Maybe not Tailscale-related but having a hard time finding where to post this, or even googling it.
Tailscale SSH to this Ubuntu server works fine either via tailnet IP or via MagicDNS hostname.
But doesn’t work at all for local/LAN connections. Getting a connection timeout.
I’ve tried googling this, but expectedly - all troubleshooting relates to fixing Tailscale SSH, not the other way round All other devices that are on the same LAN & Tailnet are fine. I can ping the server and telnet the port. I’ve checked that SSH is enabled for my user, and have re-enabled root SSH for troubleshooting without success.
I’m having this same issue. From outside my internal network, I can use tailscale ssh to access a WSL2 environment with tailscale configured. When I’m on my internal network, I cannot.
On Debian I have the same (or very similar) issue. SSH on LAN stops working randomly. Then suddenly it starts working again.
When its not working, I can still connect to the LAN IP, from another routed taillscale src.
What i’ve found, is that the local network on the tailscale client added to the tailscale route table (table 52). Its shown by doing “ip route show table 52”. (e.g. 192.168.1.0/24). This network is advertised to tailscale.
The easiest fix i’ve found is just to go an add an new DNS, and then simply delete again. It will make tailscale reload its config, and thereby removing the entry in the tailscale route table.