[COST-4289] - Skip OCP clusters without a matching billing report dat…

…e for OCP on Cloud processing (#4735)

* [COST-4289] Skip clusters for OCP on Cloud matching/summary tasks w/o a matching report period

koku/masu/processor/ocp/ocp_cloud_parquet_summary_updater.py

@@ -119,7 +119,27 @@ def get_infra_map(self, start_date, end_date):
             self.provider_type in Provider.CLOUD_PROVIDER_LIST and self._provider_uuid not in infra_provider_uuids
         ):
             infra_map = self._generate_ocp_infra_map_from_sql(start_date, end_date)
-        return infra_map
+
+        summary_infra_map = {}
+        with OCPReportDBAccessor(self._schema) as accessor:
+            # Only include OCP providers for OCP on Cloud summary if we have a matching report period
+            for ocp_provider_uuid in infra_map:
+                if accessor.report_periods_for_provider_uuid(ocp_provider_uuid, start_date):
+                    summary_infra_map[ocp_provider_uuid] = infra_map[ocp_provider_uuid]
+                else:
+                    ctx = {
+                        "schema": self._schema,
+                        "ocp_provider_uuid": ocp_provider_uuid,
+                        "provider_uuid": infra_map[ocp_provider_uuid][0],
+                        "provider_type": infra_map[ocp_provider_uuid][1],
+                    }
+                    LOG.info(
+                        log_json(
+                            msg=f"no matching report periods available for cluster - removing from OCP on {infra_map[ocp_provider_uuid][1]} summary list",  # noqa: E501
+                            context=ctx,
+                        )
+                    )
+        return summary_infra_map
 
     def determine_truncates_and_deletes(self, start_date, end_date):
         """Clear out existing data in summary tables."""

koku/masu/processor/parquet/ocp_cloud_parquet_report_processor.py

@@ -227,6 +227,15 @@ def get_ocp_provider_uuids_tuple(self):
                         )
                     )
                     continue
+                # Check we have data for the ocp provider/cluster matching the current report period
+                if not accessor.report_periods_for_provider_uuid(ocp_provider_uuid, self.start_date):
+                    LOG.info(
+                        log_json(
+                            msg=f"no matching report periods available for cluster - skipping OCP on {self.provider_type} parquet processing",  # noqa: E501
+                            context=ctx,
+                        )
+                    )
+                    continue
                 ocp_provider_uuids.append(ocp_provider_uuid)
         return tuple(ocp_provider_uuids)
 

koku/masu/test/processor/ocp/test_ocp_cloud_parquet_report_summary_updater.py

@@ -518,6 +518,18 @@ def test_get_infra_map_from_providers(self):
         self.assertIn(str(self.ocp_on_aws_ocp_provider.uuid), infra_map)
         self.assertEqual(infra_map.get(str(self.ocp_on_aws_ocp_provider.uuid)), expected_mapping)
 
+    def test_get_infra_map_for_summary(self):
+        """Test that an infrastructure map is returned for active providers."""
+        updater = OCPCloudParquetReportSummaryUpdater(
+            schema=self.schema, provider=self.ocp_on_aws_ocp_provider, manifest=None
+        )
+        start_date = "1900-12-30"
+        end_date = "1900-12-31"
+
+        infra_map = updater.get_infra_map(start_date, end_date)
+        # We shouldnt collect any providers for summary with these given dates.
+        self.assertEqual(len(infra_map.keys()), 0)
+
     def test_partition_handler_str_table(self):
         new_table_sql = f"""
 create table {self.schema}._eek_pt0 (usage_start date not null, id int) partition by range (usage_start);

koku/masu/test/processor/parquet/test_ocp_cloud_parquet_report_processor.py

@@ -307,6 +307,21 @@ def test_process_no_cluster_info(self, mock_data_processor, mock_create_parquet,
         mock_data_processor.assert_not_called()
         mock_create_parquet.assert_not_called()
 
+    @patch.object(OCPReportDBAccessor, "report_periods_for_provider_uuid")
+    @patch.object(OCPReportDBAccessor, "get_cluster_for_provider")
+    @patch.object(OCPCloudParquetReportProcessor, "create_ocp_on_cloud_parquet")
+    @patch.object(OCPCloudParquetReportProcessor, "ocp_on_cloud_data_processor")
+    def test_process_no_matching_cluster_report_period(
+        self, mock_data_processor, mock_create_parquet, mock_cluster_info, mock_report_periods
+    ):
+        """Test that ocp on cloud data is not processed when there is no matching report period for the cluster."""
+        # this is a yes or no check so false is fine
+        mock_cluster_info.return_value = True
+        mock_report_periods.return_value = False
+        self.report_processor.process("", [pd.DataFrame()])
+        mock_data_processor.assert_not_called()
+        mock_create_parquet.assert_not_called()
+
     @patch.object(OCPReportDBAccessor, "get_cluster_for_provider")
     @patch.object(OCPReportDBAccessor, "get_openshift_topology_for_multiple_providers")
     @patch.object(OCPCloudParquetReportProcessor, "has_enabled_ocp_labels")