I'm not sure what I'm looking for there? According to the site I sent a link to, Cat5e or Cat 6, which you have. The interfaces are running at gigabit rate and there are no errors logged, so the configuration appears to be OK, to me at least? I thought this connection was quite a bit slower than the other interface connected to a switch? You are sure it is running at a gigabit rate? And two similar systems with similar cables behave the same, which seems to eliminate specific hardware/cable faults? Actually, it only says that everything is working as designed. According to the link I sent you, gigabit uses a regular patch cable for a machine to machine connection and not a crossover cable. If the cables are similar to 2 similar computers, and exhibit similar problems,
Dave Howorth wrote: then I would try to change a variable, if it is worth it to you to find the problem (since it works through the switch). Since it is easy to swap cables (which may cause some config problems), does it work as fast through the problem connection's NIC if connected to the switch? If it is just as fast, it must be the cable (is it routed near inductive noise?). Is it the same slowness if the direct connection cable is plugged into the NIC that was connected to the switch?
The cables are cat 6. That should be the best. These are just some thoughts. I may be way off target. HTH, but YMMV.
-- Joe Morris Registered Linux user 231871