What is a thread scheduler?

A thread is a piece of code that is executed independently of other threads. While doing so, it uses some CPU resource. If there are many threads running at the same time, you need to be able to suspend a thread so that the others get a chance to run.

This suspension should happen if the thread is blocked — for instance, waiting for data coming from the disk or the network, or waiting for a monitor that is preventing it from executing a synchronized piece of code.

That is the role of the thread scheduler: making sure that the CPU resource is evenly shared among threads, and that the active threads are indeed using the CPU resource and not waiting on it, doing nothing.

This capacity of suspending a thread is called preemptiveness. A thread scheduler can be preemptive or non-preemptive.

Virtual threads use a specific scheduler that is not preemptive. That is the reason why you should never run long-running in-memory computations on a virtual thread: if you do, you could block the entire carrier thread — and with it, the other virtual threads scheduled on it.