the most likely cause of incorrect checksums is if you have tcp checksum offloading enabled on your nic in which case all captured packets that are sent by the host where you capture data will have an "incorrect" checksum since you capture the outgoing packets in the stack before the packet hits the nic where the checksum will be calculated.
you can disable checksum verification in the preferences for TCP.
since this is so common, current svn versions of ethereal will not flag these packets as checksum incorrect if the packet checksum is 0. if the packet checksum is not 0 and the reason for the incorrect checksums is still that you use offloading then your kernel might leak two bytes of data per frame that is sent (and captured by ehtereal) which might be a suboptimal setting which the kernel developers may want to look into.
Hello,
When analyzing traces collected in hp-ux using the command "nettl", ethereal shows it as a bad checksum, but looking the same frame in the hp-ux sing the command "netfmt" the checksum is ok.
I work for HP and I cannot duplicate this problem with my system, including analyzing some traces that the customer has sent to me. In my PC running windows XP, I can see the checksum correctly, but the same file at the customer displays "bad checksum". I have already also the customer to update the version of ethereal, but it didn't solve the problem. I can read the same file in many systems, with no problem, but the customer is able to duplicate it on many systems also.
For sure, if in the native hp-ux system the checksum is ok, the problem should be at ethereal, but I have no idea what else I can do to try to help the customer.
Do you have any about this problem?
Regards,
Luiz Sergio _______________________________________________ Ethereal-users mailing list Ethereal-users@xxxxxxxxxxxx http://www.ethereal.com/mailman/listinfo/ethereal-users
Powered by MHonArc 2.6.10