I think you should create a normal bridge, and use that bridge with the guest.

say create a br0, put the physical interface in that br0, say eth0

you put the ip on br0, and the host will be reachable, then the extra ip on the guest, and that should work.


On Tue, Feb 25, 2014 at 1:40 PM, Tomasz Chmielewski <mangoo@wpkg.org> wrote:
I have a server with a hosting company, Hetzner.

The servers at this hosting company have a public IP, let's say,
A.B.C.D/255.255.255.x.

Additionally, one can order extra IPs like below:

1) additional subnet (let's say X.Y.Z.0 / 28)

2) single IP (let's say, E.F.G.H)


With additional subnet, assigning the IP to libvirt guest is simple:

- assign X.Y.Z.1 on the host
- assign X.Y.Z.2 to the guest, with X.Y.Z.1 gateway - done, works fine.


However, I'm having problems assigning a single IP to the guest (when
there is no extra subnet on the host, but just an extra IP):

- if I assign E.F.G.H/255.255.255.255 to the host, the IP is reachable
  from the internet

- if I assign E.F.G.H to the guest (with a proper gateway), the guest is
  not able to communicate with the internet.


Which networking mode should I use in this setup? Bridge, macvtap? I've
tried both, without success (i.e. in macvtap mode, arp who-has are
passed from the guest to the gateway, replies received on the host, but
they are not passed to the guest).

Has anyone ever set libvirt with a "typical" hosting provider, when
only an extra IP, from a non-host subnet, is available for a guest?


--
Tomasz Chmielewski
http://wpkg.org

_______________________________________________
libvirt-users mailing list
libvirt-users@redhat.com
https://www.redhat.com/mailman/listinfo/libvirt-users