Hi everyone,

I am experiencing a persistent networking issue with a fresh installation of Proxmox Mail Gateway on a Hetzner Dedicated Root Server. Despite having 29 other virtual machines running perfectly on the same host and vSwitch, this new PMG instance cannot reach its gateway.

The Setup (Example IPs):

• Environment: Hetzner Dedicated Server
• Network: Hetzner vSwitch (VLAN ID: 4000)
• Bridge: vmbr0 (VLANs are handled via tag=4000 in the VM NIC settings).
• VM Config: * IP: 192.168.1.85/24
  
  • Gateway: 192.168.1.1
  • VLAN Tag: 4000

What I have tried so far (The "Nothing Works" list):

1. Fresh Install: Performed a completely fresh installation of PMG from ISO; same issue.
2. vSwitch Check: Verified that the dedicated server is correctly "Tagged" to the vSwitch 4000 in the Hetzner Robot panel.
3. MAC Address Binding: Tried using a dedicated MAC address from Hetzner for the IP, and also tried without it. Neither worked.
4. NIC Models: Tested with both VirtIO and Intel E1000.
5. Comparison: If I assign the same configuration to a different test VM on the same host/vSwitch, it works immediately.

The Symptom:

• ARP Failure: ip neigh show consistently returns 192.168.1.1 dev ens18 INCOMPLETE.
• No Connectivity: The VM can ping itself, but cannot reach the gateway or external network.

It seems as though the ARP requests for this specific VM are not being answered by the gateway, even though 29 other VMs on the same host are working fine. It feels like the Hetzner vSwitch is "ignoring" this specific VM's traffic or MAC, but I cannot figure out why.

Has anyone encountered this specific issue with Hetzner vSwitch and Proxmox? Any debugging suggestions for the bridge or vSwitch layer?

Hi,

try to use only one VLAN method, either VLAN tagging on the VM NIC or a VLAN subinterface on the Proxmox host, not both.
Also, try to set a static MAC address for the VM (Hetzner vSwitch could drop traffic if MAC learning doesn’t match).