diff --git a/ci/azure/install.yml b/ci/azure/install.yml
index 958e3c180fa..60559dd2064 100644
--- a/ci/azure/install.yml
+++ b/ci/azure/install.yml
@@ -19,7 +19,6 @@ steps:
         --upgrade \
         matplotlib \
         numpy \
-        pandas \
         scipy
     python -m pip install \
         --no-deps \
@@ -30,7 +29,8 @@ steps:
         git+https://github.com/Unidata/cftime \
         git+https://github.com/mapbox/rasterio \
         git+https://github.com/hgrecco/pint \
-        git+https://github.com/pydata/bottleneck
+        git+https://github.com/pydata/bottleneck \
+        git+https://github.com/pandas-dev/pandas
   condition: eq(variables['UPSTREAM_DEV'], 'true')
   displayName: Install upstream dev dependencies
 
diff --git a/xarray/coding/cftimeindex.py b/xarray/coding/cftimeindex.py
index c680a7e0bcf..2e42702caac 100644
--- a/xarray/coding/cftimeindex.py
+++ b/xarray/coding/cftimeindex.py
@@ -253,6 +253,7 @@ def __new__(cls, data, name=None):
         result = object.__new__(cls)
         result._data = np.array(data, dtype="O")
         result.name = name
+        result._cache = {}
         return result
 
     def _partial_date_slice(self, resolution, parsed):
diff --git a/xarray/tests/test_cftimeindex.py b/xarray/tests/test_cftimeindex.py
index 43d6d7b068e..d31bf9471ea 100644
--- a/xarray/tests/test_cftimeindex.py
+++ b/xarray/tests/test_cftimeindex.py
@@ -450,7 +450,6 @@ def test_sel_date_scalar(da, date_type, index):
     assert_identical(result, expected)
 
 
-@pytest.mark.xfail(reason="https://github.com/pydata/xarray/issues/3751")
 @requires_cftime
 def test_sel_date_distant_date(da, date_type, index):
     expected = xr.DataArray(4).assign_coords(time=index[3])
@@ -513,12 +512,7 @@ def test_sel_date_scalar_backfill(da, date_type, index, sel_kwargs):
     [
         {"method": "pad", "tolerance": timedelta(days=20)},
         {"method": "backfill", "tolerance": timedelta(days=20)},
-        pytest.param(
-            {"method": "nearest", "tolerance": timedelta(days=20)},
-            marks=pytest.mark.xfail(
-                reason="https://github.com/pydata/xarray/issues/3751"
-            ),
-        ),
+        {"method": "nearest", "tolerance": timedelta(days=20)},
     ],
 )
 def test_sel_date_scalar_tolerance_raises(da, date_type, sel_kwargs):
@@ -526,7 +520,6 @@ def test_sel_date_scalar_tolerance_raises(da, date_type, sel_kwargs):
         da.sel(time=date_type(1, 5, 1), **sel_kwargs)
 
 
-@pytest.mark.xfail(reason="https://github.com/pydata/xarray/issues/3751")
 @requires_cftime
 @pytest.mark.parametrize(
     "sel_kwargs",
@@ -574,12 +567,7 @@ def test_sel_date_list_backfill(da, date_type, index, sel_kwargs):
     [
         {"method": "pad", "tolerance": timedelta(days=20)},
         {"method": "backfill", "tolerance": timedelta(days=20)},
-        pytest.param(
-            {"method": "nearest", "tolerance": timedelta(days=20)},
-            marks=pytest.mark.xfail(
-                reason="https://github.com/pydata/xarray/issues/3751"
-            ),
-        ),
+        {"method": "nearest", "tolerance": timedelta(days=20)},
     ],
 )
 def test_sel_date_list_tolerance_raises(da, date_type, sel_kwargs):
@@ -614,7 +602,6 @@ def range_args(date_type):
     ]
 
 
-@pytest.mark.xfail(reason="https://github.com/pydata/xarray/issues/3751")
 @requires_cftime
 def test_indexing_in_series_getitem(series, index, scalar_args, range_args):
     for arg in scalar_args:
diff --git a/xarray/tests/test_interp.py b/xarray/tests/test_interp.py
index 9cc4933f462..0502348160e 100644
--- a/xarray/tests/test_interp.py
+++ b/xarray/tests/test_interp.py
@@ -586,7 +586,6 @@ def test_datetime_single_string():
     assert_allclose(actual.drop_vars("time"), expected)
 
 
-@pytest.mark.xfail(reason="https://github.com/pydata/xarray/issues/3751")
 @requires_cftime
 @requires_scipy
 def test_cftime():
@@ -613,7 +612,6 @@ def test_cftime_type_error():
         da.interp(time=times_new)
 
 
-@pytest.mark.xfail(reason="https://github.com/pydata/xarray/issues/3751")
 @requires_cftime
 @requires_scipy
 def test_cftime_list_of_strings():
@@ -635,7 +633,6 @@ def test_cftime_list_of_strings():
     assert_allclose(actual, expected)
 
 
-@pytest.mark.xfail(reason="https://github.com/pydata/xarray/issues/3751")
 @requires_cftime
 @requires_scipy
 def test_cftime_single_string():
@@ -697,7 +694,6 @@ def test_datetime_interp_noerror():
     a.interp(x=xi, time=xi.time)  # should not raise an error
 
 
-@pytest.mark.xfail(reason="https://github.com/pydata/xarray/issues/3751")
 @requires_cftime
 def test_3641():
     times = xr.cftime_range("0001", periods=3, freq="500Y")