virt-manager: handling of bonding and bridge devices
On Thu, Jan 10, 2008 at 05:33:07PM +0100, Guido Guenther wrote:
> On Tue, Dec 18, 2007 at 04:34:14PM +0100, Guido Guenther wrote:
> > Hi,
> > currently virt-manager uses some linux specific heuristics to find out
> > if a device is part of a bridge. This immediatley fails if the
> > underlying device is not a physical device but e.g. a bonding device.
> > Hal doesn't seem to care about either bridges nor bonding devices. I
> > wonder where to fix this? Add another linux specific bonding hack
> > to determine if a eth devices is part of a bonding device which might be
> > part of a bridge? Would be easy enough to do but I wonder if it wouldn't
> > be better if hal would export this information in the first place?
> Hmm...no comments - would another hack that treats bridges on top of
> bond devices be o.k.?
The way virt-manager enumerates network devices only deals with physical
NICs which are optionally part of a bridge.
It is possible to extend this to deal with NICs that have then had a number
of VLANs setup, or are part of a bonded pair, but I don't have any machines
with such setups myself so not in a situation to write & test any code.
If someone wishes to add support for this, the place to look is the file
src/virtManager/connection.py. In particular the _device_added method
is called once for each physical NIC on the machine. You'd have to extend
that to also look for associated bonds, vlans. So patches welcomed...
|=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=|
|=- Perl modules: http://search.cpan.org/~danberr/ -=|
|=- Projects: http://freshmeat.net/~danielpb/ -=|
|=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=|
et-mgmt-tools mailing list