all Python 3multiprocessingterm

multiprocessing.Condition

synchronization

A condition variable shared between processes: one process waits in wait() or wait_for(), another wakes it with notify() after changing shared state.

Syntax

cond = ctx.Condition(lock=None); with cond: cond.wait_for(pred, timeout); cond.notify_all()

wait() and notify() may only be called with the lock held, otherwise AssertionError, not RuntimeError as in threading. The condition itself must live in shared memory (Value, Array, Manager): a plain variable is per process, leaving the waiter nothing to check. wait_for(predicate, timeout) beats a bare wait(): it rechecks the condition after spurious wakeups and returns its value. Pass the primitive to the child as a Process argument at creation; it cannot go through a Queue or Pool arguments — RuntimeError "should only be shared between processes through inheritance".

Examples

import multiprocessing as mp
def producer(cond, ready):
    with cond:
        ready.value = 1
        cond.notify_all()
if __name__ == '__main__':
    ctx = mp.get_context('spawn')
    cond, ready = ctx.Condition(), ctx.Value('i', 0, lock=False)
    p = ctx.Process(target=producer, args=(cond, ready))
    p.start()
    with cond:
        print(cond.wait_for(lambda: ready.value == 1, timeout=10))  # → True
    p.join()

import multiprocessing as mp
if __name__ == '__main__':
    cond = mp.get_context('spawn').Condition()
    try:
        cond.wait(0.1)
    except AssertionError as e:
        print(e)  # → must acquire() condition before using wait()

See also

Open in the full glossarydocs.python.orgReport an inaccuracy