XFAIL test_dataframe_cow_chain

tests/benchmarks/test_dataframe.py

@@ -1,7 +1,8 @@
+import pytest
 from dask.sizeof import sizeof
 from dask.utils import format_bytes
 
-from ..utils_test import cluster_memory, timeseries_of_size, wait
+from ..utils_test import HAS_PYARROW12, cluster_memory, timeseries_of_size, wait
 
 
 def print_dataframe_info(df):
@@ -69,6 +70,9 @@ def test_filter(small_client):
     wait(result, small_client, 10 * 60)
 
 
+@pytest.mark.xfail(
+    HAS_PYARROW12, reason="https://github.com/coiled/benchmarks/issues/1002"
+)
 def test_dataframe_cow_chain(small_client):
     memory = cluster_memory(small_client)  # 76.66 GiB
 

tests/benchmarks/test_parquet.py

@@ -12,17 +12,9 @@
 import pandas
 import pytest
 from coiled import Cluster
-from packaging.version import Version
 
 from ..conftest import dump_cluster_kwargs
-from ..utils_test import run_up_to_nthreads, wait
-
-try:
-    import pyarrow
-
-    HAS_PYARROW12 = Version(pyarrow.__version__) >= Version("12.0.0")
-except ImportError:
-    HAS_PYARROW12 = False
+from ..utils_test import HAS_PYARROW12, run_up_to_nthreads, wait
 
 
 @pytest.fixture(scope="module")
@@ -51,8 +43,7 @@ def parquet_client(parquet_cluster, cluster_kwargs, upload_cluster_dump, benchma
 
 
 @pytest.mark.xfail(
-    HAS_PYARROW12,
-    reason="50x slower than PyArrow 11; https://github.com/coiled/benchmarks/issues/998",
+    HAS_PYARROW12, reason="https://github.com/coiled/benchmarks/issues/998"
 )
 @run_up_to_nthreads("parquet_cluster", 100, reason="fixed dataset")
 def test_read_spark_generated_data(parquet_client):

tests/utils_test.py

@@ -12,6 +12,14 @@
 from dask.datasets import timeseries
 from dask.sizeof import sizeof
 from dask.utils import format_bytes, parse_bytes
+from packaging.version import Version
+
+try:
+    import pyarrow
+
+    HAS_PYARROW12 = Version(pyarrow.__version__) >= Version("12.0.0")
+except ImportError:
+    HAS_PYARROW12 = False
 
 
 def scaled_array_shape(