fix: Improve caching of work_group lookup (#678)

dbt/adapters/athena/impl.py

@@ -19,7 +19,6 @@
 from dbt_common.clients.agate_helper import table_from_rows
 from dbt_common.contracts.constraints import ConstraintType
 from dbt_common.exceptions import DbtRuntimeError
-from mypy_boto3_athena import AthenaClient
 from mypy_boto3_athena.type_defs import DataCatalogTypeDef, GetWorkGroupOutputTypeDef
 from mypy_boto3_glue.type_defs import (
     ColumnTypeDef,
@@ -218,14 +217,11 @@ def apply_lf_grants(self, relation: AthenaRelation, lf_grants_config: Dict[str,
             lf_permissions.process_permissions(lf_config)
 
     @lru_cache()
-    def _get_work_group(self, client: AthenaClient, work_group: str) -> GetWorkGroupOutputTypeDef:
+    def _get_work_group(self, work_group: str) -> GetWorkGroupOutputTypeDef:
         """
         helper function to cache the result of the get_work_group to avoid APIs throttling
         """
-        return client.get_work_group(WorkGroup=work_group)
-
-    @available
-    def is_work_group_output_location_enforced(self) -> bool:
+        LOGGER.debug("get_work_group for %s", work_group)
         conn = self.connections.get_thread_connection()
         creds = conn.credentials
         client = conn.handle
@@ -237,8 +233,15 @@ def is_work_group_output_location_enforced(self) -> bool:
                 config=get_boto3_config(num_retries=creds.effective_num_retries),
             )
 
+        return athena_client.get_work_group(WorkGroup=work_group)
+
+    @available
+    def is_work_group_output_location_enforced(self) -> bool:
+        conn = self.connections.get_thread_connection()
+        creds = conn.credentials
+
         if creds.work_group:
-            work_group = self._get_work_group(athena_client, creds.work_group)
+            work_group = self._get_work_group(creds.work_group)
             output_location = (
                 work_group.get("WorkGroup", {})
                 .get("Configuration", {})