Each of my important VMs disks replicates every 15 mins to the second host as a “warm” recovery image. Also, during migration the VM hard drive and config are sent over the replication NICs I believe.

I suppose I don’t “need” the third NIC for replication, but old habits die hard.

Yep, everything is identical across the nodes and I’m using ZFS pools for VM storage.

I also have a dedicated NIC for cluster and replication traffic. So 3 NICs per host; WAN, LAN, and Replication

I run pfSense on a 2 node Proxmox “cluster” (cluster in quotes because I don’t have quorum for automatic failover). Each host has a dedicated NIC for the firewall’s WAN port attached to my modem which is in bridge mode. When I need to do maintenance on the node hosting the FW I do a live migration to the other node. I drop one ping during the migration.

Honestly, when I was designing it I didn’t think it would work…but here we are…lol.