Hi all.
Some days ago i have done some tests using the user mode network stack
with multiple network interfaces and i noticed some strange behaviours
that i would report to you. I use QEMU 0.7.2 under Debian Sid, using
the package provided by the distribution, and the guest OS is Debian
testing.
I've observed the following things using "-nics n":
1) all the networks interfaces are detected correctly during the boot
phase: eth0, eth1, ... ethn-1, ethn;
2) the DHCP server doesn't work with the first n-1 interfaces but it's
OK with the last. For example, with "-nics 4" DHCP works only with
eth3 and not with eth0,1,2 and with "-nics 2" works only with eth1.
To try out, in the guest system leave all nics unconfigured, then
launch the DHCP client on each nic separately. For example, lauching
dhclient 2.0p15 on the first n-1 interfaces, after some wait, it
terms saying "No DHCPOFFERS received". With ethn works.
3) in general, seems that the network (not only DHCP) doesn't work
correctly with first n-1 interfaces. Trying to leave all network
interfaces unconfigured and trying to configure each of them
individually with IP 10.0.2.20, netmask 255.255.255.0, 10.0.2.2
as gateway and 10.0.2.3 as DNS:
- all nics can ping its IP address;
- only ethn can ping 10.0.2.2, the others gets "Destination host
unreachable" from the IP of the network card;
- none of the nics can ping 10.0.2.3: all gets "Destination host
unreachable" from the IP of the network card, except ethn, that
doesn't obtain responses (but the name resolution seems to work);
- only ethn can go out the 10.0.2.x network (internet, for example).
Anyone has experienced such behaviour that appear to me as bug(s)? It's
a known thing (googling around the mailng-list and internet i have
found nothing)?
Cesare.
_______________________________________________
Qemu-devel mailing list
Qemu-devel@nongnu.org
http://lists.nongnu.org/mailman/listinfo/qemu-devel
