Strange IPv6 reachability problem

[Seth Mattinen]

On 6/19/2011 4:22 PM, Jeroen Massar wrote:
>
> If that is 'rude' then you need to adjust your sensitivity meter.
>
> This list is called ipv6-ops, not 'ipv6-user-helpdesk' as such you
> should be aware of the simple network debugging tools and have the
> ability to provide the most minimal amount of information when debugging
> issues.

The simple tools didn't help explain what I was seeing. See:


source: 2607:fe70:0:beef::a
dest: 2001:500:61:28::70

traceroute to 2001:500:61:28::70 (2001:500:61:28::70), 30 hops max, 40 
byte packets
  1  2607:fe70:0:beef::1  0.859 ms  0.782 ms  0.694 ms
  2  2607:fe70::6:1  0.172 ms  0.176 ms  0.159 ms
  3  2607:fe70::3:2  1.938 ms  1.937 ms  1.922 ms
  4  2600:6:300::6  6.690 ms * *
  5  2600:0:2:1239:144:232:6:61  6.851 ms  6.898 ms  6.886 ms
  6  2600:0:2:1239:144:232:20:113  9.564 ms 2600:0:2:1239:144:232:24:34 
  9.471 ms 2600:0:2:1239:144:232:9:224  9.436 ms
  7  2600:0:2:1239:144:232:18:31  9.953 ms  9.522 ms  9.497 ms
  8  2001:504:0:1::6939:1  9.460 ms  10.796 ms  10.754 ms
  9  2001:470:0:32::2  18.455 ms  19.211 ms  17.397 ms
10  2001:470:0:38::2  23.232 ms  22.637 ms  20.560 ms
11  2001:4f8:0:1::49:2  81.698 ms  81.695 ms  81.185 ms
12  2001:500:61:28::70  82.461 ms  82.429 ms  82.426 ms


source: 2607:fe70:1002:0:202:b3ff:fe30:f633
dest: 2001:500:61:28::70

traceroute to 2001:500:61:28::70 (2001:500:61:28::70), 30 hops max, 40 
byte packets
  1  2607:fe70:1002::1  2.385 ms  2.341 ms  2.308 ms
  2  2607:fe70:1001::1  10.189 ms  10.176 ms  10.178 ms
  3  2607:fe70::14:1  10.536 ms  10.526 ms  11.711 ms
  4  2607:fe70::5:2  14.571 ms  14.612 ms  14.616 ms
  5  2607:fe70::a:1  11.650 ms  11.631 ms  11.618 ms
  6  2607:fe70::4:1  11.582 ms  15.339 ms  16.466 ms
  7  2600:6:300::6  20.470 ms * *
  8  2600:0:2:1239:144:232:6:61  16.270 ms  16.079 ms  16.100 ms
  9  2600:0:2:1239:144:232:9:224  16.996 ms 2600:0:2:1239:144:232:24:34 
  20.877 ms 2600:0:2:1239:144:232:9:224  20.790 ms
10  2600:0:2:1239:144:232:18:31  25.681 ms  25.753 ms  23.043 ms
11  2001:504:0:1::6939:1  23.044 ms  23.801 ms  20.814 ms
12  2001:470:0:32::2  29.885 ms  25.846 ms  25.799 ms
13  * * *
14  * * *



source: 2607:fe70:10:0:230:48ff:fe9e:51e8
dest: 2001:500:61:28::70

traceroute to 2001:500:61:28::70 (2001:500:61:28::70), 30 hops max, 40 
byte packets
  1  2607:fe70:10::1  0.539 ms  0.543 ms  0.550 ms
  2  2607:fe70::6:1  0.164 ms  0.138 ms  0.102 ms
  3  2607:fe70::3:2  0.805 ms  0.747 ms  0.701 ms
  4  2600:6:300::6  5.829 ms * *
  5  2600:0:2:1239:144:232:6:63  7.261 ms  7.231 ms  7.196 ms
  6  2600:0:2:1239:144:232:20:181  10.044 ms 2600:0:2:1239:144:232:24:32 
  9.507 ms 2600:0:2:1239:144:232:20:181  9.566 ms
  7  2600:0:2:1239:144:232:8:151  10.679 ms  10.653 ms  10.622 ms
  8  2001:504:0:1::6939:1  11.247 ms  8.724 ms  9.455 ms
  9  2001:470:0:32::2  17.622 ms  18.167 ms  17.652 ms
10  * * *
11  * * *


They look pretty similar right up until it fails. What's so special 
about the second and third source addresses that it won't route back but 
the first one will? The first and third even take identical paths out on 
my side, yet one works and one doesn't. So I checked netflow (which 
you're assuming is broken for some reason even though its data exactly 
matches the traceroutes and pings) and didn't see any traffic coming 
back in.

I'm announcing it as a whole /32 so I'd expect it to all come back to 
me, but it's not, so my original question (or complaining on a public 
list, "ipv6-user-helpdesk" question, or whatever) still stands.

~Seth