I'm trying to figure out the best way to ensure corosync has its own network. Please let me know if any of these points weren't ideal and why:
- I created a VLAN with no gateway (just a space on my switch with a unique subnet) 10.5.0.X and set all 8x of my nodes to addresses on that subnet.
- I created a Linux Bridge for the VM + Management Network, but did NOT create a bridge for the cluster NIC
- How can I tell if corosync and migration are really running though the 10.5.0.X network?
- On my previous cluster, I configured the cluster IPs and Management IPs on the same subnet (different NICs) and one day I unplugged the cluster NIC and the cluster continued on like nothing happened. It was really disappointing to see that all traffic was going through the Management NIC.