nv-ingest-client 2025.8.6.dev20250806__tar.gz → 2025.8.8.dev20250808__tar.gz

{nv_ingest_client-2025.8.6.dev20250806/src/nv_ingest_client.egg-info → nv_ingest_client-2025.8.8.dev20250808}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nv-ingest-client
-Version: 2025.8.6.dev20250806
+Version: 2025.8.8.dev20250808
 Summary: Python client for the nv-ingest service
 Author-email: Jeremy Dyer <jdyer@nvidia.com>
 License:                                  Apache License

{nv_ingest_client-2025.8.6.dev20250806 → nv_ingest_client-2025.8.8.dev20250808}/src/nv_ingest_client/client/interface.py

@@ -535,10 +535,6 @@ class Ingestor:
 
         proc_kwargs = filter_function_kwargs(self._client.process_jobs_concurrently, **kwargs)
 
-        _return_failures = return_failures
-        if self._vdb_bulk_upload:
-            return_failures = True
-
         results, failures = self._client.process_jobs_concurrently(
             job_indices=self._job_ids,
             job_queue_id=self._job_queue_id,
@@ -567,15 +563,39 @@ class Ingestor:
 
         if self._vdb_bulk_upload:
             if len(failures) > 0:
-                raise RuntimeError(f"Failed to ingest documents, unable to complete vdb bulk upload: {failures}")
+                # Calculate success metrics
+                total_jobs = len(results) + len(failures)
+                successful_jobs = len(results)
+
+                if return_failures:
+                    # Emit message about partial success
+                    logger.warning(
+                        f"Job was not completely successful. "
+                        f"{successful_jobs} out of {total_jobs} records completed successfully. "
+                        f"Uploading successful results to vector database."
+                    )
+
+                    # Upload only the successful results
+                    if successful_jobs > 0:
+                        self._vdb_bulk_upload.run(results)
 
-            self._vdb_bulk_upload.run(results)
+                        if self._purge_results_after_vdb_upload:
+                            logger.info("Purging saved results from disk after successful VDB upload.")
+                            self._purge_saved_results(results)
 
-            if self._purge_results_after_vdb_upload:
-                logger.info("Purging saved results from disk after successful VDB upload.")
-                self._purge_saved_results(results)
+                else:
+                    # Original behavior: raise RuntimeError
+                    raise RuntimeError(
+                        "Failed to ingest documents, unable to complete vdb bulk upload due to "
+                        f"no successful results. {len(failures)} out of {total_jobs} records failed "
+                    )
+            else:
+                # No failures - proceed with normal upload
+                self._vdb_bulk_upload.run(results)
 
-        return_failures = _return_failures
+                if self._purge_results_after_vdb_upload:
+                    logger.info("Purging saved results from disk after successful VDB upload.")
+                    self._purge_saved_results(results)
 
         return (results, failures) if return_failures else results
 

{nv_ingest_client-2025.8.6.dev20250806 → nv_ingest_client-2025.8.8.dev20250808}/src/nv_ingest_client/util/vdb/milvus.py

@@ -729,7 +729,7 @@ def write_records_minio(records, writer: RemoteBulkWriter) -> RemoteBulkWriter:
     for element in records:
         writer.append_row(element)
     writer.commit()
-    print(f"Wrote data to: {writer.batch_files}")
+    logger.debug(f"Wrote data to: {writer.batch_files}")
     return writer
 
 
@@ -757,9 +757,10 @@ def bulk_insert_milvus(
 
     connections.connect(uri=milvus_uri)
     t_bulk_start = time.time()
+    files_to_upload = [_file for file_set in writer.batch_files for _file in file_set]
     task_id = utility.do_bulk_insert(
         collection_name=collection_name,
-        files=writer.batch_files[0],
+        files=files_to_upload,
         consistency_level=CONSISTENCY,
     )
     # list_bulk_insert_tasks = utility.list_bulk_insert_tasks(collection_name=collection_name)
@@ -769,11 +770,11 @@ def bulk_insert_milvus(
         state = task.state_name
         if state == "Completed":
             t_bulk_end = time.time()
-            print("Start time:", task.create_time_str)
-            print("Imported row count:", task.row_count)
-            print(f"Bulk {collection_name} upload took {t_bulk_end - t_bulk_start} s")
+            logger.info("Start time:", task.create_time_str)
+            logger.info("Imported row count:", task.row_count)
+            logger.info(f"Bulk {collection_name} upload took {t_bulk_end - t_bulk_start} s")
         if task.state == BulkInsertState.ImportFailed:
-            print("Failed reason:", task.failed_reason)
+            logger.error("Failed reason:", task.failed_reason)
         time.sleep(1)
 
 

{nv_ingest_client-2025.8.6.dev20250806 → nv_ingest_client-2025.8.8.dev20250808/src/nv_ingest_client.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nv-ingest-client
-Version: 2025.8.6.dev20250806
+Version: 2025.8.8.dev20250808
 Summary: Python client for the nv-ingest service
 Author-email: Jeremy Dyer <jdyer@nvidia.com>
 License:                                  Apache License