diff --git a/tensorflow_io/core/kernels/bigtable/bigtable_dataset_kernel.cc b/tensorflow_io/core/kernels/bigtable/bigtable_dataset_kernel.cc
index 596da7396..66db5e7d7 100644
--- a/tensorflow_io/core/kernels/bigtable/bigtable_dataset_kernel.cc
+++ b/tensorflow_io/core/kernels/bigtable/bigtable_dataset_kernel.cc
@@ -225,7 +225,9 @@ class Iterator : public DatasetIterator<Dataset> {
       filters.push_back(std::move(f));
     }
 
-    return cbt::Filter::InterleaveFromRange(filters.begin(), filters.end());
+    return filters.size() > 1 ? cbt::Filter::InterleaveFromRange(
+                                    filters.begin(), filters.end())
+                              : filters[0];
   }
 
   static std::pair<std::string, std::string> ColumnToFamilyAndQualifier(
diff --git a/tensorflow_io/python/ops/bigtable/bigtable_dataset_ops.py b/tensorflow_io/python/ops/bigtable/bigtable_dataset_ops.py
index 9ddbcffb5..4380b0ba1 100644
--- a/tensorflow_io/python/ops/bigtable/bigtable_dataset_ops.py
+++ b/tensorflow_io/python/ops/bigtable/bigtable_dataset_ops.py
@@ -57,7 +57,7 @@ def read_rows(
         self,
         columns: List[str],
         row_set: bigtable_row_set.RowSet,
-        filter: filters.BigtableFilter = filters.latest(),
+        filter: filters.BigtableFilter = None,
         output_type=tf.string,
     ):
         """Retrieves values from Google Bigtable sorted by RowKeys.
@@ -69,23 +69,24 @@ def read_rows(
         Returns:
             A `tf.data.Dataset` returning the cell contents.
         """
+
+        # Python initializes the default arguments once at the start of the
+        # program. If the fork happens after that (for instance when we run
+        # tests using xdist) the program deadlocks and hangs. That is why we
+        # have to make sure, all default arguments are initialized on each
+        # invocation.
+        if filter is None:
+            filter = filters.latest()
         return _BigtableDataset(
-            self._client_resource,
-            self._table_id,
-            columns,
-            row_set,
-            filter,
-            output_type,
+            self._client_resource, self._table_id, columns, row_set, filter, output_type
         )
 
     def parallel_read_rows(
         self,
         columns: List[str],
         num_parallel_calls=tf.data.AUTOTUNE,
-        row_set: bigtable_row_set.RowSet = bigtable_row_set.from_rows_or_ranges(
-            bigtable_row_range.infinite()
-        ),
-        filter: filters.BigtableFilter = filters.latest(),
+        row_set: bigtable_row_set.RowSet = None,
+        filter: filters.BigtableFilter = None,
         output_type=tf.string,
     ):
         """Retrieves values from Google Bigtable in parallel. The ammount of work
@@ -101,11 +102,17 @@ def parallel_read_rows(
             A `tf.data.Dataset` returning the cell contents.
         """
 
+        # We have to make sure that all the default arguments are initialized
+        # on each invocation. For more info see read_rows method.
+        if row_set is None:
+            row_set = bigtable_row_set.from_rows_or_ranges(
+                bigtable_row_range.infinite()
+            )
+        if filter is None:
+            filter = filters.latest()
+
         samples = core_ops.bigtable_split_row_set_evenly(
-            self._client_resource,
-            row_set._impl,
-            self._table_id,
-            num_parallel_calls,
+            self._client_resource, row_set._impl, self._table_id, num_parallel_calls
         )
 
         def map_func(idx):
diff --git a/tests/test_bigtable/bigtable_emulator.py b/tests/test_bigtable/bigtable_emulator.py
index 02cabe92f..975eb01ac 100644
--- a/tests/test_bigtable/bigtable_emulator.py
+++ b/tests/test_bigtable/bigtable_emulator.py
@@ -30,12 +30,14 @@
 CBT_EMULATOR_SEARCH_PATHS = [
     "/usr/lib/google-cloud-sdk/platform/bigtable-emulator/cbtemulator",
     "/usr/local/google-cloud-sdk/platform/bigtable-emulator/cbtemulator",
+    "/v/google-cloud-sdk/platform/bigtable-emulator/cbtemulator",
     "cbtemulator",
 ]
 
 CBT_CLI_SEARCH_PATHS = [
     "/usr/local/google-cloud-sdk/bin/cbt",
     "/usr/bin/cbt",
+    "/v/google-cloud-sdk/bin/cbt",
     "cbt",
 ]
 
diff --git a/tests/test_gcloud/test_pubsub_bigtable.sh b/tests/test_gcloud/test_pubsub_bigtable.sh
index ff504b924..71d1adec3 100755
--- a/tests/test_gcloud/test_pubsub_bigtable.sh
+++ b/tests/test_gcloud/test_pubsub_bigtable.sh
@@ -34,7 +34,7 @@ fi
 curl -sSOL https://dl.google.com/dl/cloudsdk/channels/rapid/downloads/google-cloud-sdk-236.0.0-darwin-x86_64.tar.gz
 tar -xzf google-cloud-sdk-236.0.0-darwin-x86_64.tar.gz
 google-cloud-sdk/install.sh -q
-google-cloud-sdk/bin/gcloud -q components install beta
+google-cloud-sdk/bin/gcloud -q components install beta bigtable cbt
 google-cloud-sdk/bin/gcloud -q components install pubsub-emulator
 google-cloud-sdk/bin/gcloud -q components update beta
 google-cloud-sdk/bin/gcloud -q beta emulators pubsub start &