[ARROW-9353] [Python][CI] Nightly dask integration jobs fail

• Type: Bug
• Status: Resolved
• Priority: Blocker
• Resolution: Fixed
• Fix Version/s: 1.0.0
• Component/s: C++
• Labels: pull-request-available
• External issue URL: https://github.com/apache/arrow/issues/25436
• Assignee: Joris Van den Bossche
• Reporter: Neal Richardson
• Created: 07/Jul/20 16:27
• Updated: 11/Jan/23 08:06
• Resolved: 09/Jul/20 10:14

Description

On dask latest:

=================================== FAILURES ===================================
___________________________ test_dataframe_picklable ___________________________

    def test_dataframe_picklable():
        from pickle import loads, dumps
    
        cloudpickle = pytest.importorskip("cloudpickle")
        cp_dumps = cloudpickle.dumps
    
        d = _compat.makeTimeDataFrame()
        df = dd.from_pandas(d, npartitions=3)
        df = df + 2
    
        # dataframe
        df2 = loads(dumps(df))
        assert_eq(df, df2)
>       df2 = loads(cp_dumps(df))
E       ValueError: unsupported pickle protocol: 5

opt/conda/envs/arrow/lib/python3.7/site-packages/dask/dataframe/tests/test_dataframe.py:1623: ValueError

https://github.com/ursa-labs/crossbow/runs/844367735

Dask master has many more failures: https://github.com/ursa-labs/crossbow/runs/844351816

Issue Links

• links to: GitHub Pull Request #7675