Packet loss due to MAC multicast setup of switch?

[Ralph Grothe]

Hello networkers,

we have a couple of HP-UX boxes which are still furnished with 10MBit/s LAN interfaces.
On these boxes we experience some loss of (e.g. ping sent ICMP) packets.
When an hpterm is started on these boxes this process hangs, and top lists over 50% CPU for it.
Our system analyst suspects that the MAC multicast of a switch could be the cause for the packet loss.
I have no idea why one would want to configure a switch to use the same virtual MAC address on several of its ports.
I only think to have read somewhere that this could enhance video streaming (we don't do such in our LAN).
So my dumb posting is also a plea for some short explanation of this MAC multicasting technique, and its possible impact on 10MBit/s interfaces of hosts.
The HP-UX boxes in the same subnet which are furnished with 100MBit/s LAN interfaces don't have any problem with the active network components.
Is there a way to track the cause for packet losses (maybe by use of packet sniffers, e.g. nettl, tcpdump etc.)?
Is there a way to change the operational mode of LAN interfaces through lanadmin or similar commands in order to cope with multiple MACs?

[rick jones]

indeed i am at a loss as to how multicasting could be at issue here unelss perhaps there is some sort of loop in your switches?

you might start looking at the lanadmin stats of your NIC(s) and see what sort of errors they are reporting.

you could try a tcpdump trace, but that is "post NIC"