Exception ignored in: <function Connection.__del__ at ...> connection shutdown.

[fdcastel]

Running this code:

# requires `pandas` and `sqlalchemy-firebird` packages.
import sqlalchemy as sa
import pandas as pd

def main():
    db_uri = "firebird://..."  # any database
    q = 'SELECT COUNT(*) FROM mon$attachments'

    e = sa.create_engine(db_uri)
    with e.connect() as conn:
        df = pd.read_sql(sql=sa.text(q), con=conn)
        print(df)

    print('----- [ END OF MAIN ] -----')

if __name__ == "__main__":
    main()

outputs:

   COUNT
0      3
----- [ END OF MAIN ] -----
Exception ignored in: <function Connection.__del__ at 0x000001BA5CAFD6C0>
Traceback (most recent call last):
  File "C:\Temp\sqla\python3-driver\src\firebird\driver\core.py", line 1802, in __del__
    self._att.detach()
  File "C:\Temp\sqla\python3-driver\src\firebird\driver\interfaces.py", line 1214, in detach
    self._check()
  File "C:\Temp\sqla\python3-driver\src\firebird\driver\interfaces.py", line 141, in _check
    raise self.__report(DatabaseError, self.status.get_errors())
firebird.driver.types.DatabaseError: connection shutdown

using latest master of python3-driver.

[fdcastel]

@pcisar I've tried every possible solution in SQLAlchemy, but it seems the only place to fix this is in the Python 3 driver.

Kindly correct me if I’m mistaken.

Problem

The issue is in the Connection.__del__ method, more precisely in this detach() call.

When the destructor is called during garbage collection, it attempts to detach from a connection that had already been closed by the server. This results in a "connection shutdown" error.

Solution

I replaced that detach() call with the following:

try:
    self._att.detach()
except DatabaseError:
    # Connection may already be closed by server during cleanup
    pass
finally:
    self._att = None

and everything is working correctly now:

   COUNT
0      3
----- [ END OF MAIN ] -----

I can submit a PR if you approve.