Tailscale startup loop

peter.suba · December 12, 2025, 10:49pm

I have 3 Gl.Inet routers - with Tailscale running on all of them - will, they were.

I have made several config updates which ended up with one of them (a Flint 2 GL-MT6000) now in a situation where I can not use Tailscale any more. Other aspects are (as far as I checked) working.

It was working with Tailscale using my user login to the tailnet, and I have used admon’s script to have an updated version of tailscale recently (sorry, can’t remember the exact version but I think it was around 1.88). It was all working without any issues.

I got to the place where I have made 3 changes, not entirely sure of the order any more:

I have moved the device’s tailscale authentication from a user authenticated to a tagged device (note: I have done this on the other 2 devices, a Flint 3 and a Beryl - both work without issues)
I have updated the tailscale version again using the admon script above (current 1.92.2 but not sure if this was the exact version that caused the symptoms described, see below)
I have updated the firmware to v4.8.3 - I am fairly sure this was the point where my issues started occuring but can’t be 100% sure now

Symptom: when I am trying to bring up tailscale (starting tailscaled), it seems (on the web interface) as if it came up, but it keeps flicking off then on again.

At this point, after a lot of troubleshooting and steps that I took to try and fix it, it now flicks up the warning that I need to follow the link to authorise the device, but clicking on the link gives an error message - this is fully understandable because when I look in the logs it is clear that the tailscale daemon never comes up fully and just tries again every 3 seconds.

A relevant section of the log is below:

Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 logtail started
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 Program starting: v1.92.2-t95a957cdd-tiny.by.admon.1098, Go 1.25.5: []string{"/usr/sbin/tailscaled", "--port", "41641", "--state", "/etc/tailscale/tailscaled.state"}
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 LogID: 7f7686719d458b4f4e252171824dcf45543ed96bfd6ebbaa55555b7fa2101571
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 logpolicy: using system state directory "/var/lib/tailscale"
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 dns: [rc=unknown ret=direct]
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 dns: using "direct" mode
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 dns: using *dns.directManager
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 cleanup: list tables: netlink receive: invalid argument
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 wgengine.NewUserspaceEngine(tun "tailscale0") ...
Fri Dec 12 21:45:46 2025 daemon.notice netifd: Interface 'tailscale' is enabled
Fri Dec 12 21:45:46 2025 daemon.notice netifd: Network device 'tailscale0' link is up
Fri Dec 12 21:45:46 2025 daemon.notice netifd: Interface 'tailscale' has link connectivity
Fri Dec 12 21:45:46 2025 daemon.notice netifd: Interface 'tailscale' is setting up now
Fri Dec 12 21:45:46 2025 daemon.notice netifd: Interface 'tailscale' is now up
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 dns: [rc=unknown ret=direct]
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 dns: using "direct" mode
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 dns: using *dns.directManager
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 link state: interfaces.State{defaultRoute=eth1 ifs={br-guest:[192.168.9.1/24] br-lan:[192.168.15.10/24] eth1:[192.168.1.136/24]} v4=true v6=false}
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 magicsock: disco key = d:561e08b20c556b82
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 Creating WireGuard device...
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 Bringing WireGuard device up...
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 Bringing router up...
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 router: portUpdate(port=41641, network=udp6)
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 external route: up
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 router: detect nftables rule: firewall mode "nftables" not supported: create nat table: get table: get tables: netlink receive: invalid argument
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 router: nftables rule count: 0, iptables rule count: 228
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 router: iptables is currently in use
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 router: disabling tunneled IPv6 due to system IPv6 config: disable_ipv6 is set
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 Clearing router settings...
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 Starting network monitor...
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 Engine created.
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 pm: migrating "_daemon" profile to new format
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 envknob: TS_DEBUG_FIREWALL_MODE="auto"
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 logpolicy: using system state directory "/var/lib/tailscale"
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 got LocalBackend in 68ms
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 Start
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 ipnext: active extensions: portlist
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 router: portUpdate(port=41641, network=udp4)
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 Backend: logs: be:7f7686719d458b4f4e252171824dcf45543ed96bfd6ebbaa55555b7fa2101571 fe:
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 Switching ipn state NoState -> NeedsLogin (WantRunning=false, nm=false)
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 blockEngineUpdates(true)
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 wgengine: Reconfig: configuring userspace WireGuard config (with 0/0 peers)
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 health(warnable=wantrunning-false): error: Tailscale is stopped.
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 wgengine: Reconfig: configuring router
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 wgengine: Reconfig: user dialer
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 wgengine: Reconfig: configuring DNS
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 dns: Set: {DefaultResolvers:[] Routes:{} SearchDomains:[] Hosts:0}
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 dns: Resolvercfg: {Routes:{} Hosts:0 LocalDomains:[]}
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 dns: OScfg: {}
Fri Dec 12 21:45:46 2025 user.notice firewall: Reloading firewall due to ifup of tailscale (tailscale0)
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 tailscaled got signal terminated; shutting down
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 control: client.Shutdown ...
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 control: updateRoutine: exiting
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 control: mapRoutine: exiting
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 control: authRoutine: exiting
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 control: Client.Shutdown done.
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 ipnext: work queue shutdown failed: execqueue shut down
Fri Dec 12 21:45:46 2025 daemon.notice netifd: Network device 'tailscale0' link is down
Fri Dec 12 21:45:46 2025 daemon.notice netifd: Interface 'tailscale' has link connectivity loss
Fri Dec 12 21:45:46 2025 daemon.notice netifd: Interface 'tailscale' is now down
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 monitor: ip rule deleted: {Family:2 DstLength:0 SrcLength:0 Tos:0 Table:254 Protocol:0 Scope:0 Type:1 Flags:0 Attributes:{Dst:<nil> Src:<nil> Gateway:<nil> OutIface:0 Priority:5210 Table:254 Mark:16711680 Pref:<nil> Expires:<nil> Metrics:<nil> Multipath:[]}}
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 monitor: ip rule deleted: {Family:2 DstLength:0 SrcLength:0 Tos:0 Table:253 Protocol:0 Scope:0 Type:1 Flags:0 Attributes:{Dst:<nil> Src:<nil> Gateway:<nil> OutIface:0 Priority:5230 Table:253 Mark:16711680 Pref:<nil> Expires:<nil> Metrics:<nil> Multipath:[]}}
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 monitor: ip rule deleted: {Family:2 DstLength:0 SrcLength:0 Tos:0 Table:0 Protocol:0 Scope:0 Type:7 Flags:0 Attributes:{Dst:<nil> Src:<nil> Gateway:<nil> OutIface:0 Priority:5250 Table:0 Mark:16711680 Pref:<nil> Expires:<nil> Metrics:<nil> Multipath:[]}}
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 monitor: ip rule deleted: {Family:2 DstLength:0 SrcLength:0 Tos:0 Table:52 Protocol:0 Scope:0 Type:1 Flags:0 Attributes:{Dst:<nil> Src:<nil> Gateway:<nil> OutIface:0 Priority:5270 Table:52 Mark:0 Pref:<nil> Expires:<nil> Metrics:<nil> Multipath:[]}}
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 RAW-STDERR: ***
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 RAW-STDERR: *** Lines prefixed with RAW-STDERR below bypassed logtail and probably come from a previous run of the program
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 RAW-STDERR: ***
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 RAW-STDERR:
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 RAW-STDERR: [
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 flushing log.
Fri Dec 12 21:45:46 2025 daemon.err tailscaled[13270]: 2025/12/12 21:45:46 logger closing down
Fri Dec 12 21:45:46 2025 daemon.notice netifd: Interface 'tailscale' is disabled
Fri Dec 12 21:45:46 2025 user.notice firewall: Reloading firewall due to ifdown of tailscale ()

This more or less keeps repeating every 3 seconds.

