Currently Being Moderated

[Flybird74]

Hello, I have switch1: 3750G 24TS-1U and switch2: 3750G 24TS-S.

Port 25 on switch1 is connected to Port 26 on switch2, both ports are set to be trunk ports.

I have two windows 2008 servers: server1 and server2. Server1 is connected to port 18 on switch1 with IP 10.10.6.11,

server2 is connected to port 23 on switch2 with IP 10.10.6.12.

System MTU on both switches are set to be 1500, MTU on the two windows 2008 servers are also 1500.


OK, now the problem:

If I logon to server1(10.10.6.11), and issue the following command:


ping -f -l 500 10.10.6.12, return is OK
ping -f -l 600 10.10.6.12, timeout!

I get timeout on the second command with packet size 600, why? Can anyone please help to explain what is my problem here?


Thanks.

 

[VinceWhirlwind]

Connect the servers directly to each other with a crossover lead to see if you can eliminate the switch from the equation.

 

[Flybird74]

We have tried that approach and we are pretty sure the two switches caused this problem.

 

[VinceWhirlwind]

OK, how about using different devices (like two laptops) to try the same ping between each other?

As far as the switches go - show us the config. It seems unlikely that a switch would be dropping MTUs bigger than 500-odd. Even if they have IP routing configured, they will simply switch frames staying on the same subnet.

This is where a packet capture could be good.

What about the firewall on the Windows servers - you haven't recently changed it to block all ICMP or something have you? You need ICMP to work for MTU size negotiation to work.