Test your connectivity for World IPv6 Day

Bill Owens owens at nysernet.org
Mon Jun 6 04:59:26 CEST 2011


On Sun, Jun 05, 2011 at 11:04:17PM +0200, Daniel G. Kluge wrote:
> Similarly www.verisigninc.com was pointing to 2620:103:c003::10 also with a PMTUD problem, which gives you a very strange interaction: The first redirect [1] goes through, but the actual get of the page [2] is stuck forever.

The first transfer is so small it sneaks under the tunnel PMTU, but the page you're redirected to is big enough to run into it:

[cookiemonster:~] owens% sudo scamper -F ipfw -I "tbit -M 1280 -u 'http://www.verisigninc.com/' 2620:74:13:3000::40"
tbit from 2001:468:901:1:21b:63ff:fea4:4d92 to 2620:74:13:3000::40
 server-mss 1440, result: pmtud-toosmall
 app: http, url: http://www.verisigninc.com/
 [  0.050] TX SYN             64  seq = 0:0             
 [  0.136] RX SYN/ACK         64  seq = 0:1             
 [  0.136] TX                 60  seq = 1:1             
 [  0.149] TX                238  seq = 1:1(178)        
 [  0.236] RX                 60  seq = 1:179           
 [  0.237] RX                309  seq = 1:179(249)      
 [  0.237] TX                 60  seq = 179:250         
 [ 70.237] TX FIN             60  seq = 179:250         
 [ 70.324] RX FIN             60  seq = 250:180         
 [ 70.324] TX                 60  seq = 180:251         

[cookiemonster:~] owens% sudo scamper -F ipfw -I "tbit -M 1280 -u 'http://www.verisigninc.com/en_US/index.xhtml' 2620:74:13:3000::40"
tbit from 2001:468:901:1:21b:63ff:fea4:4d92 to 2620:74:13:3000::40
 server-mss 1440, result: pmtud-fail
 app: http, url: http://www.verisigninc.com/en_US/index.xhtml
 [  0.050] TX SYN             64  seq = 0:0             
 [  0.136] RX SYN/ACK         64  seq = 0:1             
 [  0.136] TX                 60  seq = 1:1             
 [  0.149] TX                255  seq = 1:1(195)        
 [  0.237] RX                 60  seq = 1:196           
 [  0.240] RX               1500  seq = 1:196(1440)     
 [  0.240] TX PTB           1280  mtu = 1280
 [  0.240] RX               1500  seq = 1441:196(1440)  
 [  3.240] RX               1500  seq = 1:196(1440)     
 [  3.240] TX PTB           1280  mtu = 1280
 [  9.240] RX               1500  seq = 1:196(1440)     
 [  9.240] TX PTB           1280  mtu = 1280
 [ 21.241] RX               1500  seq = 1:196(1440)     
 [ 21.241] TX PTB           1280  mtu = 1280
 [ 45.242] RX               1500  seq = 1:196(1440)     

And yes, I had to ask the author of scamper to explain what the heck was going on with "pmtu-toosmall" in the first result. But just like juniper.com, when this site goes live anyone behind a tunnel that has less than 1500 byte MTU won't be able to load the page.

Bill.


More information about the ipv6-ops mailing list