Bridged Network on HA-cluster


#1

I am running OpenNebula 5.6 as a 3 node HA configuratioon, running on 3 metal machines were each have a different NIC (eno1, enp6s0 and enp3s0) .
Even if the name of the bridge network interface on the 3 hosts are the same (br0), the NIC:s are not. That means that I can not specify the physical device when I configure a Bridged network. As I haven’t configured it, I can not ping it from my network. Bellow is my network layout:

Internet --> Firewall/router --> Internal Network 192.168.1.0/24 --> OpenNebula Bridge (br0) --> OpenNebula Network 192.168.1.0/24

More information about OpenNebula network configuration and bridge is available at https://pastebin.com/GtP41STa

Is there a way/recommendations on how I can make this work?


(Ruben S. Montero) #2

So, I think I do not fully understand your issue:

  • For OpenNebula Bridged networks you can just define the bridge and add different interfaces to it. The bridged drivers can work only with the bridge name.

  • However this configuration is for the hypervisors and not the front-ends. In HA you should not run VMs in the front-ends. So the bridged networks in OpenNebula should not interfere with your HA setup.

  • I guess your problem can be the vip.sh script that sets the floating IP for the cluster, in this case it requires de same name. You can easily update the script to pick the ritgh name depending on the host/IP.I’m referring to this scrtipt.

Cheers