Help with P2P Routing in China

General Discussion
1

I use NB on macOS and Ubuntu. Latency used to be fine but either due to a version update or some Chinese change, the latency is becoming crazy. My home is 2km from my office, and I used to be able to SSH and nomachine in over NB. But in the last week or so latency is >700ms and ssh is basically unusable and nomachine too. I assume NB cannot find a P2P route now so goes through a US server thus the latency. Does anyone have advice how to improve P2P discovery in this case or at least know what route is being selected?

2

Could be that your office made the ports/firewall stricter, are you the IT admin or a user?

3

I’m a User. I’m also having problems connecting to a totally different site around 20km away (not the same admin) but still in my city. I think this may be GFW at the ISP level is starting to disrupt UDP traffic (as VPNs that use UDP are also starting to become unreliable). Is there a way to debug the “hops” or how netbird is routing between two peers to understand why I have >1s ping and lots of drops?

4

It appears netbird will start to fail here is china as traffic gets selectively disrupted. The GFW is targeting IPv4 and sadly Netbird doesn’t seem to support any form of IPv6 routing.

5

You could check with

However, after a certain point this becomes a routing/firewall question. And this is so deep of a level I woudn’t be able to assist you with that.

6

Thanks I will check the troubleshooting steps. It is clear my connection is being relayed, my detailed log for from home to my work server shows:

Connection type: Relayed
  ICE candidate (Local/Remote): -/-
  ICE candidate endpoints (Local/Remote): -/-
  Relay server address: rels://streamline-vl-nrt-0.relay.netbird.io:443

I reinstalled my Zerotier client and on the same network using the Zerotier IPv6 route my ssh connection is like butter (the peer list shows DIRECT as the connection type, 10ms ping! compared to ~700ms Netbird). However I can also connect on IPv4 with low ping times so there is some other routing algorithm difference that is not dependent on IPv6.

I have also noticed recently I need to force netbird down && netbird up which can enable some local routes (so I am even having problems on the same local switch, where my local IP can connect (1ms ping) but netbird IPs do not respond at all…