예제 #1
0
def test_ingestion_manager_run_success():
    df = pd.DataFrame({"float": pd.Series([2.0], dtype="float64")})
    manager = IngestionManagerPandas(
        feature_group_name="MyGroup",
        sagemaker_session=sagemaker_session_mock,
        data_frame=df,
        max_workers=10,
    )
    manager.run()
예제 #2
0
def test_ingestion_manager_run_multi_process_with_multi_thread_success(
    fs_runtime_client_config_mock, ):
    df = pd.DataFrame({"float": pd.Series([2.0], dtype="float64")})
    manager = IngestionManagerPandas(
        feature_group_name="MyGroup",
        sagemaker_fs_runtime_client_config=fs_runtime_client_config_mock,
        max_workers=2,
        max_processes=2,
    )
    manager.run(df)
def test_ingestion_manager_run_success():
    df = pd.DataFrame({"float": pd.Series([2.0], dtype="float64")})
    manager = IngestionManagerPandas(
        feature_group_name="MyGroup",
        sagemaker_fs_runtime_client_config=fs_runtime_client_config_mock,
        max_workers=10,
    )
    manager.run(df)

    manager._run_multi_process.assert_called_once_with(data_frame=df, wait=True, timeout=None)
def test_ingestion_manager_run_failure():
    df = pd.DataFrame({"float": pd.Series([2.0], dtype="float64")})
    manager = IngestionManagerPandas(
        feature_group_name="MyGroup",
        sagemaker_session=sagemaker_session_mock,
        data_frame=df,
        max_workers=10,
    )
    with pytest.raises(RuntimeError) as error:
        manager.run()
    assert "Failed to ingest some data into FeatureGroup MyGroup" in str(error)
예제 #5
0
def test_ingestion_manager_with_profile_name_run_failure():
    df = pd.DataFrame({"float": pd.Series([2.0], dtype="float64")})
    manager = IngestionManagerPandas(
        feature_group_name="MyGroup",
        sagemaker_fs_runtime_client_config=fs_runtime_client_config_mock,
        max_workers=1,
        profile_name="non_exist",
    )

    try:
        manager.run(df)
    except Exception as e:
        assert "The config profile (non_exist) could not be found" in str(e)
예제 #6
0
def test_ingestion_manager_run_failure():
    df = pd.DataFrame({"float": pd.Series([2.0], dtype="float64")})
    manager = IngestionManagerPandas(
        feature_group_name="MyGroup",
        sagemaker_fs_runtime_client_config=fs_runtime_client_config_mock,
        max_workers=1,
    )

    with pytest.raises(IngestionError) as error:
        manager.run(df)

    assert "Failed to ingest some data into FeatureGroup MyGroup" in str(error)
    assert error.value.failed_rows == [1]
    assert manager.failed_rows == [1]