Understanding ThreadPoolExecutor Worker Thread Recycling in JDK 1.8

1. runWorker(Worker w)

When a worker thread starts, it enters the runWorker(Worker w) method, which loops while tasks are available; if no task is found or an exception occurs, the loop exits and processWorkerExit(w, completedAbruptly) removes the thread.

The first task comes from firstTask , after which tasks are fetched via getTask() . If getTask() returns null , the thread terminates.

2. getTask() returns null

Two situations cause getTask() to return null :

The pool is in STOP, TIDYING, TERMINATED state, or SHUTDOWN with an empty work queue.

The number of workers exceeds the core size and either the worker count is above the maximum or the thread has timed out while the queue is empty.

3. Scenario‑based analysis of thread recycling

3.1 No shutdown() – RUNNING state, all tasks completed

If the pool remains RUNNING and tasks finish, excess workers (above corePoolSize ) time out, the timed flag is true, and a CAS operation reduces the worker count; successful CAS returns null and the thread is reclaimed.

CAS guarantees that the core size is never undershot, because failed CAS attempts simply retry the loop.

3.2 shutdown() invoked – all tasks completed

Calling shutdown() interrupts idle workers. If a worker is blocked in getTask() , the interrupt wakes it, the thread checks the termination conditions, and processWorkerExit removes it.

3.2.1 All tasks finished, thread blocked

The interrupt awakens the thread, getTask() returns null , and the worker count is decremented.

3.2.2 Tasks still running

Running workers continue until they finish their current tasks; after each task they call getTask() again, which may return null once the queue is empty, leading to termination via processWorkerExit . The pool’s tryTerminate() method may interrupt any remaining idle thread to ensure all blocked workers are eventually awakened and reclaimed.

4. Summary

ThreadPoolExecutor recycles a worker when getTask() returns null . In a RUNNING pool without shutdown, excess workers time out and are removed via CAS; after shutdown() , all workers are interrupted, and each eventually exits after completing or failing to obtain further tasks.