Configuring bridges - front-end and node config differences

I’m running an ON cluster on three servers under CentOS7, I’ve been fighting with the networking for a while now and I’m trying to understand how to configure networking so that VMs on any node can speak to VMs on any other.

The question I have is regarding the statement at the end of step 5 of this doc where it says:

Remember that this is only required in the Hosts, not in the Front-end. Also remember that the exact name of the resources is not important ( br0 , br1 , etc…), however it’s important that the bridges and NICs have the same name in all the Hosts.

So on the nodes I have to have the bridge and the physical network configured with the same names, this makes sense because you specify these in the virtual network configuration (BRIDGE and PHYDEV) and that is passed to every node, but it doesn’t make sense on the front-end server unless that physical machine is not being used as a node.

Ah, I think I’ve answered my own question there. There is a divide between the nodes that run the VMs and the front-end server that manages them. However, if you use your frontend server as a node as well then i must abide by the configuration requirements for the nodes as well.

I will leave this here as it may help others.

1 Like