Merge branch 'main' into tchow/catalog-switch

Author: tchow-zlai

api/python/ai/chronon/cli/compile/parse_teams.py

@@ -99,7 +99,7 @@ def update_metadata(obj: Any, team_dict: Dict[str, Team]):
     if isinstance(obj, Join):
         join_namespace = obj.metaData.outputNamespace
         # set the metadata for each join part and labelParts
-        def set_join_part_metadata(join_part_gb, output_namespace):
+        def set_group_by_metadata(join_part_gb, output_namespace):
             if join_part_gb is not None:
                 if join_part_gb.metaData:
                     # Only set the outputNamespace if it hasn't been set already
@@ -112,11 +112,11 @@ def set_join_part_metadata(join_part_gb, output_namespace):
 
         if obj.joinParts:
             for jp in (obj.joinParts or []):
-                set_join_part_metadata(jp.groupBy, join_namespace)
+                set_group_by_metadata(jp.groupBy, join_namespace)
 
         if obj.labelParts:
             for lb in (obj.labelParts.labels or []):
-                set_join_part_metadata(lb, join_namespace)
+                set_group_by_metadata(lb.groupBy, join_namespace)
 
     if metadata.executionInfo is None:
         metadata.executionInfo = ExecutionInfo()

api/python/ai/chronon/repo/gcp.py

@@ -437,11 +437,15 @@ def run(self):
                 """
                 )
                 check_call(
-                    f"gcloud dataproc jobs wait {job_id} --region={GcpRunner.get_gcp_region_id()}"
+                    f"gcloud dataproc jobs wait {job_id} --region={GcpRunner.get_gcp_region_id()} "
+                    f"--project={GcpRunner.get_gcp_project_id()}"
                 )
 
                 # Fetch the final job state
-                jobs_info_str = check_output(f"gcloud dataproc jobs describe {job_id} --region=us-central1 --format=json").decode("utf-8")
+                jobs_info_str = (check_output(
+                    f"gcloud dataproc jobs describe {job_id} --region={GcpRunner.get_gcp_region_id()} "
+                    f"--project={GcpRunner.get_gcp_project_id()} --format=json")
+                                 .decode("utf-8"))
                 job_info = json.loads(jobs_info_str)
                 job_state = job_info.get("status", {}).get("state", "")