[ARROW-13198] [C++][Dataset] Async scanner occasionally segfaulting in CI

• Type: Bug
• Status: Closed
• Priority: Blocker
• Resolution: Cannot Reproduce
• Fix Version/s: 6.0.0
• Component/s: C++
• Labels: dataset datasets
• External issue URL: https://github.com/apache/arrow/issues/28888
• Assignee: Weston Pace
• Reporter: David Li
• Created: 28/Jun/21 16:03
• Updated: 11/Jan/23 08:31
• Resolved: 27/Sep/21 18:21

Description

See attached log; it's failing in test_open_dataset_partitioned_directory[threaded-async] AMD64 Conda Python 3.7 Pandas latest.log

2021-06-28T15:53:47.0090857Z opt/conda/envs/arrow/lib/python3.7/site-packages/pyarrow/tests/test_dataset.py::test_scan_iterator[True-True] PASSED [ 27%]
2021-06-28T15:53:48.5943186Z /arrow/ci/scripts/python_test.sh: line 32:  7137 Segmentation fault      (core dumped) pytest -r s -v ${PYTEST_ARGS} --pyargs pyarrow
2021-06-28T15:53:49.1303267Z 139
/pyarrow/tests/test_dataset.py::test_open_dataset_partitioned_directory[threaded-async] 

Issue Links

• duplicates: ARROW-13248 [Python] Segfault in test_dataset::test_scan_iterator on python 3.9
• is related to: ARROW-13417 [C++] The merged generator can sometimes pull from source sync-reentrant