Hi Simon,
For the traffic labels, I assign then bridge.
Physical Bond Bridge
Eth1 -------> Bond0 -------> br0
/ Bond0.534 -> br0.534
Eth2 ---- /
eth1/2 are connected to trunk ports on my switch, bond0.534 is my
management VLAN/IP so the KVM host gets bond0.534 for
ssh/management/storage traffic.
When setting up the zone in CloudStack, assign the traffic label of
"bond0.534" for management, then for both Public and Guest, the label
"br0". Then I go create a shared guest network, provide the VLAN number and
IP addressing scheme.
I can get to the point of the System VMs running, however, they are never
able to talk to anything on the public or guest networks.
On Mon, Mar 14, 2016 at 1:44 PM, Simon Weller <[email protected]> wrote:
> C hris,
>
> What do you define as the traffic labels when you configured the network
> within the Cloudstack zone wizard?
> Can you provide some more information on what you setup?
>
> - Si
> ________________________________________
> From: Chris Watts <[email protected]>
> Sent: Monday, March 14, 2016 9:22 AM
> To: [email protected]
> Subject: Advanced networking with VLANs under KVM
>
> Hi Everyone, I having a helluva time trying to understand how cloudstack
> wants
> todo networking. Specifically when it comes to adding your zones/physical
> networks.
>
> My KVM hosts have 2 physical nics, bonded together as bond0, and
> they are VLAN trunk ports with about 30 VLANs coming in
> I have bond0 as a member of a bridge, br0
>
> I assign public/guest/management traffic to the physical network,
> management has
> no label (native vlan) then guest/public get br0
>
> Adding guest/public networks using the vlan ID's
> but nothing seems to work when the system VM's try to spin up....so I'm
> assuming I'm probably not following how cloudstack wants to interface with
> the
> physical network..but I'm suck.
>
> I need my cloudstack VM's to sit directly on the VLANs so they can talk to
> my
> other physical hosts.
>
> I've seen references to creating physical interfaces/bridges for each VLAN
> on the hypervisor (IE br0.111 for vlan 111, br0.112 etc) and using those,
> but I figured there was probably a better way.
>
> My assumption would be that I'd be able to tell cloudstack/KVM that the
> bridge is really a trunk and that cloudstack would take care of the tagging
> of traffic etc.
>
> Thanks
>
