I like tailscale and have been testing it for a few months. I’m also using headscale as the control plane.
Unfortunately the android client is somewhat unreliable. It works most of the time but once in a while, connections to your tailnet will fail for a bit and require retries. If you ping a machine in your tailnet during this problem, it will show packet loss and then start working after a few pings. This unfortunately makes it difficult to have a reliable split DNS setup.
I’ve done everything to try and understand what happens without success. It seems like state is lost somewhere and a few packets flowing will fix it. Running a constant ping from Android to my tailnet “fixes” the problem, but is not a great workaround.
Just something to keep in mind before you jump headfirst.