Assert dask client at early stage. (#5048)

python-package/xgboost/dask.py

@@ -105,6 +105,12 @@ def _get_client_workers(client):
     return workers
 
 
+def _assert_client(client):
+    if not isinstance(client, (type(get_client()), type(None))):
+        raise TypeError(
+            _expect([type(get_client()), type(None)], type(client)))
+
+
 class DaskDMatrix:
     # pylint: disable=missing-docstring, too-many-instance-attributes
     '''DMatrix holding on references to Dask DataFrame or Dask Array.
@@ -142,6 +148,7 @@ def __init__(self,
                  feature_names=None,
                  feature_types=None):
         _assert_dask_support()
+        _assert_client(client)
 
         self._feature_names = feature_names
         self._feature_types = feature_types
@@ -362,11 +369,12 @@ def train(client, params, dtrain, *args, evals=(), **kwargs):
 
     '''
     _assert_dask_support()
-
+    _assert_client(client)
     if 'evals_result' in kwargs.keys():
         raise ValueError(
             'evals_result is not supported in dask interface.',
             'The evaluation history is returned as result of training.')
+
     client = _xgb_get_client(client)
     workers = list(_get_client_workers(client).keys())
 
@@ -432,6 +440,7 @@ def predict(client, model, data, *args):
 
     '''
     _assert_dask_support()
+    _assert_client(client)
     if isinstance(model, Booster):
         booster = model
     elif isinstance(model, dict):