You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Describe the bug
I was seeing test failures in one of our integration test environment where we were running on an actual cluster. Specifically with test test_hash_groupby_typed_imperative_agg_without_gpu_implementation_fallback. This test assumes that the spark.rapids.sql.incompatibleOps.enabled=false. In this test environment that flag had been turned to true in the global spark configs which caused the test to fail. In this case it wasn't obvious to me that this test relied on that behavior.
If the tests are assuming this is false we should have a global check for it and fail with useful message if it finds its not set that way.
Describe the bug
I was seeing test failures in one of our integration test environment where we were running on an actual cluster. Specifically with test test_hash_groupby_typed_imperative_agg_without_gpu_implementation_fallback. This test assumes that the spark.rapids.sql.incompatibleOps.enabled=false. In this test environment that flag had been turned to true in the global spark configs which caused the test to fail. In this case it wasn't obvious to me that this test relied on that behavior.
If the tests are assuming this is false we should have a global check for it and fail with useful message if it finds its not set that way.
Note that we have a check https://github.com/NVIDIA/spark-rapids/blob/branch-22.04/integration_tests/src/main/python/asserts.py#L219 that doesn't like setting the conf manually.
The text was updated successfully, but these errors were encountered: