Enable Tailscale client metrics

[xeberth]

Now you can expose and collect Tailscale client metrics for use with monitoring systems such as Prometheus for your Tailscale network.

These metrics provide insights into client behavior, health, and performance. For example, you can view information about client connectivity to subnet routers configured in your tailnet.

This enable also the option to monitor tailscale from grafana dashboards

How we can enable this option from add-on config?

---

Sources: https://tailscale.com/kb/1482/client-metrics

[frenck]

Oh man they keep adding options huh?

I'm at the point that I'm thinking about just not adding more options. It really kills the ease of use and becoming a monster of things.

[frenck]

I would recommend to create a feature request upstream with tailscale, to ask if they can extend an WebUI that provides options like these.

[lmagyar]

@xeberth It's already part of the web UI, but there is a bug in Tailscale web server, when the --readonly flag is used, ie. tailscale web --readonly is used (as the add-on starts the web ui), tailscale forgets to start up the 5252 port listener for the metrics data on the tailnet IP of the client.

$ /opt/tailscale web
2025/02/21 23:54:16 starting tailscaled web client at http://100.x.x.x:5252
2025/02/21 23:54:16 web server running on: http://localhost:8088

$ /opt/tailscale web --readonly
2025/02/21 23:54:23 web server running on: http://localhost:8088

curl 100.x.x.x:5252/metrics should work with --readonly also

Please open an issue at the tailscale repo with the above information.

[lmagyar]

Created the TS issue: tailscale/tailscale#15108

[lmagyar]

The --readonly bug is confirmed upstream, this issue can be closed now.