Replies: 3 comments
-
The primary reason is that running in kernel level networking mode (i.e. A bunch of this comes down to the default routes that the UDM and Tailscale install in kernel level mode and the resulting connectivity issues between |
Beta Was this translation helpful? Give feedback.
-
Understood. That is a good reasoning. I got better understanding now. Appreciate you comment and detailed description of the culprit with kernel level tailscale. Thank you. |
Beta Was this translation helpful? Give feedback.
-
Sorry to bring this old topic. I run tailscale on UDM SE without problem using the default userspace mode. Devices in the tailscale network can reach machines in the advertised subnets but not the other way around. Obviously, it's normal since the tailscale routes aren't installed in UDM because it's userspace. I was going to switch to kernel mode and use tunnel interface and routes but I understand it's not recommended for different reasons. Now my question is, is there a way to make this work with userspace mode? Bonus question: when in usersapce, I suppose the UDM source NAT the connections to the machines not running tailscale? |
Beta Was this translation helpful? Give feedback.
-
As in the title. I wonder because I want to create a masquerade to use it as exit node, but I don't know how.
Also wonder about the good reasoning for using
--tun=userspace-networking
instead regular interface and /dev/net/tun tailscale0 interface.Similar topic was mentioned here: #61 (comment)
Beta Was this translation helpful? Give feedback.
All reactions