update

graphrag/general/community_reports_extractor.py

@@ -150,6 +150,7 @@ class CommunityReportsExtractor(Extractor):
         try:
             await asyncio.gather(*tasks, return_exceptions=False)
         except Exception as e:
+            logging.error(f"Error in community processing: {e}")
             for t in tasks:
                 t.cancel()
             await asyncio.gather(*tasks, return_exceptions=True)

graphrag/general/extractor.py

@@ -145,6 +145,7 @@ class Extractor:
             try:
                 await asyncio.gather(*tasks, return_exceptions=False)
             except Exception as e:
+                logging.error(f"Error in worker: {str(e)}")
                 for t in tasks:
                     t.cancel()
                 await asyncio.gather(*tasks, return_exceptions=True)

graphrag/general/index.py

@@ -286,6 +286,7 @@ async def run_graphrag_for_kb(
     try:
         await asyncio.gather(*tasks, return_exceptions=False)
     except Exception as e:
+        logging.error(f"Error in asyncio.gather: {e}")
         for t in tasks:
             t.cancel()
         await asyncio.gather(*tasks, return_exceptions=True)

graphrag/utils.py

@@ -531,6 +531,7 @@ async def set_graph(tenant_id: str, kb_id: str, embd_mdl, graph: nx.Graph, chang
     try:
         await asyncio.gather(*tasks, return_exceptions=False)
     except Exception as e:
+        logging.error(f"Error in get_embedding_of_nodes: {e}")
         for t in tasks:
             t.cancel()
         await asyncio.gather(*tasks, return_exceptions=True)
@@ -549,6 +550,7 @@ async def set_graph(tenant_id: str, kb_id: str, embd_mdl, graph: nx.Graph, chang
     try:
         await asyncio.gather(*tasks, return_exceptions=False)
     except Exception as e:
+        logging.error(f"Error in get_embedding_of_edges: {e}")
         for t in tasks:
             t.cancel()
         await asyncio.gather(*tasks, return_exceptions=True)

rag/flow/hierarchical_merger/hierarchical_merger.py

@@ -14,6 +14,7 @@
 #  limitations under the License.
 
 import asyncio
+import logging
 import random
 import re
 from copy import deepcopy
@@ -183,6 +184,7 @@ class HierarchicalMerger(ProcessBase):
             try:
                 await asyncio.gather(*tasks, return_exceptions=False)
             except Exception as e:
+                logging.error(f"Error in image2id: {e}")
                 for t in tasks:
                     t.cancel()
                 await asyncio.gather(*tasks, return_exceptions=True)

rag/prompts/generator.py

@@ -752,6 +752,7 @@ async def run_toc_from_text(chunks, chat_mdl, callback=None):
     try:
         await asyncio.gather(*tasks, return_exceptions=False)
     except Exception as e:
+        logging.error(f"Error generating TOC: {e}")
         for t in tasks:
             t.cancel()
         await asyncio.gather(*tasks, return_exceptions=True)

rag/raptor.py

@@ -209,6 +209,7 @@ class RecursiveAbstractiveProcessing4TreeOrganizedRetrieval:
             try:
                 await asyncio.gather(*tasks, return_exceptions=False)
             except Exception as e:
+                logging.error(f"Error in RAPTOR cluster processing: {e}")
                 for t in tasks:
                     t.cancel()
                 await asyncio.gather(*tasks, return_exceptions=True)

rag/svr/sync_data_source.py

@@ -664,6 +664,7 @@ async def dispatch_tasks():
     try:
         await asyncio.gather(*tasks, return_exceptions=False)
     except Exception as e:
+        logging.error(f"Error in dispatch_tasks: {e}")
         for t in tasks:
             t.cancel()
         await asyncio.gather(*tasks, return_exceptions=True)

rag/svr/task_executor.py

@@ -305,6 +305,7 @@ async def build_chunks(task, progress_callback):
     try:
         await asyncio.gather(*tasks, return_exceptions=False)
     except Exception as e:
+        logging.error("Saving chunk {}/{}/{} got exception".format(task["location"], task["name"], d["id"]))
         for t in tasks:
             t.cancel()
         await asyncio.gather(*tasks, return_exceptions=True)
@@ -830,6 +831,7 @@ async def insert_es(task_id, task_tenant_id, task_dataset_id, chunks, progress_c
             try:
                 await asyncio.gather(*tasks, return_exceptions=False)
             except Exception as e:
+                logging.error(f"do_handle_task delete_image failed since task {task_id} is unknown.")
                 for t in tasks:
                     t.cancel()
                 await asyncio.gather(*tasks, return_exceptions=True)