Goroutine Scheduler (GMP)

sysmon is a monitoring thread the Go runtime starts at program start and runs for the process lifetime. It holds no P, so GOMAXPROCS does not bound it and it can still act when every P is busy: polling the network, retaking Ps and forcing a periodic collection.

That the failure depends on the runtime interrupting goroutines with a signal at arbitrary instructions, since asyncpreemptoff=1 leaves only function-call preemption. Treat the setting as a bisect that narrows the bug, not as a fix to ship.

On every 61st scheduling tick a P takes one goroutine from the global run queue before looking at its own, and a P that empties its local queue drains a global batch. Without the periodic check, local work would win forever.

It runs the test in a bubble with its own fake clock. That clock moves only when every goroutine in the bubble is durably blocked, then jumps to the next timer deadline, so no real waiting happens.

Default to the runtime-derived value fleet-wide and treat pins as documented exceptions. The default is only trustworthy if every service builds on a toolchain that reads the container CPU limit and every workload actually has such a limit; otherwise pin.