What I have tried:

I have tried downgrading the tailscale version and the updating again (see why I can’t remember the exact version)
I have deleted /etc/tailscale/tailscaled.state (and the restarted etc)
I have tried authenticating the machine manually through registering the node through ssh using an auth key (manually starting tailscaled then using the tailscale command etc) - this has made the device actually appear on the tailscale console, but made no difference really to the symptom as far as when I switched “on” tailscale on the web UI, it kept being in the startup loop. I eventually cleared the machine from the tailscale console, removed the state on the router again etc.
I have looked at LUCI and compared the setup with a Flint 7, which is not a full match of course, and the tailscale device is actually up on the flint 7, but the only thing I could really see that was weird is that flint 7 has a tailscale0 device, whereas flint 6 has a tailscale interface (but no device). I have also manually set up a zone in the firewall on flint6 to be the same as flint 7 - just after suspecting the issue may be related to firewall (but I know the gl_tailscale script does play with firewall rules to a degree).

I am close to giving up.

Ideally, I’d like to avoid factory reset - partially because I am not physically with the router and I’d like to avoid needing to do something that needs me being there. Also to avoid having to re-configure everything as everything else but tailscale seems to be OK. I’m just trying to get to a point where tailscale would be back to as if it was never touched… Or fix forward, but it has not been successful so far of course.

I did not remove gl_tailscale and tailescale packages yet as I’ve not seen many success stories on this forum on that approach.

Any ideas would be appreciated.

admon · December 12, 2025, 10:52pm

If you run a firmware upgrade, you need to re-run my script again to get the newest tailscale version since downgrading can cause issues (which might happen during a firmware update)

Instead of factory reset you can restore your old Tailscale version as well:

wget -q https://get.admon.me/tailscale -O update-tailscale.sh ; sh update-tailscale.sh --restore

Might resolve your issue for now.

Maybe choosing a different Tailscale version might help as well:

wget -q https://get.admon.me/tailscale -O update-tailscale.sh ; sh update-tailscale.sh --select-release

peter.suba · December 13, 2025, 12:29am

Factory resetting the tailscale version did not help unfortunately...

I got:

ℹ️  Stopping tailscale
ℹ️  Restoring tailscale binary from rom
✅ tailscale binary restored
ℹ️  Restoring tailscaled binary from rom
✅ tailscaled binary restored
✅ gl_tailscale script restored
ℹ️  Removing entries from /etc/sysupgrade.conf
✅ Tailscale restored to factory default.
⚠️  Restarting tailscale might or might not work
⚠️  You might need to re-authenticate your device
ℹ️  Starting tailscale
✅ Script finished successfully. The current tailscale version (software, daemon) is:
1.80.3
  go version: go1.23.6

After this, I have exactly the same effect.

When switching tailscale on, it shows the “popup” for device linking:

but when clicking on the link, it fails to get the device binding link (in other words, it does exactly what it did before the downgrade).

There is a lot less in the logs as there is no tailscaled entries, but it boils down to the same symptoms, these keep repeating:

ri Dec 12 23:53:51 2025 daemon.notice netifd: Interface 'tailscale' is enabled
Fri Dec 12 23:53:51 2025 daemon.notice netifd: Network device 'tailscale0' link is up
Fri Dec 12 23:53:51 2025 daemon.notice netifd: Interface 'tailscale' has link connectivity
Fri Dec 12 23:53:51 2025 daemon.notice netifd: Interface 'tailscale' is setting up now
Fri Dec 12 23:53:51 2025 daemon.notice netifd: Interface 'tailscale' is now up
Fri Dec 12 23:53:51 2025 user.notice firewall: Reloading firewall due to ifup of tailscale (tailscale0)
Fri Dec 12 23:53:52 2025 daemon.notice netifd: Network device 'tailscale0' link is down
Fri Dec 12 23:53:52 2025 daemon.notice netifd: Interface 'tailscale' has link connectivity loss
Fri Dec 12 23:53:52 2025 daemon.notice netifd: Interface 'tailscale' is now down
Fri Dec 12 23:53:52 2025 daemon.notice netifd: Interface 'tailscale' is disabled

In other words - looks like the same issue regardless of tailscale version (to be honest I did try a few tailscale versions with the help of your script before with no material difference).

admon · December 13, 2025, 12:30am

If you run tailscale login does it work then?

peter.suba · December 13, 2025, 3:33pm

nope - tailscaled is not running.

What I can do, and I did it now again:

Login and run tailscaled interactively:

/usr/sbin/tailscaled --verbose=3 --state=/etc/tailscale/tailscaled.state --port=41641

On a separate terminal, run tailscale up. This gives me the authentication link:

I can follow the link and authenticate the device. The tailscale console now shows the device as online. I can also ping the device over the tailscale network.

