[ovs-discuss] Re : Re : A few question and experience about Open vSwitch

[DarkBls]

Thank for the reply.

Yes of course it's unusual. But mainly for testing purpose. I tested both taged and untagged port.

I want to use the kernel bonding implementation for stability and policy in my firm. Furthermore, the kernel implementation give for now more control on it. (LACP, Fail over etc..). Is the the feature officially supported ?



----- Message d'origine ----
De : Ben Pfaff <blp at nicira.com>
À : DarkBls <darkbls at yahoo.com>
Cc : discuss at openvswitch.org
Envoyé le : Mardi, 6 Octobre 2009, 18h17mn 45s
Objet : Re: Re : [ovs-discuss] A few question and experience about Open vSwitch

DarkBls <darkbls at yahoo.com> writes:

> [root] # cat ovs-vswitchd.conf
> bridge.mybr.port=bond0
> bridge.mybr.port=tap0
>
> #Config VLAN
> #vlan.eth1.trunks=210
> #vlan.tap-alcide.trunks=210
> #vlan.eth1.trunks=0
> #vlan.tap-alcide.trunks=0
>
> #Config en non tague
> vlan.bond0.tag=210
> vlan.tap0.tag=210
> #vlan.eth1.tag=0
> #vlan.tap-alcide.tag=0

This is an unusual configuration.  First, the bridge has only two
ports.  In itself, that's not odd, but both of them have the same
tagged vlan.  That will have no useful effect in practice,
because there are no trunk ports.

Also, it looks like you have configured a bond using the Linux
kernel bonding implementation.  That might work, but we have not
tested it.  Usually, I would expect that you would use the Open
vSwitch implementation of bonding, which can be configured by
adding settings to the configuration file.  The
ovs-vswitchd.conf(5) manpage explains how.