Tailscale not working as expected?

Hello, I’m trying to reach other devices in my tailnet from homeassistant as described in the Tailscale integration intro, but I’m not able to ping them.

[core-ssh ~]$ ping 100.78.182.56

PING 100.78.182.56 (100.78.182.56): 56 data bytes

^C

— 100.78.182.56 ping statistics —

3 packets transmitted, 0 packets received, 100% packet loss

Interestingly I can ping the homeassistant IP from other tailnet nodes, which makes homeassistant reachable from other nodes.

Now, is the description of the add-on incorrect? or is it the behaviour incorrect?

Just as reference I’m copying/paste what is written in the add-on description:

"
The Tailscale integration integrates the Tailscale API with Home Assistant; giving you the possibility to monitor and automate on the state of the devices in your Tailscale VPN network (Tailnet).

This integration DOES NOT make your Home Assistant accessible via Tailscale VPN remotely!

If you want to access your Home Assistant instance remotely, you will need to install Tailscale itself on your own. For instructions on how to do this, please consult the Tailscale documentation."

Thanks,
Raffaele.

1 Like

I Got Tailscale installed and Can access HA Remote via Tailscale, very cool!

But I was trying to access a remote sql server from HA via Tailscale, that seems to be a no go :-),

Any info on outgoing from HA via Tailscale ?

1 Like

Similar issue here, I can connect to home assistant via the tailscale vpn, I can ping home assistant from any other device but I cannot ping anything outside the local network from home assistant. Interestingly I can ping device inside the local network, via their tailscale 100.xx ip but not tailscale devices outside the local network…odd

1 Like

I have a similiar issue.

  1. Every member of the tailscale network can succesfully reach my home assistance instance by pinging it.

like Matz88:

  1. I cannot ping another member of the specific tailscale network from my home assistance (100.xxx).
  2. I can ping, though,
    a) another instance on a subnet, where my home assistance instance is running on and
    b) anything outside the tailscale network, i.e. 1.1.1.1, from my home assistance.

this solved my problem:

  1. I had to connect as a subnet router
  2. I had to use the --accept-routes flag

detailed procedures as described here: https://tailscale.com/kb/1019/subnets/

Hi @nico.z old post I know but do you remember how you added the accept-routes as the add-on doesn’t let you.
I’ve enabled allow subnets in the tailscale machine settings but that’s not doing it I think I need to add that but not sure how you add that with Hass OS
Thanks

Greetings,

I have been experiencing the same issues for at least 2-3 months. Cannot connect OUT of a tailscale docker container to the rest of my tailnet.

I had a fully working 3 node HASS setup between 3 geographically separate instances, all communicating properly over MQTT in early January. Each of these 3 instances are using a different type of install:

Node 1: Supervised - Tailscale 1.38.4 (addon)
Node 2: Docker Only - Tailscale 1.18.1 (docker)
Node 3: HASS OS - Tailscale 1.38.4 (addon)

Basically, Nodes 2 and 3 connect to the MQTT broker on Node 1. Sometime in the last 4 months I updated the Tailscale add-on on Nodes 1 and 2 and it hasn’t worked since. Note that I have not tried the --accept-routes solution not because it doesn’t work but because I was trying to figure out why (or when) it broke in the first place.

Troubleshooting:

In Node 3’s location first verified that a Windows tailscale client works (to rule out firewall issues), which it did. I then spun up another base Debian install in that location (call this Node 4). Installed the tailscale/tailscale container (as I had done with Node 2). Node 4 could not connect to any other tailnet node. Yuck! So guessed that it may not be a HASS OS issue after all but a docker issue? Then installed Tailscale 1.40.0 on the host. Worked fine. Yay!

I have a working install on Node 2 with Tailscale 1.18.1. So on Node 4 I uninstalled 1.38.4 from the host and then spun up 1.18.1 in Docker. It worked, indicating a downgrade on Node 3 may fix the problem. I then restored a 1.22.2 Tailscale image from a backup on Node 3 and that worked as well.

Conclusion:

While I don’t know which Tailscale image or HASS Add-on version caused the issue, I believe that there is an issue that was introduced recently that could be causing this issue. I’m guessing that most users are not creating outbound connections from HASS to other machines over Tailscale, so this issue may only affect a few of us. Fortunately for me I had backups to roll back on HASS OS.

I realize this is not a solution for all users, and that the add-on may still be broken. Good luck.

I’m running into the same issue as @conwaylw and @OH1MAC:

Home Assistant with the Tailscale add-on can’t connect to Tailnet devices, even though other Tailscale machines can successfully connect to the same Tailnet devices.

Did anyone arrive at a solution for this?

With the new update, I was trying to connect to my NAS over tailscale, but it seems I am also affected by this. Maybe these problems are related to this issue. But if someone got a solution, I would be eager to try it for myself!

Ny resolutions for this. I’m just starting with the whole HA route and Tailscale won’t work.

I can’t get it to work either. I have home assistant and my computer in my tail and they are both indicated as active, but I can’t connect to my HA instance from my computer. I activated the subnet route for the IP address of my HA instance, but there are three other routes that it identifies, but I don’t know what they are so I haven’t turned them on. Do I need to turn on the exit node?

Same issue here, anyone found a solution? Trying to connect from homeassistant to ssh other machines on my tailscale network. All other machines work fine just outbound homeassistant does not work. Is this a limitation of the addon?

Recently I encountered the same issue. I think I might have a solution.
Manually turning off the Userspace networking fixes the problem.
For me enabling and then disabling the Userspace networking mode switch, and after that saving the config and restarting the service fixed the problem.

8 Likes

This is the solution, thanks

I turned on then off users cape networking mode and this allowed me to fix my mqtt connection issues!

Thanks!!

This is what worked for me too! thanks I had wasted 3 hours on this lol

Oh bother. I thought my problem was solved.
My HA Tailescale keeps shutting down. Maybe I need to uninstall.

This worked for me as well, but very strange solution. Will be interesting to see if it now sticks or if this will have to be done periodically.

This hack works for me as well, thank you! It’s hidden under “Show unused optional configuration options”, and none of those options appear to be set by default. Enabling and disabling it in the UI therefore writes userspace_networking: false into the config. Looks very confusing in the UI though.