Cannot access these sites!? Why?

[vishalrao]

ah yes, i remember the 28 byte number (IP header or whatever) - actually i think its 20 bytes for something and anothr 8 bytes for something else so the "physical" MTU is 1492... (as the modem shows for me)... and the OS/software MTU should be that number minus 28, which is 1464... so the numbers look right...i used to leave my MTUs at 1400 just to have a buffer... but i'll try 1464 on both my connections (bsnl home 1800 and home 500) to see if i can get some extra speed

 

[max]

hmm...these values should be detected correctly by Windoze.... Ah well its windoze after all

 

[blr_p]

Originally posted by SiriusB@Nov 6 2005, 08:10 PM
I have a method to find the optimim mtu. Type in

ping google.com -f -l 1452

and see. If the answer is 'DF is set' or something, you need to lower the last number. Else if the ping is successful, increase the number and try again. You will find a number that _just_ makes it through. This is the mtu for the path. For me 1452 was it.
Let me know if I have a flawed theory here.
[snapback]30886[/snapback]

I tried this test and it seems my ideal MTU is 1442 which gives a successful ping. Anything above that gives

Pinging google.com [64.233.187.99] with 1443 bytes of dataPacket needs to be fragmented but DF set.Packet needs to be fragmented but DF set.Packet needs to be fragmented but DF set.Packet needs to be fragmented but DF set.Ping statistics for 64.233.187.99:    Packets: Sent = 4, Received = 0, Lost = 4 (100% loss),

I need to check this with the Airtel CC as it's really dependant on what MTU the particular model of exchange is being used. I suspect this is why ppl are getting slightly different ideal values.

http://www.mynetwatchman.com/kb/adsl/pppoemtu.htm
Thx for this link gives a good explanation but has to take into account what MTU the exchange is using. But it only gives a 0.6% improvement. I was hoping for more.

Originally posted by max@Nov 7 2005, 12:37 PM
hmm...these values should be detected correctly by Windoze.... Ah well its windoze after all
[snapback]30965[/snapback]

[/quote]
Heh, how does windows know you are using PPP over ATM. There is an extra wrapper (imposed by the ISP) that ATM uses around TCP/IP.

The same tweak would be required on *nix as well. All windows or any other OS implementing TCP/IP does is make sure that tcp protocol only is being used properly.

 

[Ashvon]

Hi Folks,I would like to CLOSE this thread....The final action to resolve my problem was a phone-call to AIRTEL customer-service. The personnel was excellent and took me through the steps to Log-in into the Modem-(WebDistributor) settings and do some changes. Then, I was asked to TELNET into this IP...and do some commands....and Then reboot the laptop....And it worked like magic!!All the websites that were giving a problem now load smoothly!Thanks-Ashish

 

[max]

mind giving us a rough idea as to what was done?

 

[blr_p]

Heh yeah....wonder if he will come back.It's interesting he does not mention any registry changes, sounds like LAN/WAN settings. Tho i wonder why the telnet was reqd, almost everything can be done via the browser.

 

[Ashvon]

Folks,Here is what was done....the "TCP MSS Clamp" setting in the router was changed from a FALSE to a TRUE. (This was entirely done thru the web-interface for the router-managment.)I am not sure, what was the command for TELNET that was executed. (I did not write it down anywhere. Hope this helps yr knowledge-base!)Thks-Ash

 

[max]

do you remember the command? atleast the starting of it?

 

[vishalrao]

searching online (like start.com) for info on "mss" or "mss clamping"

clip:

MSS - Maximum Segment Size. This is used in negotiating what the MTU of a connection between two hosts will be. Essentially this is saying "please don't send me packets bigger than X." This should typically be set to 40 less than your MTU to allow room for headers.

hey this is cool... people focus on MTU to get speed... maybe if you use MSS settings more sites will work better?

im impressed the airtel guy was able to help this guy out over the phone!

hehe http://www.ietf.org/rfc/rfc2923.txt if you want details!

 

[max]

wow. this is a network layer modification! I am surprised that the CC guy knew it!