EDACerton

Interesting... it seems like you must be missing one of the Tailscale log files. I've never seen that happen before, congratulations on the find I'm working on a fix now to make the interface handle that situation better. I should have it released to the preview channel later today. Edit: 2024.09.22 is released and should fix this.

Turn off the "Use Tailscale DNS settings" option.

What OS and browser are you using to access the UI?

That’s normal behavior then. The Tailscale address is going to work similarly to your local address — if you use a port to connect locally, you’ll use that port to connect via Tailscale.

This sounds like normal behavior. Do you normally need to add a port to conect to the Unraid UI?

I'd try the steps in the post right above this one to start. If that doesn't work, and you have access to the server itself: Install the "Plugin Diagnostics" app from CA. Break the server Log in using a keyboard, then run "plugin-diagnostics tailscale" That will put a diagnostics file in /boot/logs/ that you can retrieve after the server is working again.

Something is weird in your diagnostics... the "tailscale status" is blank, and I don't see any IPv4 entries from Tailscale in your routing table. I'd try this: Run "Erase Configuration" in the Tailscale settings. Delete the server from the Tailscale admin console. Restart the server Log in again and re-advertise the routes.

If you're just using Tailscale, anything that you'd be using to access Unraid remotely. There are other approaches too.... VPS + reverse proxy + Tailscale, Cloudflare tunnels... it all depends on what you're trying to accomplish.

Tailscale will bypass CGNAT on its own... just install Tailscale on your Unraid server and on whatever you're using to connect to it.

That would sound like you're losing your state storage... did you keep that configured? Auth keys are one-time-use unless created as reusable.

Request granted, filters can now be defined either plugin-wide or on specific files/commands. Examples are included on the Github README.

At present, it uses a more aggressive form of the anonymization used by the built-in diagnostics. I can look at adding some customization to that, though... that should be easy to do (especially at plugin level, file-level will take a little more work).

Yes. Tailscale has been using this for a while, but I figured that other folks might benefit too

Plugin Diagnostics is a utility plugin to help plugin authors collect diagnostic data when users request support. Plugin authors can activate diagnostics for their plugin by adding a diagnostics.json file to their installation folder in /usr/local/emhttp/plugins. A description of the file syntax is available via Github: https://github.com/dkaser/unraid-plugin-diagnostics

Please post diagnostics from the plugin settings.

You only need to run the command once. Remember to approve the route in the Tailscale admin console. The plugin is effectively a "new" computer from Tailscale's perspective, so the approval for the docker container doesn't get copied over.

FYI, unless every friend needs to share their subnet, you can just connect using the Tailscale IPs, the local network doesn't matter.

Thanks for the diagnostics... if you update to the latest release (2024.09.09), that issue should be fixed.

I don’t think you’re understanding my question. There are two different ways to access SSH via Tailscale. You can install Tailscale, leave “Tailscale SSH” off, and then connect to SSH using Unraid user/pass. You can alternately turn “Tailscale SSH” on, which replaces the built-in Unraid SSH server with one that uses Tailscale grants to control access. Tailscale SSH has its use cases, but is more complex to manage. What are you hoping to accomplish by using that alternative functionality?

Is there a particular reason that you're using Tailscale SSH instead of the SSH server included with Unraid?

Turn off --accept-routes.

It's easier for folks to help if you tell us what app you're trying to install However, in this case, the only apps in the CA feed with a MaxVer of 6.12.10 are the Python plugins that were maintained by dlandon. Those plugins are now deprecated, so will no longer be updated: https://github.com/dlandon/python3/pull/2 Almost all plugins in CA are maintained in Github, so you always have the option of going to the repository and making a fork, then manually installing it after making whatever modifications you want. (This should go without saying, but the consequences of this are on you -- if you force something incompatible to install, bad things could happen.) If all you want is Python, you can install it via NerdTools:

The log issue is just a cosmetic problem with the log page (it’s trying to display a compressed log because the current file is small). That will be fixed in the next release. For the alert, follow the instructions: deactivate flash backup, delete cloud backup, and reactivate the flash backup. The alert will trigger on each update until that is completed.

Did you apply the "workaround" from that other post (disabling "Unraid services listen on Tailscale IP")? That would stop the plugin from adding tailscale1 to the listening interfaces.