Do you have a specific error from a log? I was not aware that CloudStack would look for interfaces w/ eth*, em*. In the code it just does "ifconfig -a" to list the devices. By creating a bond, the mac address CloudStack finds will probably change then I could imagine something could possibly fail.
Darren On Tue, Oct 22, 2013 at 1:39 PM, Marty Sweet <msweet....@gmail.com> wrote: > Hi Guys. > > I am planning on upgrading my 4.1.1 infrastructure to 4.2 over the weekend. > > When testing my 4.1.1 setup I ran across a problem where a TOR switch > failure would cause an outage to the management server. The agents use 2 > NICs for all management traffic using bonds. > When I tried to configure the management server to use a bond0 in simple > active-passive mode (like I use for my agent management network), > cloudstack-management would not start due to 'Integrity Issues', which at > the time I located back to a IntegitryChecker which ensures the interfaces > of eth* em* or some others were taking the IP of management server. > > My question is does this limitation still exist and if so, can it be > overcome by adding bond* to the list of allowed interface names and > compiling the management server from source? > I would love to hear input to this, it seems bizarre to me that it is > difficult to add simple but effective network redundancy to the management > server. > > For scenario basis, this is the basic redundant network setup I have for my > Agents: > 4x KVM Hosts all with 4 NICs - 2 bonds (Private/Public Traffic) > > Example Host: > ------------------Interconnect--------------- > TOR 1 --------- TOR 2 > --------------------- --------------------- > | Management | > | Tagged VLANs | > ---------------------------------------------------- > KVM Cloudstack Hypervisor > ---------------------------------------------------- > | Public Traffic | > | Tagged VLANS | > | LACP Aggregation | > ---------------------------------------------------- > Core Router > ---------------------------------------------------- > > There are also LACP links with STP rules between the TOR switches are the > core device to allow for interconnect failure so the TORs do not become > isolated, but I have excluded that for simplicity. > > > I would have thought it would be easy to create a bond for my management > node and connect the two NICs to both the TOR switches, but that didn't > work in 4.1.1 due to my reasons above. > > Thanks! > Marty
