On Tue, Mar 28, 2017 at 8:06 AM, Edward Haas <ehaas@redhat.com> wrote:

I think that Gianluca is referring to: https://gerrit.ovirt.org/#/c/74707

We seem to block the scenario where a slave of a bond has a VLAN on top of it.

Thanks,
Edy.

On Mon, Mar 27, 2017 at 5:04 PM, Dan Kenigsberg <danken@redhat.com> wrote:
I am not sure I understand your question, though it reminds me of

Bug 1372798 - Setupnetworks not removing the "BRIDGE=" entry in ifcfg
file when changing a untagged network to tagged

which Vdsm version do you have? can you share your supervdsm.log?

Yes, I mean a scenario where I have already in place a vlan on an interface that would become part of a bond created by oVirt

And I want to retain that vlan configuration.

I remember a thread where something similar was referred, with a parameter inside vdsm.conf, but I'm not able to find it...

I don't remember if that setting was targeted at ignoring the particular network interface as a whole, actually..

I would like to have eth3.100 and eth4.100 as single interfaces pointing to an iSCSI storage, but also use (throuch oVirt configuration) a bond0 device with eth3 and eth4 as slaves

So that VMs vlans with ID xx would be defined on bond0.xx and bridges with those bonds as interfaces

Does the gerrit change reported by you, Edward, go to correct the incompatibility?

In case in which version do I find it?

In case not possible can I sort of manually force this config and have oVirt accept it, even if not able to automagically create it?

Thanks,

Gianluca