Hi all, I’m a homelab enthusiast who’s become the de facto IT person at my small company. We’re currently paying ~$250/month for Unifi’s UID Enterprise one-click VPN, and after great personal experience with Netbird at home, I started testing it as an always-on VPN solution for our traveling sales team. I’ve hit two issues I can’t get past and would really appreciate some help.
Setup Overview
Both my home and work environments are nearly identical:
| Home | Work | |
|---|---|---|
| Router | Unifi UDM Pro | Unifi UDM Pro |
| Hypervisor | Proxmox | Proxmox |
| VPN Host | Debian VM (DMZ) | Fedora VM (DMZ) |
| Stack | Docker + Traefik + Netbird self-hosted | Docker + Traefik + Netbird self-hosted |
| Routing Peer | Proxmox LXC running Netbird client | Proxmox LXC running Netbird client |
| Management URL | https://netbird.home.io | https://netbird.company.work |
| Port Forwards | 80, 443, 3478 → Docker host | 80, 443, 3478 → Docker host |
Both sites have a main LAN (internal zone) and a DMZ LAN. On my laptop I run two Netbird profiles — Home and Work — and can switch between them.
Problem 1 — No P2P When Connecting Cross-Site (Both Directions)
| Location | Profile Active | Result |
|---|---|---|
| At home | Home profile |  P2P connection |
| At work | Work profile | P2P connection |
| At work | Home profile |  Relayed only |
| At home | Work profile | Relayed only |
P2P works fine when I’m local to the network the profile belongs to, but always falls back to relay when connecting cross-site — in both directions. This makes me think something on the UDM Pros is blocking UDP hole punching. I do have IDS/IPS enabled at both sites, and I’ve already unchecked P2P under Active Detection, but the problem persists. Has anyone identified the specific UDM setting responsible for this?
Problem 2 — Severe Throughput Asymmetry Over Relay
Since both connections are relayed, I ran iperf3 to check throughput:
-
At work → Home profile (relay): ~150–170 Mbps
Acceptable -
At home → Work profile (relay): ~1–2 Mbps
Not acceptable
The setups are nearly identical, so I’m not sure what’s causing such a dramatic difference. The only variables I can identify on the work side are: it’s still running UID Enterprise alongside Netbird, and the Docker host is Fedora instead of Debian. Could either of those be the culprit, or is there something else I should be looking at?
Any guidance on either issue would be greatly appreciated!
*******************************************************
Yes I did you AI to clean up and organize my thoughts