Are Dupe ICMP Echos still an issue?

[wsanders]

I am seeing occasional duplicate ICMP Echos when I ping a host. My guest is Fedora 20 running in NAT mode under Virtualbox 4.3.8 on Windows 7:

Code: Select all

64 bytes from 10.192.0.2: icmp_seq=781 ttl=60 time=32.2 ms
64 bytes from 10.0.10.155: icmp_seq=781 ttl=60 time=31.2 ms (DUP!)
64 bytes from 10.192.0.2: icmp_seq=782 ttl=60 time=32.1 ms
64 bytes from 10.0.10.155: icmp_seq=782 ttl=60 time=42.1 ms (DUP!)

Once I even got three with the same sequence:

Code: Select all

64 bytes from 10.0.10.155: icmp_seq=762 ttl=60 time=29.9 ms
64 bytes from 10.192.0.2: icmp_seq=762 ttl=60 time=32.6 ms (DUP!)
64 bytes from 10.0.10.155: icmp_seq=762 ttl=60 time=29.5 ms (DUP!)

10.0.10.155 is the distant host, 10.192.0.2 is the Virtualbox "router" as I understand it. (I changed the "inside" subnet with ./VBoxManage.exe modifyvm "Fedora" --natnet1 "10.192/16" to rule out IP address conflicts, we use 10.0/16 as our network address space.) It's interesting that the echos with the pseudo-router as origin have about the same RTT as the "real" echo.

Is this still an issue with 4.3.8 or some side effect? I am running a Nagios instance in my guest, which uses ICMP to establish "connectivity" to determine whether a host is up, but I can always switch to a TCP 3-way handshake fpr that purposes.

Thanks in advance -w