Sporadic packet loss of large packets between Google and HE

[thardie]

My local LAN has connectivity using tunnelborker. Works fine most of the time, but I'm having problems with some clients accessing Google servers. I've narrowed it down to what looks to be a large packet problem between HE and google. Here's a MTR with small (default) packets:

                                                       My traceroute  [v0.75]
orcas (::)                                                                                                 Fri Jun 17 22:40:45 2011
Keys:  Help   Display mode   Restart statistics   Order of fields   quit
                                                                                           Packets               Pings
Host                                                                                    Loss%   Snt   Last   Avg  Best  Wrst StDev
1. 2001:470:1f05:de3::2                                                                  0.0%   133    0.3   0.3   0.2   0.5   0.0
2. thardie-1.tunnel.tserv3.fmt2.ipv6.he.net                                              0.0%   133   32.8  31.2  25.4  53.2   5.4
3. gige-g5-19.core1.fmt2.he.net                                                          0.0%   133   21.5  27.4  15.6  66.6  10.3
4. 10gigabitethernet1-2.core1.pao1.he.net                                                0.0%   133   22.7  24.8  16.7  51.5   6.3
5. core2-1-1-0.pao.net.google.com                                                        0.0%   133   18.2  26.1  16.4  82.5  12.2
6. 2001:4860::1:0:21                                                                     0.0%   132   35.2  26.2  18.1  80.8   9.5
7. 2001:4860:0:1::1b7                                                                    0.0%   132   18.8  24.9  17.8 178.9  15.4
8. 2001:4860:4001:803::1011                                                              0.0%   132   19.9  24.7  17.6  64.0   7.4

When the problem is happening, if I do a MTR with large packets (1000 bytes) between hops 4 and 5, loss will go up dramatically (around 20-30%), and it follows to all the hops below.

It doesn't happen all the time. Seems there's some congestion on that link?

Thanks,

Terry

[mtindle]

Can you please send a traceroute showing the loss with the larger MTU and any further information you might have?  If it's only happening with larger packets, it sounds more like a PMTU issue and I can see if there's something I can track down with Google.