tadeja opened a new issue, #50037: URL: https://github.com/apache/arrow/issues/50037
### Describe the bug, including details regarding any error messages, version, and platform. `test_table_uses_memory_pool` fails occasionally with `.../test_pandas.py:3964: AssertionError` Observed on job `ARM64 macOS 14`, where `pytest -r s -n auto ...` uses xdist to run the tests in parallel (8207, so ~2700 per worker). [Latest example May 25](https://github.com/apache/arrow/actions/runs/26399836679/job/77709190993?pr=50033#step:10:210), same commit [re-run passed](https://github.com/apache/arrow/actions/runs/26399836679/job/77727866997?pr=50033#step:10:208). [Second example May 20](https://github.com/apache/arrow/actions/runs/26191758355/job/77061638937#step:10:230) [Third example May 20 is on main](https://github.com/apache/arrow/actions/runs/26135872194/job/76870837319#step:10:230) ```bash def test_table_uses_memory_pool(): N = 10000 arr = pa.array(np.arange(N, dtype=np.int64)) t = pa.table([arr, arr, arr], ['f0', 'f1', 'f2']) prior_allocation = pa.total_allocated_bytes() ... x = None # noqa gc.collect() > assert pa.total_allocated_bytes() == prior_allocation E assert 1472 == 1664 ... ``` Resembles #44728: `prior_allocation` captures before `gc.collect()`, so lingering allocations from earlier tests in the same xdist worker could be counted in the baseline but later freed by `gc.collect()` making final smaller than baseline. ### Component(s) Python, Continuous Integration -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
