Anyone from ARIN listening?
Bernhard Schmidt
berni at birkenwald.de
Mon Oct 29 12:08:38 CET 2007
Pekka Savola wrote:
> ARIN's IPv6 connectivity appears to be pretty broken. I recall this has
> been the case for some time now. The largest IPv6 packet that gets
> through seems to be 1480 bytes. Bigger ones (or more likely, return
> packets) are silently blackholed, I don't see any ICMPv6 Packet too Big
> messages either. They advertise full TCP MSS of 1440 to boot.
>
> As a result, trying to open www.arin.net with a graphical web browser
> (lynx is OK because it keeps the page size down) stalls for infinite time.
Same here, forward path is 12816 680 3549 30071 10745
bschmidt at lxbsc01:~> tracepath6 www.arin.net
1?: [LOCALHOST] pmtu 1500
1: vl-23.csr1-2wr.lrz-muenchen.de 1. 81ms
2: xr-gar1-te1-3.x-win.dfn.de 0.843ms
3: 2001:638:c:c043::2 8.371ms
4: 2001:450:2002:25::1 8.316ms
5: equi6ix.dc.gblx.net asymm 6 101.371ms
6: ibr01-ve96.asbn01.occaid.net 101.796ms
7: arin-ic-1605-wdc.customer.occaid.net 102.812ms
8: no reply
9: no reply
return path is symetric (30071 3549 680 12816) according to GRH.
Another location (AS29259, 2001:1b10::/32) works fine (tracepath6
stalls, but curl http://www.arin.net/index.shtml works)
1?: [LOCALHOST] pmtu 1500
1: backbone2-gige-0-3-15.teleport-iabg.de 1.146ms
2: mchn-s1-rou-1030.DE.eurorings.net 2.783ms
3: ffm-s1-rou-1030.DE.eurorings.net asymm 2 11.868ms
4: ge-6-0-0.fra10.ip6.tiscali.net asymm 5 13.385ms
5: so-0-0-0.par77.ip6.tiscali.net asymm 6 22.599ms
6: so-5-0-0.par22.ip6.tiscali.net asymm 7 23.013ms
7: so-2-0-0.was10.ip6.tiscali.net asymm 8 100.493ms
8: ibr01-ve96.asbn01.occaid.net 109.494ms
9: arin-ic-1605-wdc.customer.occaid.net asymm 8 109.225ms
10: no reply
11: no reply
return path would be 30071 41692 8767 29259 here.
Third location (AS31333, 2001:4d88::/32) stalls again, I don't have
access to their BGP but trace shows Tiscali-OCCAID-ARIN as well. Return
path is 30071 3549 15598 31333.
All that would point to OCCAID-GBLX. http://www.occaid.org loads fine
for me as well, but there is a tunnel involved inside OCCAID
9: ibr01-ve96.asbn01.occaid.net asymm 8 109.027ms
10: ibr01-ve96.asbn01.occaid.net asymm 8 109.017ms pmtu 1480
10: bbr01-p2-1.nwrk01.occaid.net asymm 9 117.058ms
if the path is symmetric return traffic would see pMTU to 1480 inside
OCCAID, so pMTU issues on OCCAID-GBLX might not be seen.
Regards,
Bernhard
More information about the ipv6-ops
mailing list