Dup icmp_seq on pings?

Running bridged networking on the VM's and thre are duplicate sequence numbers coming back.  Happens worst between VM's on the same server, though getting duplicates are occuring when pinging hosts outside the server.  The only thing I'm doing that might be considered unusual is using the bonding driver to do round-robin the host's dual nics.  Is anyone else experiencing this issue?   If so, what's the resoultion?

Here's my rig:

                                                                                ---> eth0

Vmware Server (23869) -> vmnet8 -> bond0 (nic) --->|

                                                                                ---> eth1

Here's an example ping from one of the VMs

64 bytes from 192.168.0.190: icmp_seq=3 ttl=64 time=0.672 ms

64 bytes from 192.168.0.190: icmp_seq=4 ttl=64 time=0.872 ms

64 bytes from 192.168.0.190: icmp_seq=4 ttl=64 time=0.889 ms (DUP!)

64 bytes from 192.168.0.190: icmp_seq=5 ttl=64 time=0.917 ms

64 bytes from 192.168.0.190: icmp_seq=5 ttl=64 time=0.948 ms (DUP!)