On Mon, Nov 02, 2009 at 08:44:26AM -0500, Laine Stump wrote:

With the patches I sent to netcf-devel last night, along with lutter's (or DV's, take your pick ;-) netcf patch to eliminate the segfault, my patches for reporting interface info now pass make check with no errors. I believe lutter will be making a netcf release tomorrow or so, after he gets back on solid ground.

There is one issue left that I'm unsure what to do about - while the xml returned from netcf for bond interfaces does list all the slave interfaces of the bond and their types (and any required info for that type), the parser for bond info in libvirt requires either an miimon or arpmode node inside the bond, eg:

<interface type="bond" name="bond0"> <bond> <miimon freq="100" updelay="10" carrier="ioctl"/> <interface type="ethernet" name="eth1"/> <interface type="ethernet" name="eth0"/> </bond> </interface>

I haven't been able to find a source of this information yet and, while it may be useful, I don't see it as necessary. Once again I'm proposing that we make these optional in the XML, such that this would be acceptable:

<interface type="bond" name="bond0"> <bond> <interface type="ethernet" name="eth1"/> <interface type="ethernet" name="eth0"/> </bond> </interface>

Once we do that, the netcf-reported state of all 4 supported interface types (ethernet, bridge, vlan, bond) will pass libvirt's parsing test.

That sounds OK to me - the miimon vs arpmode stuff is not critical. The interface topology is the only really important thing & you've got that working now. Daniel -- |: Red Hat, Engineering, London -o- http://people.redhat.com/berrange/ :| |: http://libvirt.org -o- http://virt-manager.org -o- http://ovirt.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: GnuPG: 7D3B9505 -o- F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|