David Crawshaw

Turns out there's quite a bit to do here. 9693dc7 is a beginning, it's enough to import `gonum.org/v1/gonum/mat`. But there's parsing to be done for `i`, support for `real` and...

I wrote a comment, and GitHub ate it. :-( We can fix this, but slow down builds, by removing `-i`. We can also fix this with `-pkgdir`, but we have...

Note that `r` appears to be supported now, which in vim means "replace the current character with whatever character is typed next. What is not yet supported is `R` which...

Potential fast fix for this customer: port hint from control protocol. We send them an unstable build and see if a random port hint fixes it. Thoughts?

Ah. So I guess the fast experiment is to ship them an unstable that by default uses a random port. For the real fix, do we want to make this...

I think the fast experiment is to ship this one customer a binary that does it properly and see what happens. For the right fix, how do you detect 41641...

Presumably in the linked case, if the upstream NAT simply never let packets through 41641 for a client, then it would fall back to DERP and everything would be fine,...

On the receive side, it seems relatively straightforward to *always* open two magicsock ports and report them both to control, one 41641, the other random. The problem is, if the...

Could you initiate the service stop, then after a few seconds SIGABRT the binary to get a stack trace of it during the slow shutdown?

Two most likely things that spring to mind then are the system has brought down a mount that tailscaled depends on (a log file?) or networking, which tailscaled is waiting...