I have a bug with futures.as_completed() or futures.wait() that will be blocked indefinitely when all Futures are completed or cancelled.
Here the steps to reproduce:
After having submited Futures with ThreadPoolExecutor.submit()
, I'm waiting for my Futures with futures.as_completed()
or futures.wait()
. In an other thread, I call ThreadPoolExecutor.shutdown()
with cancel_futures=True
and then in this same process, I'm waiting for Futures to complete with a timeout. The wait will return after the delay passed, with 2 lists: completed Futures and cancelled Futures. There is no more pending Futures. However, the first as_completed()
(or wait()
) in the main thread is still blocking.
In the Python documenation , it is stated for return_when=ALL_COMPLETED
:
The function will return when all futures finish or are cancelled .
And for as_completed()
Returns [...] futures as they complete (finished or cancelled futures).
Which corresponds to my situation. Is it a bug or am I missing something? I tried to call shutdown()
in the same thread, it doesn't change anything.
Code sample:
import signal
import time
from concurrent import futures
from concurrent.futures import Future, ALL_COMPLETED
from concurrent.futures import ThreadPoolExecutor
from typing import Dict, Set
class SubThreads:
def __init__(self):
self.running_futures_url: Dict[str, Future] = {}
self.webpage_crawler_th_pool = ThreadPoolExecutor(2)
def shutdown(self):
print("Waiting for lasts URL threads")
self.webpage_crawler_th_pool.shutdown(wait=False, cancel_futures=True)
finished_futures, still_running_futures = futures.wait(
self.running_futures_url.values(), return_when=ALL_COMPLETED, timeout=5,
)
print("Shutdown done, remaining threads", len(still_running_futures))
def crawl_url(self, url):
print("Crawling webpage", url)
time.sleep(3)
print("Webpage crawled", url)
return "URL Crawled"
def run(self):
urls = ['1', '2', '3', '4', '5']
for url in urls:
running_th = self.webpage_crawler_th_pool.submit(self.crawl_url, url)
self.running_futures_url[url] = running_th
print("Waiting for URLs to be crawled")
# for _future in futures.as_completed(self.running_futures_url.values()):
# print("Future result:", _future.result()) # Will only return and print first 2 started (and completed) Futures
finished_futures, still_running_futures = futures.wait(
self.running_futures_url.values(), return_when=ALL_COMPLETED
)
print("SubThread finished (never called)", finished_futures, still_running_futures)
sub_thread = SubThreads()
def signal_handler(sig, frame):
print("Signal caught, exiting ...", sig)
sub_thread.shutdown()
signal.signal(signal.SIGINT, signal_handler)
signal.signal(signal.SIGTERM, signal_handler)
sub_thread.run()
I would not use shutdown
with wait=False
.
See the docs:
shutdown(wait=True)
Stops accepting new tasks. It waits for all the running tasks to complete if wait is True.
So since you pass wait=False
, your as_completed
will wait forever, because the ThreadPoolExecutor
will never stop running until you call shutdown
with wait=True
.
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.