Hey all,

	I've run into a problem with running a Fedora Core 6 VM (on an FC6
engine).

	I'm using the veth devices and this seems to work for my CentOS VMs
(4.5 and 5) and Fedora Core 5 systems but the eth interfaces fail to
configure in an FC6 VM and it appears to be some problem with the sysfs
entries.  When I run ifup on the interface, this is what I get...

[root@amethyst /]# ifup eth0
sysfs attribute broadcast: No such file or directory
Error, some other host already uses address xxx.xxx.xxx.xxx.

	Looking through the "ifup-eth" script, the immediate problem is the
arping command is failing with the "sysfs" error and that failure is
reported out by the script as "some other host already...." and then
exiting.

	If I comment out the "arping" section, I STILL get a "sysfs attribute
broadcast: No such file or directory" error from another arping command
further down but the interface does get configured with the IP address.
If I comment both stanzas containing arping, the errors go away and the
interface appears to configure correctly, I've just lost some duplicate
IP address error detection.

	It also fails (silently, this time) if I set it up with dhcp as the
boot protocol.  The interface is "up" but has no IPv4 address assigned
to it.  Fortunately, they do get IPv6 addresses assigned to it and
autoconfigure properly (one of the primary reasons why I choose veth
over vnet), which is why I hadn't noticed it much earlier (I largely
manage my systems over IPv6).

	I only had one FC6 VM earlier, for testing, and it was only accessed
over IPv6, so I never noticed the failure before.  The new one, on a
different VE0, actually needed to be accessible over IPv4 and THATS when
I noticed the problem and ran it down to the arping problem and sysfs.

	Looking in /sys on the host machine and the vm I see these differences:

VE0:

[root@remus ~]# ls /sys/class/net/eth0
address    carrier  features  iflink     operstate   tx_queue_len  weight
addr_len   device   flags     link_mode  statistics  type
broadcast  dormant  ifindex   mtu        subsystem   uevent

VM:

[root@amethyst ~]# ls /sys/class/net/eth0
statistics  subsystem  uevent

	Not good...  Looks like some things are missing from the veth device
drivers for sysfs.

	Mike
-- 
Michael H. Warfield (AI4NB) | (770) 985-6132 |  mhw@WittsEnd.com
   /\/\|=mhw=|\/\/          | (678) 463-0932 |  http://www.wittsend.com/mhw/
   NIC whois: MHW9          | An optimist believes we live in the best of all
 PGP Key: 0xDF1DD471        | possible worlds.  A pessimist is sure of it!