LANG-134: Finish original inbound monitoring row on pipeline error (#2729)

* fix(inbound-error-state): Finish original inbound monitoring row on pipeline error (LANG-134)

* fix(pipeline): finalize monitoring on early exits and cancellation

Co-authored-by: multica-agent <[email protected]>

---------

Co-authored-by: multica-agent <[email protected]>

src/langbot/pkg/pipeline/monitoring_helper.py

@@ -6,6 +6,7 @@ without cluttering the main pipeline code.
 
 from __future__ import annotations
 
+import asyncio
 import traceback
 import typing
 import time
@@ -118,6 +119,25 @@ async def prepare_monitoring_identity(
 class MonitoringHelper:
     """Helper class for monitoring operations"""
 
+    @staticmethod
+    async def finish_query(ap, query, message_id: str, status: str) -> None:
+        """Bound terminal writes and retry transient failures without blocking cleanup."""
+        if not message_id:
+            return
+        for attempt in range(3):
+            try:
+                async with asyncio.timeout(2):
+                    await ap.monitoring_service.update_message_status(
+                        get_query_execution_context(query),
+                        message_id=message_id,
+                        status=status,
+                        level='error' if status == 'error' else 'info',
+                    )
+                return
+            except Exception as exc:
+                ap.logger.warning(f'Terminal monitoring write failed (attempt {attempt + 1}): {exc}')
+        ap.logger.error(f'Unable to finish monitoring message {message_id}: {status}')
+
     @staticmethod
     async def record_query_start(
         ap: app.Application,
@@ -283,6 +303,9 @@ class MonitoringHelper:
     ) -> str:
         """Record query processing error, returns message_id"""
         try:
+            original_message_id = (query.variables or {}).get('_monitoring_message_id')
+            if original_message_id:
+                await MonitoringHelper.finish_query(ap, query, original_message_id, 'error')
             session_id = f'{query.launcher_type.value if hasattr(query.launcher_type, "value") else query.launcher_type}_{query.launcher_id}'
 
             # Get sender name from message event

src/langbot/pkg/pipeline/pipelinemgr.py

@@ -497,6 +497,28 @@ class RuntimePipeline:
             processing_mode.reset(token)
 
     async def _process_query(self, query: pipeline_query.Query):
+        """Close monitoring on every exit, including cancellation and plugin veto."""
+        from .monitoring_helper import MonitoringHelper
+
+        status = 'error'
+        try:
+            await self._process_query_body(query)
+            if not query.variables.get('_monitoring_has_error'):
+                status = 'success'
+        finally:
+            try:
+                # Never bypass the placement fence to repair a stale execution.
+                async with asyncio.timeout(2):
+                    await self._assert_execution_active(query)
+                await MonitoringHelper.finish_query(
+                    self.ap, query, query.variables.get('_monitoring_message_id', ''), status
+                )
+            except Exception as exc:
+                self.ap.logger.error(f'Failed to finalize query {query.query_id}: {exc}')
+            finally:
+                await self.ap.query_pool.remove_query(query)
+
+    async def _process_query_body(self, query: pipeline_query.Query):
         """处理请求"""
         await self._assert_execution_active(query)
         # Get monitoring metadata
@@ -602,8 +624,10 @@ class RuntimePipeline:
                     self.ap.logger.error(f'Failed to record query response: {e}')
 
         except WorkspaceError as e:
+            query.variables['_monitoring_has_error'] = True
             self.ap.logger.info(f'Dropped query {query.query_id} because its Workspace execution binding is stale: {e}')
         except Exception as e:
+            query.variables['_monitoring_has_error'] = True
             inst_name = query.current_stage_name if query.current_stage_name else 'unknown'
             self.ap.logger.error(f'Error processing query {query.query_id} stage={inst_name} : {e}')
             self.ap.logger.error(f'Traceback: {traceback.format_exc()}')
@@ -631,7 +655,6 @@ class RuntimePipeline:
 
         finally:
             self.ap.logger.debug(f'Query {query.query_id} processed')
-            await self.ap.query_pool.remove_query(query)
 
     def _record_lane_failure(self, query: pipeline_query.Query, reason: str) -> None:
         """Land one failed ``runner/execute`` stage on the owning execution trace."""