connection selection behavior in Veeam 12.3 when using the Proxmox plugin

[peter.stenstrom]

Hi,

I’m running into what appears to be connection selection behavior in Veeam 12.3 when using the Proxmox plugin during restore operations.

Setup (simplified):

Veeam 12.3 running as a VM in a VMware environment
Veeam server has two NICs
NIC1: Management network (same subnet as vCenter/ESXi/proxies)
NIC2: Dedicated high-speed link (40 Gbps) towards Proxmox environment
Proxmox workers are reachable via NIC2 network

Problem:
When initiating a restore to Proxmox, the worker receives multiple IPs for the Veeam server and appears to attempt connections sequentially. It consistently tries the management NIC first, which is either slow or requires firewall traversal, resulting in a ~20 minute delay before falling back to NIC2.

What I’ve tested:

Preferred Networks → no effect on this behavior
Network Traffic Rules → no effect
BindToLocalInterface (registry) → partially helps but not consistently
Routing adjustments → only reliable method so far (forcing OS to remove alternative paths)

Understanding:
It seems the Proxmox plugin control channel does not respect Preferred Networks and instead relies on OS-level routing and/or internal IP selection logic.

Questions:

Is this expected behavior for the Proxmox plugin in Veeam 12.3?
Is there any supported way to control which source IP/interface is used for the initial worker connection?
Are there any plans to make the plugin respect Preferred Networks or similar configuration for control traffic?

Right now, the only deterministic workaround is to remove all alternative routes at the OS level, which feels a bit heavy-handed.

Appreciate any insight.

Regards /Peter