If at this point I exit the SSH connections, and go back to using the web UI, trying to switch on Tailscale, we get basically the same behaviour.

The status row constantly says “connecting”, I get the amber “dot” next to the Tailscale app. The page shows that the machine is bound to my account now, but in the logs I have the exact same loop showing connection-disconnection-connection-disconnection happening.

I have been here before in my troubleshooting attempts, next I was removing the machine from the tailscale console, cleared the state, where we got back to the device authentication attempts with the same connection-disconnection loop. Which is again why I suspect this is not specifically to do with the tailscale version or the core tailscale implementation itself, rather something to do with the GLinet port/implementation/wrapper.

admon · December 13, 2025, 5:40pm

Could you send me your current gl_tailscale file? Maybe my script broke something

peter.suba · December 14, 2025, 3:30am

I can send you the file - let me know how. Thank you for looking.

That said, I’d be surprised if that was the root cause. I have now done an (eyeball) comparison of the script between the device that has the issue and another device that works without the issue.

The only difference I could find is that the device WITH the issue has the following lines in addition to the one that works well:

    for count in `seq 1 10`
    do
        [ $(pgrep -f "tailscaled") ] && break
        sleep 1
    done

This is right after the line:

if [ "$action" = "restart" ];then

This for loop does not exist in the other file.

However, for the life of me I could not see how this may cause the issue - as far as my dev understanding (20+ years old rust) go, this only really waits for tailscaled to be running before continuing, max 10 seconds. Can’t see this triggering the shutdown of the connection.

Then again - maybe I am missing something else as I am eyeballing it.

peter.suba · December 14, 2025, 1:19pm

… mind you, the compared system which is working, is still on 4.8.1 - I did not dare to update to 4.8.3 yet.

ETB · December 18, 2025, 10:46pm

Have an almost identical issue looping issues with Tailscale on router reboot/restart. I have found that stopping the tailscale interface (using LuCi > Network > Interfaces) gives Tailscale a wee break to do whatever the hell its trying to do. (You don’t need to restart it, just click the stop button).

peter.suba · December 19, 2025, 10:47am

So - what is the order of things that you are doing to get it to work? I mean, you’d stop the tailscale interface in LuCi, then go to the main web interface and start tailscale - is that it, or do you need to start tailscale on the main web interface then stop in LuCi etc…?

ETB · December 21, 2025, 8:41am

Tailscale was enabled in the main interface - but with a yellow indicator instead of a green on. I clicked ‘stop’ and then a minute later noticed that Tailscale was up and working. Kinda bizarre tbh.

peter.suba · December 22, 2025, 9:17pm

oh wow - yes I can confirm if I do exactly the same, then I have Tailscale working.

So step 1 - start Tailscale on the main interface - it displays the amber indicator and keeps trying to reconnect.

Step 2- I go to Luci, Stop the tailscale interface

The tailscale interface is then brought back online automatically and tailscale is connected, shows as such on the main web interface.

This seems bizarre - I wonder if someone can progress this issue further and we can get this fixed, because I am a bit worried tinkering further with more advanced functions (like exit node setup) as I am not sure of the behaviour or reliability.

ETB · December 23, 2025, 7:09am

I tracked down how I figured it out - and then promptly forgot until I had some time to check my post history today - yay holidays!

So there is a post here about redirecting your tailscale clients through the router and through Adguard - i.e. so you get adblocking on the go Tailscale - Using AdGuard DNS is not working - #5 by Jopie7863

There is a part where the staff member says:
”Note: Currently you might currently observe that Tailscale is rebooting itself, resulting in a repeated "up" and "down" cycle.
To address this, please click "Stop" when this button is available. This will allow Tailscale to have a break to reconnect and create the tailscale0 device, and then make this interface work again.”

So that’s the best I can do in terms of figuring out what’s going on and why the ‘Stop’ works!

Hope that helps.

peter.suba · December 23, 2025, 10:35pm

Thank you - that thread is an interesting read. I am not quite sure how to progress with troubleshooting/solutioning, so let me just share my thoughts for now - maybe someone else will also chime in with thoughts.

