Bridge an existing vmbr2 to your VNet (The Layer-2 Patch) on Proxmox

I have 2 ports already bridged as vmbr0 and vmbr1. But now I want to have another vmbr2 as an additional virtual interface to be bridged as per my design logic. i also have implemented SDN. How can I achieve this? Is that really possible?

Yes, it is absolutely possible, and it is actually very easy to do!

When you create a VNet in Proxmox SDN, Proxmox automatically creates a real Linux bridge behind the scenes. If you name your VNet vmbr2, Proxmox will literally create a bridge named vmbr2 on all your nodes.
If you already created a VNet with a different name (like bridge1) and you must have it named exactly vmbr2 for your scripts or configuration, you have to follow my applied ways to achieve this.

But before that let me explain my set up first-

• I have 12 clustered nodes and have bridged 2 of my physical interfaced calling vmb0 and vmbr1
• I have also created an SDN zone, calling smartxln, with the MTU 1460
• I have create a Vnet- name bridge1, zone- smartxln, Tag-33 (Unchecked VLAN Aware)

Bridge an existing vmbr2 to your VNet (The Layer-2 Patch)

• Since I already have a physical/local bridge named vmbr2 that you created manually, and you have an SDN VNet named something else (let’s say vnet1), you can link them together natively using Proxmox’s configuration files.
• Since the GUI won’t let you cross-bridge two bridges easily, you can do it via SSH on each of the 12 nodes:
  • Open /etc/network/interfaces on a node.
  • It was achieved-by using a veth Link
  • Open /etc/network/interfaces on your nodes and modify your configuration to match this approach.

# 1. Define the virtual cable ends
auto veth-v2-to-b1
iface veth-v2-to-b1 inet manual
link-type veth
veth-peer-name veth-b1-to-v2
mtu 1450

# 2. Configure vmbr2 and plug one end of the cable into it
auto vmbr2
iface vmbr2 inet manual
bridge-ports veth-v2-to-b1
bridge-stp off
bridge-fd 0
mtu 1450

# 3. Use an 'up' hook to plug the other end into the SDN bridge1 when it becomes available
auto bridge1
iface bridge1 inet manual
post-up ip link set dev veth-b1-to-v2 master bridge1 || true
post-up ip link set dev veth-b1-to-v2 up || true

Apply the configuration
Save and close the file (Ctrl + O, Enter, Ctrl + X). Then safely apply the changes:

ifreload -a

This works safely for your Production Environment:

• No nested bridges: It satisfies the Linux kernel by using a virtual cable (veth) to bridge the gap instead of slamming two bridges together.
• Dynamic Binding: The post-up || true logic ensures that if bridge1 isn’t fully initialized on a node yet (like your issue on canode1), it will not crash your network reload. As soon as bridge1 activates, the script links them up.
• MTU Safety: Keeping everything explicitly pinned at mtu 1450 guarantees your unmanaged switch won’t drop packets due to VXLAN overhead encapsulation sizes.

Run ifreload -a on all nodes after updating the configuration