python / python/cpython

Python `io.open` intermittently throw `PermissionError` on windows

Open
#136,965 13 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

OS-windows pending topic-IO type-bug
Dominant language
Python
Stars
77.2k
Forks
35.9k
PR merge metrics
PR metrics pending

Description

Bug report

Bug description:

The io.open method intermittently raises a PermissionError on the Windows operating system. I've attached a Python script that reproduces this issue on version 3.12.8, and 3.11.9 (didn't test for any other version)

My investigation suggests that the problem arises from consecutive write operations performed within very short timeframes where each operation does fresh io.open to write updates -- For example, when attempting to write 10 updates back-to-back to a file with fresh io.open each time.

On Windows, explorer.exe runs in the background to update indexing and metadata upon file modification. If explorer.exe has already acquired a handle to the file, any new request to write to it will result in a PermissionError.

While we could implement a retry mechanism with backoff for io.open in our code, I'm curious why Python doesn't natively handle such background process contention.

import time
import threading
import random
import multiprocessing
import os

# Sample message text
MESSAGE = (
    r"file://C:\Users\sample\Desktop\temp\tmpgzy8f_lt\929cec70297d4938b31545f858d6b38c.txt,"
    r"gs://sample-bucket/loadwintest/hdahsdhahdahsdhahd/temp/dahsdhahdhadhahdh/dhahsdahdhahdhahdahdhahdha.txt,"
    r"2025-07-21T10:44:43.263442Z,2025-07-21T11:00:55.588213Z,dahshdahdhahdhahshah+hfQ==,25000000000,25000000000,OK,"
)

# Output file path
output_file = "winman/test_output11.txt"


# Writer process – gets messages from queue and writes to file
def writer(queue, end_time):
    while time.time() < end_time:
        msg = queue.get()
        with open(output_file, "at", encoding='utf-8', newline='\n') as f:
            f.write(msg + '\n')


# Thread that puts messages into the queue
def thread_worker(queue, thread_id, process_id, end_time):
    while time.time() < end_time:
        notify_msg = f"[P{process_id}-T{thread_id}] {MESSAGE}"
        queue.put(notify_msg)
        time.sleep(random.choice([0.3, 0.5, 0.4]))


# Process that spawns multiple threads
def process_worker(queue, num_threads, process_id, end_time):
    threads = []
    for tid in range(num_threads):
        t = threading.Thread(target=thread_worker, args=(queue, tid, process_id, end_time))
        threads.append(t)
        t.start()

    for t in threads:
        t.join()


def main():
    runtime_seconds = 600
    end_time = time.time() + runtime_seconds

    num_processes = 4
    threads_per_process = 4

    queue = multiprocessing.Queue()

    # Start writer process
    writer_proc = multiprocessing.Process(target=writer, args=(queue, end_time))
    writer_proc.start()

    # Start worker processes
    processes = []
    for pid in range(num_processes):
        p = multiprocessing.Process(target=process_worker, args=(queue, threads_per_process, pid, end_time))
        p.start()
        processes.append(p)

    # Wait for all worker processes
    for p in processes:
        p.join()

    # Terminate writer after work is done
    writer_proc.terminate()
    writer_proc.join()


if __name__ == "__main__":
    main()
CPython versions tested on:

3.12

Operating systems tested on:

Windows

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by running the attached Python reproducer on Windows with the reported CPython versions and confirm whether the intermittent PermissionError can be reproduced. Then inspect the io.open entry point and relevant Windows file-opening behavior; done should establish a reproducible cause and a clearly scoped fix or documented limitation.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
operating-systems
Issue type
Bug
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
20/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.