Possible memory leak when using rowfactory

[robingie]

1. OracleDB version
   Oracle 11.2.1

2. Commands

• print("sys.maxsize > 232:", sys.maxsize > 232)
  True
• print("platform.platform:", platform.platform())
  platform.platform: Linux-4.18.0-348.7.1.el8_5.x86_64-x86_64-with-glibc2.28
• print("platform.python_version:", platform.python_version())
  platform.python_version: 3.9.6
• print("cx_Oracle.version:", cx_Oracle.version)
  cx_Oracle.version: 8.3.0
• print("cx_Oracle.clientversion:", cx_Oracle.clientversion())
  cx_Oracle.clientversion: (19, 6, 0, 0, 0)

3. Is it an error or a hang or a crash?
   Hang because the system can run out of memory.

4. What error(s) or behavior you are seeing?
   The Python application grows in memory size after a while. Each time a new cursor is created and the cursor.rowfactory is used, the memory requirement increases.

5. Include a runnable Python script that shows the problem.

with conn.cursor() as cur:
        cur.execute(query2, start_id=start_id)

        # This line causes a growing memory footprint
        cur.rowfactory = lambda *args: dict(zip([d[0] for d in cur.description], args))

        # This line avoids the memory leak
        data = [dict((cur.description[i][0], value) for i, value in enumerate(row)) for row in cur.fetchall()]

[lime-n]

I too have been having the same issue when running this. I have experienced black-screen as well as blue-screen because of the growing memory size.

What's the progress on this?

[sharadraju]

@lime-n Please upgrade cx_Oracle to the new version, now called python-oracledb, see the release announcement and run your test case.
If this issue still persists, please provide us a runnable test case with the complete platform details as shown in the initial comment of this bug.

[lime-n]

Hi @sharadraju,

I have been using python-oracledb and this issue will regardlessly persist. When I have some time, I'll upload a reproducible example.

[stale]

This issue has been automatically marked as inactive because it has not been updated recently. It will be closed if no further activity occurs. Thank you for your contributions.

[stale]

This issue has been automatically closed because it has not been updated for a month.