[BUG] mortgage_test.py failing on Databricks 14.3

[razajafri]

Running mortgage_test.py::test_mortgage fails with a JavaPackage object is not callable error.

To run the test please run WITH_DEFAULT_UPSTREAM_SHIM=0 TEST_PARALLEL=0 TESTS=mortgage_test.py ./jenkins/databricks/test.sh

This is very similar to the #8910 issue which is specific to EMR

../../src/main/python/mortgage_test.py::test_mortgage[DATAGEN_SEED=1737684235, TZ=UTC, IGNORE_ORDER, INCOMPAT, APPROXIMATE_FLOAT, ALLOW_NON_GPU(ANY), LIMIT(100000)] FAILED [100%]

=================================== FAILURES ===================================
________________________________ test_mortgage _________________________________

mortgage = <conftest.MortgageRunner object at 0x7fae9642fbb0>

    @incompat
    @approximate_float
    @limit
    @ignore_order
    @allow_non_gpu(any=True)
    def test_mortgage(mortgage):
>     assert_gpu_and_cpu_are_equal_iterator(
              lambda spark : mortgage.do_test_query(spark))

../../src/main/python/mortgage_test.py:27: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
../../src/main/python/asserts.py:607: in assert_gpu_and_cpu_are_equal_iterator
    _assert_gpu_and_cpu_are_equal(func, 'ITERATOR', conf=conf, is_cpu_first=is_cpu_first)
../../src/main/python/asserts.py:506: in _assert_gpu_and_cpu_are_equal
    from_cpu = run_on_cpu()
../../src/main/python/asserts.py:491: in run_on_cpu
    from_cpu = with_cpu_session(bring_back, conf=conf)
../../src/main/python/spark_session.py:150: in with_cpu_session
    return with_spark_session(func, conf=copy)
/usr/lib/python3.10/contextlib.py:79: in inner
    return func(*args, **kwds)
../../src/main/python/spark_session.py:134: in with_spark_session
    ret = func(_spark)
../../src/main/python/asserts.py:221: in <lambda>
    bring_back = lambda spark: limit_func(spark).toLocalIterator()
../../src/main/python/asserts.py:202: in with_limit
    df = sorted_func(spark)
../../src/main/python/asserts.py:192: in with_sorted
    df = func(spark)
../../src/main/python/mortgage_test.py:28: in <lambda>
    lambda spark : mortgage.do_test_query(spark))
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

self = <conftest.MortgageRunner object at 0x7fae9642fbb0>
spark = <pyspark.sql.session.SparkSession object at 0x7fae96e46a10>

    def do_test_query(self, spark):
      from pyspark.sql.dataframe import DataFrame
      jvm_session = _get_jvm_session(spark)
      jvm = _get_jvm(spark)
      acq = self.mortgage_acq_path
      perf = self.mortgage_perf_path
      run = jvm.com.nvidia.spark.rapids.tests.mortgage.Run
      if self.mortgage_format == 'csv':
          df = run.csv(jvm_session, perf, acq)
      elif self.mortgage_format == 'parquet':
>         df = run.parquet(jvm_session, perf, acq)
E         TypeError: 'JavaPackage' object is not callable

../../src/main/python/conftest.py:496: TypeError

[razajafri]

Might be similar to #11988

Even though the stacktrace looks different we suspect the underlying problem is related to a jar not being found on the classpath.