Mailing List Archive

On Wed, 29 Jun 2005, Daniel Roesen wrote:

> I really wonder why a german ISP needs a tunnel to Canada as well. :-)

Oh, it's probably multihoming. ;-)

traceroute to 2001:6c8::1 (2001:6c8::1) from 2001:410:9000:127::10, 30
hops max, 16 byte packets
1 ncrgate-vlan270.nrc.ca (2001:410:9000:127::1) 0.374 ms 0.256 ms
0.226 ms
2 nrcgate-pap-1.ipv6.nrc.ca (2001:410:9000::9) 0.378 ms 0.412 ms
0.257 ms
3 2001:478:149::12 (2001:478:149::12) 7.394 ms 17.696 ms 7.004 ms
4 3ffe:3900:a3::1 (3ffe:3900:a3::1) 37.905 ms 35.671 ms 35.533 ms
5 3ffe:3900:a8::2 (3ffe:3900:a8::2) 36.45 ms 36.176 ms 36.639 ms
6 chicr1-ge0-chirt1.es.net (2001:400:0:77::1) 75.68 ms 75.563 ms
77.873 ms
7 snvcr1-oc192-chicr1.es.net (2001:400:0:54::1) 75.543 ms 75.591 ms
75.759 ms
8 lbl2-snvcr1.es.net (2001:400:0:87::2) 77.367 ms 76.952 ms 78.482 ms
9 esnetv6r1-lbl2.es.net (2001:400:0:93::2) 76.936 ms 77.545 ms 77.129 ms
10 sl-bb1v6-rly-t-19.sprintv6.net (3ffe:2900:a:4::1) 446.747 ms 423.794
ms 310.536 ms
11 nl-ams06d-re1-t-9.ipv6.aorta.net (2001:730::1:61) 239.302 ms 249.268
ms 236.927 ms
12 tun13.sltnxx1.inet6.tele.dk (2001:6c8:0:fffd::12) 272.414 ms 272.815
ms 270.791 ms


wfms

On Wed, Jun 29, 2005 at 08:25:05AM -0400, William F. Maton Sotomayor wrote:
> >I really wonder why a german ISP needs a tunnel to Canada as well. :-)
>
> Oh, it's probably multihoming. ;-)

You won't find better places than .nl and .de currently to get sane
IPv6 transit. And actually, Noris _has_ other sane transit aside
Teleglobe directly in Frankfurt. :-)

> 12 tun13.sltnxx1.inet6.tele.dk (2001:6c8:0:fffd::12) 272.414 ms 272.815
> ms 270.791 ms

Yeah, works from most sources. Someone in the ASPAC area is blackholing
packets it seems. I have seen difficulties like that on other routes as
well (e.g. from Sprint to C&W, via Asia). Didn't have the energy though
to try finding the root cause.


Best regards,
Daniel

--
CLUE-RIPE -- Jabber: dr@cluenet.de -- dr@IRCnet -- PGP: 0xA85C8AA0

On Wed, Jun 29, 2005 at 02:20:07PM +0200, Daniel Roesen wrote:
> 5 if-11-0-1-459.6bb1.AD1-Amsterdam.ipv6.teleglobe.net (2001:5a0:200::15) 3.342 ms 5.451 ms 3.043 ms
> 6 gin-mtt-6bb1.ipv6.teleglobe.net (2001:5a0:300::1) 90.268 ms 115.456 ms 90.385 ms
> 7 tu-viagenie.ipv6.noris.de (2001:780::b) 116.046 ms 140.328 ms 120.178 ms
> 8 3ffe:b00:c18::6b (3ffe:b00:c18::6b) 199.959 ms 201.08 ms 193.336 ms
>
> Seems like hop 7 (noris.de, small german ISP) forwards a ~fullish table
> to their Upstream Teleglobe, who happily accept anything obviously.

And somethings, things are completely different than they look. The
mistake of believing DNS reverse. :-)

Word is, that the Noris-Tunnel is long gone, but Teleglobe's router
(hop 7) possibly still having it configured, and router answering with
wrong IP address as source for the ICMPv6 hop count exceeded. Looking
at the hop RTTs that makes sense too. Hop 5 is Amsterdam, Hop 6 might be
US east coast, hop 7 Teleglobe US west coast. Hop 8 is already Viagenie
Canada.

The AS_PATH as seen at io.nl confirms that:
AS path: 6453 10566 6939 4716 6175 3292 I
No Noris inside there.

Teleglobe: could you take a look at your hop 7 router?
Noris: sorry for the noise, you are not involved. :-)


Best regards,
Daniel

--
CLUE-RIPE -- Jabber: dr@cluenet.de -- dr@IRCnet -- PGP: 0xA85C8AA0

Hi,

found out why traceroute ends. Sprint has no route back to noc.sixxs.net
from where I'm tracerouting:

noc.sixxs.net has IPv6 address 2001:838:1:1:210:dcff:fe20:7c7c

*>i2001:830::/32 2001:478:FFFF::9
4294967294 90 0 4555 6939 3257 25560 21501 i
*>i2001:840::/32 2001:478:FFFF::9
4294967294 90 0 4555 6939 6939 5381 i

No 2001:838::/32.

A Sprint<->Teleglobe peering would fix that.


Best regards,
Daniel

--
CLUE-RIPE -- Jabber: dr@cluenet.de -- dr@IRCnet -- PGP: 0xA85C8AA0

On 29/06/2005, at 10:11 PM, Daniel Roesen wrote:

> Yeah, works from most sources. Someone in the ASPAC area is
> blackholing
> packets it seems. I have seen difficulties like that on other
> routes as
> well (e.g. from Sprint to C&W, via Asia). Didn't have the energy
> though
> to try finding the root cause.

Works for me but unfortunately it goes via Asia and North America due
to a shorter peer based AS path seen from Asia.

traceroute to 2001:6c8::1 (2001:6c8::1) from
2001:388:4000:4002:204:23ff:fea9:6440, 30 hops max, 16 byte packets
1 gigabitethernet0-2.er2.aarnet.cpe.aarnet.net.au
(2001:388:4000:4002:20f:23ff:fea3:ec00) 0.501 ms 0.286 ms 0.192 ms
2 gigether0-2-3.bb1.a.adl.aarnet.net.au
(2001:388:1:2003:204:e0ff:fe00:1023) 19.129 ms 18.66 ms 18.579 ms
3 pos0-1-0.bb1.a.mel.aarnet.net.au (2001:388:1:6::2) 28.201 ms
27.303 ms 27.309 ms
4 pos0-1-0.bb1.b.syd.aarnet.net.au (2001:388:1:a::2) 40.07 ms
39.806 ms 40.421 ms
5 gigabitethernet3-0.bb3.a.syd.aarnet.net.au (2001:388:1:1d:
205:ddff:feff:2c54) 21.314 ms 21.262 ms 21.313 ms
6 pos2-0.bb1.a.suv.aarnet.net.au (2001:388:1:16:212:43ff:fe40:6200)
57.656 ms 57.749 ms 57.606 ms
7 pos1-0.bb1.b.hnl.aarnet.net.au (2001:388:1:17:205:ddff:fe6d:2008)
117.256 ms 116.922 ms 117.139 ms
8 * 2001:388:1:c803:290:6900:c8ed:e49d
(2001:388:1:c803:290:6900:c8ed:e49d) 117.935 ms *
9 * 3ffe:8140:101:d::4 (3ffe:8140:101:d::4) 196.084 ms *
10 3ffe:8140:101:6::3 (3ffe:8140:101:6::3) 252.745 ms 252.818 ms
252.661 ms
11 hitachi1.otemachi.wide.ad.jp (2001:200:0:1800::9c4:2) 236.813
ms 236.847 ms 237.167 ms
12 pc6.otemachi.wide.ad.jp (2001:200:0:1802:2d0:b7ff:fe88:eb8a)
236.809 ms 236.624 ms 236.642 ms
13 ix6.v6.dti.ad.jp (2001:200:0:1800::4691:1) 237.058 ms 237.069
ms 237.139 ms
14 lan-gate.core1.otemachi.v6.dti.ad.jp (2001:2e8:20:10::3) 237.169
ms 237.057 ms 237.034 ms
15 otemachi-gate.core1.otemachi4.v6.dti.ad.jp (2001:2e8:20::11:2)
237.873 ms 237.234 ms 236.82 ms
16 lan-gate.jpnap.otemachi4.v6.dti.ad.jp (2001:2e8:20:11::2) 237.38
ms 237.136 ms 237.146 ms
17 fa-1-3-2.a13.tokyjp01.jp.ra.verio.net (2001:218:2000:5000::3d)
253.225 ms 253.358 ms 253.123 ms
18 ge-7-0-0.a21.tokyjp01.jp.ra.verio.net (2001:218:2000:3002::22)
253.183 ms 253.45 ms 253.076 ms
19 xe-0-0-0.r21.tokyjp01.jp.bb.verio.net (2001:218:0:6000::21)
252.748 ms 252.969 ms 253.072 ms
20 p64-2-1-0.r21.sttlwa01.us.bb.verio.net (2001:218:0:2000::19d)
252.245 ms 252.17 ms 278.542 ms
21 * * *
22 p16-3-0-0.r80.nycmny01.us.bb.verio.net (2001:418:0:2000::f2)
316.97 ms 316.773 ms 316.742 ms
23 p16-7-1-1.r21.londen03.uk.bb.verio.net (2001:418:0:2000::2ca)
390.201 ms 390.273 ms 390.199 ms
24 fa-1-0.r00.londen03.uk.b6.verio.net (2001:728:0:7000::b600)
387.729 ms 387.737 ms 387.552 ms
25 tu-0.teledanmark.londen03.uk.b6.verio.net (2001:728:0:4000::e)
421.215 ms 421.784 ms 421.113 ms

Mark.

Hi,

> 7 tu-viagenie.ipv6.noris.de (2001:780::b) 116.046 ms 140.328 ms 120.178 ms

that was an very old ipv6 aaddress of an early tunnel which doesn't
exist anymore (since a very long time).

> 8 3ffe:b00:c18::6b (3ffe:b00:c18::6b) 199.959 ms 201.08 ms 193.336 ms
> 9 3ffe:80a::e (3ffe:80a::e) 286.115 ms 285.733 ms 286.887 ms
> 10 * * *
>
> Seems like hop 7 (noris.de, small german ISP) forwards a ~fullish table
> to their Upstream Teleglobe, who happily accept anything obviously.

when i left noris, there was no BGP session to Teleblobe at all and
i don't guess, that they have setup such a session.

Cheers,

Marco