6bone Database cleansing?

[Jeroen Massar]

As I was about to create a nice 'ping-over-the-6bone' tool for ourselfs
(IPng.nl) I first tested how many of the hosts actually where alive.
With 'hosts' I mean the 'application: ping' lines found in the nightly
6bone.db.

I conducted the test at around 01:22 CET the 13the of April from
ping.ipng.nl (IPv6) and came up with the following results:

jeroen@tunnelserver:~$ cat 6bone.db | grep application | grep ping | wc
-l
    791

Which (should) mean there are 779 hosts defined in application ping
lines.
I passed all these hosts to a patched version of fping6 which skips
hosts without addrinfo (no AAAA ;)
pinging using 5 icmpv6 packets, which produced the following results

jeroen@tunnelserver:~$ cat output.txt | grep "no addrinfo" | wc -l
    329
jeroen@tunnelserver:~$ cat output.txt | grep "unreachable" | wc -l
    205
jeroen@tunnelserver:~$ cat output.txt | grep "alive" | wc -l
    257

Of these 'unreachables' where:

jeroen@tunnelserver:~$ cat output.txt | grep unreachable | grep -e "^5f"
| wc -l
     26
are of the historical 5fxx:: prefix, I will excuse those.
This still means, assuming all peering was up at the moment, which as
far as I can tell was the case,
that about 175+ hosts weren't reachable for one way or the other and
those 329 that didn't resolve ouch.
I still hope it's a glitch, but a next test today revealed almost the
same results.

And that's only the "application ping" part, now for a more interresting
take, at least I wanted to nag about the fact that
the ipv6-site SURFNET (real) == KRAAKPAND (false) in the database, but
that apparently has been resolved already ;)

I also noticed that not all object/subnetdelegations are noted in the
database nor refered to other whois servers.

With these current stats I will prolly make a wellknown-hosts pinger as
pinging every random site available is a bit overkill anyways.

Greets,
 Jeroen