[Feature] More Information of Unsupported Tailscale Client

[IamTaoChen]

Use case

I have an older tailscale client, when I upgrade to v0.25.0-beta.1, there is an error.

headscale  | 2025/02/05 10:44:40 http: response.Write on hijacked connection from fmt.Fprintln (print.go:305)
headscale  | 2025-02-05T10:44:40-05:00 ERR http internal server error error="noise upgrade failed: unsupported client version: 85" code=500

1. If the error message can print the client information(name,IP .etc..), I can locate the client quickly.

Description

see up.

Contribution

• I can write the design doc for this feature
• I can contribute this feature

How can it be implemented?

No response

[kradalby]

Hi, this has already been added in the latest beta #2391.

[kradalby]

I've opened another PR adding the version number so there is some more information, but unfortunately, at that part in the code we dont want to do any database lookup and we do not yet know any identifiers, so the version is what is available.

[mfld-pub]

Was just about to open an issue for the same. I had a hunch which clients are doing this. It's openwrt / gl.inet devices. Their repos are stuck on a low rev of tailscale 1.6x. I was able to manually update them to stop the headscale log from being flooded. In this case theres only 90 machines on the tailnet but if you have hundreds it really would help to be able to know who is misbehaving. It makes sense that at this part of the session you don't want to to intensive lookups. Difficult to solve.

Good thing not many devices are 15 tailscale versions behind. @IamTaoChen if you are dealing with gl.inet devices you can see this solution for inspiration. Other ARM devices you can manually take the tailscale binaries, shrink them with upx and replace.

Edit: 💡 is it feasible to have an additional flag that has headscale nodes list return last seen client version ? This would prevent intensive lookups at each event but let the admin identify the devices as needed.