In the other thread there is a key point that the loop is starting after the manual creation of the tailscale0 interface in LuCi.
Another note mentions that the tailscale0 interface is no longer needed to be created after firmware 4.8.2 - so this seems to corraborate my thought that my issues started when I updated to 4.8.3 as clearly the tailscale internal setup have changed in the router which triggered the issue for me.
After all the changes and troubleshooting I have done, now I can no longer be sure if the tailscale0 interface I have was manually created by me (similar to the instructions in the thread you reference) at some point, or whether it exists because the firmware update created it. I could try and delete it and see what happens, but somehow not very confident in this path…
Someone mentioned recently that the tailscale startup loop does come back after restart - I haven’t yet tried restart, I am sure that will be coming, so I could do a few things that would all potentially trigger the loop (e.g. adding more tailscale features as well)… one commenter needed to factory reset the router to fix, so I’ll probably only try it once I am physically near the router which won’t be for a few weeks…

To be honest, I’d ideally like GL-iNet to fix the route cause in firmware as clearly they are aware of something in the setup causing this loop in their tailscale customisation in various circumstances. That would be the standard here…

peter.suba · January 6, 2026, 1:13am

Update.

I can confirm that after the device restarts, the issue comes back, and I have to manually go in and again stop the tailscale0 interface on Luci to stop the flip-flopping behaviour. This is especially a problem for me as this is usually a remote site so I would have to find a different mechanism to remote in. Which… was supposed to be tailscale, actually.

Today, I had a power outage so everything, even my alternative remote failed and restarting it to remote in, I needed someone there physically.

I wonder if, while there is no actual fix, it is possible to script the “stop” action of the tailscale interface and trigger it somehow based on either the observed flip-flop behaviour, or at least after startup at the right time. Maybe someone with a more advanced knowledge of OpenWRT (for scripting the equivalent of the “stop” button action) and/or Linux on general (for the triggering) can suggest something clever?

ETB · January 21, 2026, 12:41am

Can confirm that if you reboot the router, tailscale starts looping up and down again (shows yellow in the main GIU) and the only thing you can do to get it up and running again is to stop the network interface using LuCI which allows it to settle down and connect. Very fustrating.

I have also tried manually updating the tailscale version (as described here: Script: Update Tailscale on (nearly) all devices ) but the same behaviour persists - so its not the version of Tailscale that is at fault here.

Hoping a person smarter than me can help you! A script is a good idea - but long-term Gli.net need to fix.

I’m on v4.8.3 at the moment but the patch notes for the next version don’t mention this issue or tailscale at all ( GL.iNet download center ).

admon · January 21, 2026, 9:16am

Can you give a brief overview of what you changed in comparison to stock firmware?
Never had this issue, and I would say it's not a Tailscale issue at all.

What does the log show for tailscaled?

will.qiu · January 21, 2026, 10:18am

You can try editing it and disabling the “Bring up on boot” option to see whether this fixes the issue of the interface repeatedly coming up again after a reboot.

We more strongly recommend deleting the Tailscale interface that was manually created in LuCI → Network in the previous version.
As you mentioned, the firmware now automatically creates the tailscale0 device and the corresponding firewall zone when the Tailscale feature is enabled, and automatically adds the device to the firewall zone, so manual creation is no longer necessary.

Since this setup is configured through LuCI, and we are unable to account for all possible scenarios that may exist in LuCI, thanks for your understanding.

ETB · January 22, 2026, 3:38am

OK thanks for that @will.qiu

Deleting the tailscale interface and firewall zone worked great.

@peter.suba here is the steps I took to resolve the issue:

Gli.Net GUI:
Applications > set 'Enable Tailscale' to 'off' > Apply changes (ensure it's not showing green or yellow when you open the applications sub menu).

LuCi:
Network > Interfaces: Delete the tailscale interface
Network > Firewall: Delete the tailscale firewall zone

Gli.Net GUI: Applications > enable Tailscale (and other options you need like 'Allow remote access LAN/WAN'

Wait for Tailscale to be green in the applications sub-menu.

Reboot router - look for the green indicator in the Applications sub-menu, green = working and should now work every reboot.