EDACerton

Just to clarify here... this is all happening via Settings -> Tailscale? Please provide plugin diagnostics: https://edac.dev/unraid/plugin-diagnostics/usage/

What do you get if you run this on the server? wget https://pkgs.tailscale.com/stable/tailscale_1.82.0_amd64.tgz

The video is out of date, correct instructions are here: https://edac.dev/unraid/tailscale/plugin/

Do you have "Accept Routes" turned on? If so, turn it off. Otherwise, please follow the instructions in the pinned post and provide plugin diagnostics.

Thanks for mentioning this. It looks like that’s an issue with an Ubuntu kernel update, not Tailscale 1.82 (it also affects things like k8s).

I won’t be able to look at diagnostics until tomorrow, but it seems like your plugin is still attached to the old tailnet. (The plugin and the “Use Tailscale” option in Docker are two different things that don’t affect the each other.) I would do two things: 1. Use the “Erase Configuration” option in the advanced Tailscale settings. 2. Make sure “Accept DNS” is disabled.

The plugin works the same way in v7. The "Tailscale integration" is actually a setting in Docker containers that will install Tailscale directly into the container (which has some use cases, but the way the "Use Tailscale" setting does that is bad for a variety of reasons, and there are better ways to do it -- see https://edac.dev/unraid/tailscale/docker-options/ ). Yes, the settings are all preserved.

Something in 6.11 doesn't like the latest update. I've updated the plugin to prevent the update from installing before 6.12. In the meantime, if you remove the Tailscale plugin, then go to Plugins -> Install Plugin and enter this: https://raw.githubusercontent.com/dkaser/unraid-tailscale/2025.03.05/plugin/tailscale.plg That will roll you back to the previous version of the plugin.

25% of installs are still on some verson of 6.12... I guess those folks should be grateful that I manage the plugin and not you 🤣

Please provide plugin diagnostics, following the instructions in the pinned post. Also, please be aware that I will be dropping support for 6.11 in an upcoming update (it's now 2 versions behind), so you might want to look at updating your server.

This is probably because of your tailscale_pihole instance that's running on br0. I'm guessing that's set up as your DNS server in the tailscale admin console (probably by the local IP address). I'd probably set up a Tailscale sidecar for that container, then put the Tailscale IP in the admin console: https://edac.dev/unraid/tailscale/docker-sidecar/

Yes, you have "Accept Routes"/"Allow Tailscale Subnets" turned on... you should disable that.

I need the Tailscale diagnostics (see the pinned post). Those include tailscale logs, additional information about tailscale settings, etc. However, in the meantime — 99% of the time, the symptoms that you described are a result of enabling the “Use Tailscale Subnets” option in the advanced settings of the plugin. If that’s on, try turning it off.

Thanks for the suggestion, this will be included in the next release

It looks like tailscaled isn't starting on your system... what happens if you run this from the console? /usr/local/sbin/tailscaled -statedir /boot/config/plugins/tailscale/state -tun tailscale1

The plugin is currently being held on 1.78 because of a problem introduced in 1.80. Tailscale is aware of this, I’m trying to get into from them on when to expect a fix. The preview version of the plugin includes 1.80 if you really need to update to that version.

Yes, this is one of the key benefits of Tailscale -- you don't have to open ports in the router. In general, the plugin is sufficient, unless you want to share specific containers with other tailnets. (e.g., if you wanted to give a friend access to a single container, but not be able to see everything). No, you don't need a separate network in Unraid. No, you don't have to set Tailscale up as a subnet router -- you can connect to your services via http(s)://tailscaleIP:port/, just like you would http(s)://localIP:port/.

FYI -- adding /usr/local/emhttp/plugins/tailscale/bin to the system PATH is unnecessary... tailscale and tailscaled get linked into /usr/local/sbin/ during the plugin installation.

This shouldn't be the case... disabling "Use Tailscale DNS" on the Unraid server shouldn't affect the DNS resolution on other devices, unless you're doing something weird like trying to run Adguard as a container on the Unraid server, then using the host Tailscale to resolve DNS for everything upstream... If that's what you're doing, the better way is to set your Adguard server as the DNS server in your Tailscale admin console, then enable Tailscale DNS on the clients.

I'll take it

Make sure that you have Tailscale installed and connected on the device you're using to access the Unraid WebGUI. If that doesn't work, you could try logging out on your client and logging back in. (Yes, this is annoying, and I actually just opened up a feature request on the Tailscale repo to see if we can get rid of the extra step.) Alternately, everything that you can do in the GUI you can also do on the console using "tailscale set" commands. Finally, my name isn't Ed

That seems like a fairly simple container, it should work just fine at https://name.tailnet.ts.net/ . Did you add a port when you configured the labels?

That would be where I would start... you'll need to enable that from the console: tailscale set --exit-node-allow-lan-access I'll have to add that as an option in the settings in a future update.

Delete the device from the Tailscale admin console, then use the "Erase Configuration" button in the plugin settings (switch to advanced mode). That will give you a fresh start.

Go to the console and run this: wget https://pkgs.tailscale.com/stable/tailscale_1.78.1_amd64.tgz and see what it says.