pytorch - 💡(How to fix) Fix DISABLED test_memory_plots_free_segment_stack (__main__.TestCudaAllocator)

Traceback (most recent call last): File "/var/lib/jenkins/workspace/test/test_cuda.py", line 4589, in test_memory_plots_free_segment_stack self.assertEqual(("empty_cache" in ss), (context == "all")) File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/testing/_internal/common_utils.py", line 4441, in assertEqual raise error_metas.pop()[0].to_error( # type: ignore[index] AssertionError: Booleans mismatch: True is not False

To execute this test, run the following from the base repo dir: python test/test_cuda.py TestCudaAllocator.test_memory_plots_free_segment_stack

This message can be suppressed by setting PYTORCH_PRINT_REPRO_ON_FAILURE=0

TL;DR

The most likely fix involves investigating and resolving the boolean mismatch in the test_memory_plots_free_segment_stack test case.

Guidance

• Investigate the test_memory_plots_free_segment_stack test case in test_cuda_expandable_segments.py to understand why the boolean values are mismatched.
• Run the test locally using the provided command python test/test_cuda.py TestCudaAllocator.test_memory_plots_free_segment_stack to reproduce the error.
• Study the logs from the CI workflow by following the provided debugging instructions to identify patterns or clues that may lead to the root cause of the flakiness.
• Check the conditions under which "empty_cache" in ss and context == "all" are evaluated to ensure they are correctly implemented.

Example

No code snippet is provided as the issue does not imply a specific code change.

Notes

The fix may involve modifying the test case or the underlying code that the test case is verifying, depending on the root cause of the boolean mismatch.

Recommendation

Apply workaround: The issue seems to be related to a flaky test, and without a clear root cause, applying a workaround such as re-running the test or modifying the test conditions may be necessary.