python / python/cpython

SQLite3 multithreading cache inconsistency

未關閉
#118,172 10 則留言 9 個 reaction 已指派 0 人 在 GitHub 檢視

還沒有人認領這個 Issue。

extension-modules topic-sqlite3 type-bug
主要語言
Python
星號
77.2k
分支
35.9k
PR 合併指標
PR 指標待擷取

描述

Bug report

Bug description:

When using SQLite3 in multi-threading application, the fetch reuslts are not consistent. After some testing, this seems to be caused by the caching.
The SQLite threading mode is serialized.
This issue only exists in python 3.12 and 3.13. No issue in 3.11.
Currently solution is to set cached_statements=0 when connect to the database.

A simple demo to reproduce the error:

import sqlite3
import threading

KB = sqlite3.connect(
    "example.db", check_same_thread=False
)

def execute_query():
    cursor = KB.cursor()
    cursor.execute("SELECT * FROM test_table")
    result = cursor.fetchall()
    assert result == [(1, 'test1'), (2, 'test2'), (3, 'test3')], str(result)
    return result

def run_threads():
    threads = []
    for i in range(50):
        thread = threading.Thread(target=execute_query)
        threads.append(thread)
        thread.start()

    for thread in threads:
        thread.join()

def test_multithreading():
    cursor = KB.cursor()

    cursor.execute(
        """CREATE TABLE IF NOT EXISTS test_table (id INTEGER PRIMARY KEY, value TEXT)"""
    )
    KB.commit()

    cursor.execute("""DELETE FROM test_table""")
    KB.commit()

    cursor.execute("""INSERT INTO test_table (value) VALUES ('test1')""")
    cursor.execute("""INSERT INTO test_table (value) VALUES ('test2')""")
    cursor.execute("""INSERT INTO test_table (value) VALUES ('test3')""")
    KB.commit()

    run_threads()

    KB.close()

if __name__ == "__main__":
    test_multithreading()

Test output: (Different in each run because of inconsistent fetch results)

Exception in thread Thread-3 (execute_query):
Traceback (most recent call last):
  File "/usr/lib/python3.12/threading.py", line 1073, in _bootstrap_inner
    self.run()
  File "/usr/lib/python3.12/threading.py", line 1010, in run
Exception in thread Thread-7 (execute_query):
Traceback (most recent call last):
  File "/usr/lib/python3.12/threading.py", line 1073, in _bootstrap_inner
    self._target(*self._args, **self._kwargs)
  File "/home/user/tmp/kb_test.py", line 13, in execute_query
    self.run()
  File "/usr/lib/python3.12/threading.py", line 1010, in run
Exception in thread Thread-12 (execute_query):
Traceback (most recent call last):
  File "/usr/lib/python3.12/threading.py", line 1073, in _bootstrap_inner
    assert result == [(1, 'test1'), (2, 'test2'), (3, 'test3')], str(result)
    self._target(*self._args, **self._kwargs)
AssertionError: [(1, 'test2'), (3, 'test3')]
  File "/home/user/tmp/kb_test.py", line 13, in execute_query
Exception in thread Thread-1 (execute_query):
Traceback (most recent call last):
  File "/usr/lib/python3.12/threading.py", line 1073, in _bootstrap_inner
    self.run()
    assert result == [(1, 'test1'), (2, 'test2'), (3, 'test3')], str(result)
    self.run()
AssertionError: [(3, 'test3')]
  File "/usr/lib/python3.12/threading.py", line 1010, in run
  File "/usr/lib/python3.12/threading.py", line 1010, in run
    self._target(*self._args, **self._kwargs)
  File "/home/user/tmp/kb_test.py", line 13, in execute_query
    self._target(*self._args, **self._kwargs)
  File "/home/user/tmp/kb_test.py", line 13, in execute_query
    assert result == [(1, 'test1'), (2, 'test2'), (3, 'test3')], str(result)
    assert result == [(1, 'test1'), (2, 'test2'), (3, 'test3')], str(result)
AssertionError: [(None, None), (), (2, 'test2'), (), (), (2, 'test2'), (0, None), (1, 'test1'), (2, 'test2'), (3, 'test3')]
AssertionError: [(3, 'test3')]
CPython versions tested on:

3.12, 3.13

Operating systems tested on:

Linux

貢獻指南

開啟貢獻指南

從這裡開始

  1. 先讀完整個 Issue,再讀專案的貢獻指南。
  2. 在 Issue 下留言說明你要接手 —— 這能避免兩個人做同樣的事。
  3. Fork 儲存庫,在一個分支上完成修改。
  4. 送出 Pull Request,並在描述裡引用這個 Issue 編號。

研究方向

從提供的 Python sqlite3.connect() 重現程式開始,在執行 50 執行緒查詢並變更 cached_statements 的同時,比較 Python 3.11、3.12 和 3.13。調查 serialized SQLite 執行緒模式下 sqlite3 陳述式快取的行為。當多執行緒擷取能夠穩定返回預期的三列,且不需要 cached_statements=0 時,即表示完成。

由索引模型根據 Issue 內容生成。

評估

技術堆疊
python, sqlite
領域
databases
Issue 類型
缺陷
難度
4/5
預估耗時
3-5 天
活躍度
停滯
描述清晰度
需要釐清
新手友好度
35/100

把新 issue 寄到你的電子郵件信箱

精選適合新手參與的 GitHub issue 摘要。