Hello Clive,
Thank you for your reply.
When we establish a TCP connection for a file transfer operation, the
source and target systems negotiate with each other and check each other's
MTU (""Maximum Segment size" in TCP head). However, when we run Ping
commend, the ICMP packets are sent directly.
When we run "ping -l 1401 ###.###.###.###", everything works. From the
NetMon trace, the target machine receives the packets and can fragment the
returned packets into two to meet the MTU on the local machine (1400).
When we run "ping -l 1401 -f ###.###.###.###", the target can still receive
the ICMP packets. However, since the packets cannot be fragmented, it is
unable to send out the response; this is why "Response timed out" returns.
You can verify this in the NetMon logs.
I hope the above information helps. Thanks for using our news groups!
Regards,
Joe Wu
Product Support Services
Microsoft Corporation
Get Secure! -
www.microsoft.com/security
====================================================
When responding to posts, please "Reply to Group" via your newsreader so
that others may learn and benefit from your issue.
====================================================
This posting is provided "AS IS" with no warranties, and confers no rights.
--------------------
|From: Clive <
[email protected]>
|Subject: Re: MTU Size
|Date: Fri, 05 Dec 2003 10:33:45 +0000
|Message-ID: <
[email protected]>
|References: <
[email protected]>
<
[email protected]>
|X-Newsreader: Forte Agent 1.93/32.576 English (American)
|MIME-Version: 1.0
|Content-Type: text/plain; charset=us-ascii
|Content-Transfer-Encoding: 7bit
|Newsgroups: microsoft.public.win2000.networking
|NNTP-Posting-Host: uk-15-96.client.oup.com 194.80.15.96
|Lines: 1
|Path:
cpmsftngxa07.phx.gbl!cpmsftngxa10.phx.gbl!TK2MSFTNGXA05.phx.gbl!TK2MSFTNGP08
..phx.gbl!TK2MSFTNGP11.phx.gbl
|Xref: cpmsftngxa07.phx.gbl microsoft.public.win2000.networking:47119
|X-Tomcat-NG: microsoft.public.win2000.networking
|
|How is the common maxiimun packet size communicated between the
|machines then?
|
|The reason I ask is that due to an issue with a VPN tunnel, we have to
|set the MTU to 1400 on one of our servers, but the others on that
|subnet are going to be 1500. I thought that the receiving machine
|with the smaller MTU was supposed to send back an icmp packet advising
|that it could only accept packets of size xxxx.
|
|I have seen the ping -f -l xxxx mentioned in various articles in the
|past about "tuning" the MTU on Dialups, ADSL etc, wherby the "DF set"
|response was the defining point which indicated the largest packet
|allowed between two points.
|
|After doing some netmon traces of a file transfer, I can see the
|packet size must of negotiated OK, but how is that done?
|
|Do you have any pointers for articles on how this works and why the
|result is different for machines on the same subnet?
|
|Thanks
|
|Clive
|
|
|On Fri, 05 Dec 2003 09:35:15 GMT, (e-mail address removed) (Joe Wu
|[MSFT]) wrote:
|
|>Hello Clive,
|>
|>Thank you for your post.
|>
|>Yes, it is an expected result. I have also reproduced this on my test
|>machines.
|>
|>The ping command just check the local MTU, but not the destination's MTU.
|>Therefore, when the packet's size is from 1373- 1472. It does not exceed
|>the local MTU and do not to be fragmented. So the ICMP packages are still
|>sent. Since it exceeds the MTU value on the destination, "Response timed
|>out" returns.
|>
|>Please let me know if anything is unclear. Thanks!
|>
|>Regards,
|>Joe Wu
|>Product Support Services
|>Microsoft Corporation
|>
|>Get Secure! -
www.microsoft.com/security
|>
|>====================================================
|>When responding to posts, please "Reply to Group" via your newsreader so
|>that others may learn and benefit from your issue.
|>====================================================
|>This posting is provided "AS IS" with no warranties, and confers no
rights.
|>
|>--------------------
|>|From: Clive <
[email protected]>
|>|Subject: MTU Size
|>|Date: Thu, 04 Dec 2003 18:49:40 +0000
|>|Message-ID: <
[email protected]>
|>|X-Newsreader: Forte Agent 1.93/32.576 English (American)
|>|MIME-Version: 1.0
|>|Content-Type: text/plain; charset=us-ascii
|>|Content-Transfer-Encoding: 7bit
|>|Newsgroups: microsoft.public.win2000.networking
|>|NNTP-Posting-Host: uk-15-96.client.oup.com 194.80.15.96
|>|Lines: 1
|>|Path:
|>cpmsftngxa07.phx.gbl!cpmsftngxa10.phx.gbl!TK2MSFTNGXA05.phx.gbl!TK2MSFTNGP
08
|>phx.gbl!TK2MSFTNGP09.phx.gbl
|>|Xref: cpmsftngxa07.phx.gbl microsoft.public.win2000.networking:47047
|>|X-Tomcat-NG: microsoft.public.win2000.networking
|>|
|>|
|>|If I have two Windows 2000 servers on the same subnet and set the MTU
|>|value of one of them to 1400 and leave the other at its default of
|>|1500, why do I get the following responses when pinging from the
|>|machine with the default MTU (with the -f switch)?
|>|
|>|upto 1372 bytes = reply OK (Expected)
|>|1373 - 1472 bytes = Response timed out (NOT Expecting this)
|>|1473+ bytes = Packet needs to be fragmented but DF set
|>|
|>|
|>|I was expecting a DF set response for everything over 1372 bytes
|>|(1400-28 byte overhead)?
|>|
|>|Why is there this blackhole in the middle?
|>|
|>|Clive
|>|
|
|
