8120::19:1) 49.353 ms 55.394 ms 49.947 ms
> 4 renater.gtpv6.renater.fr (2001:660:1102:4003::1) 311.005 ms 286.745 ms
> 286.65 ms
> 5 gsr-nio.gsr-nio_gtpv6.projets.renater.fr (2001:660:3007:16:1::) 415.911
> ms 287.458 ms 287.868 ms
> 6 gsr-6net.gsr-nio_gsr-6net.projets.renater.fr (2001:660:3007:12:2::)
> 286.993 ms 287.094 ms 289.085 ms
> 7 * * *
> 8 * * *
> 9 * * *
> 10 * * *
> 11 * * *
> 12 2001:798:20:200::2 (2001:798:20:200::2) 93.296 ms 93.137 ms 93.764 ms
> 13 rtg-6net.mi.garr.net (2001:760:ffff:100::5) 93.34 ms 93.508 ms *
> 14 bo-mi-g.garr.net (2001:760:ffff:ffff::12) 97.957 ms 97.034 ms 96.999
> ms
> 15 6net-rtg.bo.garr.net (2001:760:ffff:200::6) 97.926 ms 98.771 ms
> 97.261 ms
> 16 unife-bo.6net.garr.net (2001:760:fff:4::15) 100.431 ms 101.128 ms
> 99.461 ms
> 17 gw-ing-fe.ipv6.cnit.it (3ffe:8300::9) 214.396 ms 220.364 ms *
> 18 gw-fe-na.ipv6.cnit.it (3ffe:8300::4) 215.619 ms * 217.453 ms
> 19 2001:760:204:10:10:a7ff:fe16:27f4 (2001:760:204:10:10:a7ff:fe16:27f4)
> 216.207 ms 216.52 ms 213.339 ms
>
> Completely different and note to CERN, then to renater in france and then to
> italy.
> Apparently the CERN box is quite lagged:
> 8 r3gen.vianw.net (213.2.254.10) 31.608 ms 32.268 ms 31.430 ms
> 9 cern-atm7.cern.ch (192.65.185.7) 188.931 ms * 188.776 ms
>
> This should be fixed also imho, will notice the admin.
>
> You are also announcing both your /32 and your /35:
> 2001:760::/32 > 2001:778:11:4:: 2847 20965 137 LITNET GEANT GARR
> 2001:760::/35 > 3ffe:8120::19:1 513 559 137 CERN SWITCH GARR
i'll tell this to the 6net-italy netadmins.
> The best path taken here is the /35 (3rd hop :)
> And apparent from your traceroute these routes are very asynch.
> I hope your stack is sending out it's packets on the correct interface.
well, we have a single NIC but we're multihomed.
> There are good ISP's who drop packets from mismatching origins.
> I wonder why the /35 all have a completely different path from the /32.
>
> > [EMAIL PROTECTED] mauro]$ /usr/sbin/traceroute6 -s
> > 3ffe:8300:0:1:10:a7ff:fe16:27f4 3ffe:8114:2000:240:290:27ff:fe24:c19f
> > traceroute to 3ffe:8114:2000:240:290:27ff:fe24:c19f
> > (3ffe:8114:2000:240:290:27ff:fe24:c19f) from
> > 3ffe:8300:0:1:10:a7ff:fe16:27f4, 30 hops max, 16 byte packets
> > 1 3ffe:8300:0:1:202:16ff:febc:1fc1
> > (3ffe:8300:0:1:202:16ff:febc:1fc1) 0.948 ms * 1.07 ms
> > 2 3ffe:8300::5 (3ffe:8300::5) 24.308 ms * 25.03 ms
> > 3 3ffe:8100:102::1:6 (3ffe:8100:102::1:6) 189.886 ms
> > 212.803 ms 202.296 ms
> > 4 3ffe:b00:c18::61 (3ffe:b00:c18::61) 290.605 ms 300.884
> > ms 290.953 ms
> > 5 ipng.nl (2001:6e0::250:4ff:fe4a:7708) 297.034 ms
> > 295.936 ms 292.553 ms
> > 6 3ffe:8114:2000:240:290:27ff:fe24:c19f
> > (3ffe:8114:2000:240:290:27ff:fe24:c19f) 310.646 ms 316.277
> > ms 342.222 ms
>
> [EMAIL PROTECTED]:~$ traceroute6 3ffe:8300:0:1:10:a7ff:fe16:27f4
> traceroute to 3ffe:8300:0:1:10:a7ff:fe16:27f4
> (3ffe:8300:0:1:10:a7ff:fe16:27f4) from
> 3ffe:8114:2000:240:290:27ff:fe24:c19f, 30 hops max, 16 byte packets
> 1 gw-20.ams-02.nl.sixxs.net (3ffe:8114:1000::26) 20.145 ms 18.985 ms
> 18.881 ms
> 2 Amsterdam.core.ipv6.intouch.net (2001:6e0::2) 19.223 ms 19.006 ms
> 18.984 ms
> 3 gw-viagenie-cnit.ipv6.cnit.it (3ffe:830f::3) 123.431 ms 123.124 ms
> 122.881 ms
> 4 gw-cnit-viagenie.ipv6.cnit.it (3ffe:830f::2) 286.447 ms * 288.432 ms
> 5 gw-fe-na.ipv6.cnit.it (3ffe:8300::4) 310.649 ms * 310.805 ms
> 6 3ffe:8300:0:1:10:a7ff:fe16:27f4 (3ffe:8300:0:1:10:a7ff:fe16:27f4)
> 314.122 ms 311.215 ms 310.889 ms
>
> Only One AS path here, so no problems:
> 3ffe:8300::/28 > 3ffe:b00:c18::60 10566 20745
>
> But it depends on which prefix is chosen... RIR address doesn't work at all.
> Either way, 6bone is going over the US too..
> Quick look using GRH and only showing Best shows that everybody takes that
> path...
>
> > the production address should be reachable from any network which is
> > connected to the 6net. perhaps there is a routing problem between
> > your host/net and the 6net.
>
> You mean between 6net and the rest of the world as 6net has a stupid
> "we don't accept ASpaths longer than 3 hops" clause?
>
> > > > Problem exists since two weeks or so.
> > > >
> > > > Perhaps someone could check this from different 6to4
> > > > addresses and/or look into routing tables of such routers.
> > > >
> > > > See traceroutes below for more.
> > >
> > > I rather think it is a problem at cnit then in the 6to4 setup.
> > > They really should be checking their routing tables.
> > > And clean them up a lot.
> >
> > well, we must be able to track down the problem before we can
> > fix it ;-)
> > are we really sure that this is a problem with the cnit BGP RT?
>
> Trace to any 6to4 address out there, preferebly Pieter's as he
> is seeing that problem.
already done. the problem seems to be solved. that's why i was wondering
if the cause was really cnit's BGP RT.
> Traceroute should show where it goes wrong.
unfortunately, it's rather difficult to find out what's wrong if you can't
reproduce the problem in a deterministic way.
> Another way to check it is to dump your ASpath's.
> Can't look into that as it's not in GRH...
i'll try to take a look at this.
--
Aequam memento rebus in arduis servare mentem...
Mauro Tortonesi [EMAIL PROTECTED]
[EMAIL PROTECTED]
[EMAIL PROTECTED]
Deep Space 6 - IPv6 with Linux http://www.deepspace6.net
Ferrara Linux User Group http://www.ferrara.linux.it
---------------------------------------------------------------------
The IPv6 Users Mailing List
Unsubscribe by sending "unsubscribe users" to [EMAIL PROTECTED]
