Knowledge Agent Issue

[InnovationMaitools]

025-07-21 11:58:52.328 INFO {task_manager} [_listen_transcriber] Received transcript, sending for further processing
2025-07-21 11:58:52.329 INFO {task_manager} [_handle_transcriber_output] Running llm Tasks
2025-07-21 11:58:52.856 INFO {_client} [_send_single_request] HTTP Request: POST https://api.openai.com/v1/embeddings "HTTP/1.1 200 OK"
2025-07-21 11:58:52.868 INFO {base} [query] query_type :, vector
Traceback (most recent call last):
File "/usr/local/lib/python3.10/site-packages/bolna/agent_manager/task_manager.py", line 1389, in _run_llm_task
await self._process_conversation_task(message, sequence, meta_info)
File "/usr/local/lib/python3.10/site-packages/bolna/agent_manager/task_manager.py", line 1326, in _process_conversation_task
await self.__do_llm_generation(messages, meta_info, next_step, should_bypass_synth)
File "/usr/local/lib/python3.10/site-packages/bolna/agent_manager/task_manager.py", line 1222, in __do_llm_generation
async for llm_message in self.tools['llm_agent'].generate(messages, synthesize=synthesize, meta_info=meta_info):
File "/usr/local/lib/python3.10/site-packages/bolna/agent_types/knowledgebase_agent.py", line 116, in generate
response = await self.query_engine.aquery(message.content)
File "/usr/local/lib/python3.10/site-packages/llama_index/core/instrumentation/dispatcher.py", line 290, in async_wrapper
result = await func(*args, **kwargs)
File "/usr/local/lib/python3.10/site-packages/llama_index/core/base/base_query_engine.py", line 64, in aquery
query_result = await self._aquery(str_or_query_bundle)
File "/usr/local/lib/python3.10/site-packages/llama_index/core/instrumentation/dispatcher.py", line 290, in async_wrapper
result = await func(*args, **kwargs)
File "/usr/local/lib/python3.10/site-packages/llama_index/core/query_engine/retriever_query_engine.py", line 204, in _aquery
nodes = await self.aretrieve(query_bundle)
File "/usr/local/lib/python3.10/site-packages/llama_index/core/query_engine/retriever_query_engine.py", line 148, in aretrieve
nodes = await self._retriever.aretrieve(query_bundle)
File "/usr/local/lib/python3.10/site-packages/llama_index/core/instrumentation/dispatcher.py", line 290, in async_wrapper
result = await func(*args, **kwargs)
File "/usr/local/lib/python3.10/site-packages/llama_index/core/base/base_retriever.py", line 274, in aretrieve
nodes = await self._aretrieve(query_bundle=query_bundle)
File "/usr/local/lib/python3.10/site-packages/llama_index/core/instrumentation/dispatcher.py", line 290, in async_wrapper
result = await func(*args, **kwargs)
File "/usr/local/lib/python3.10/site-packages/llama_index/core/indices/vector_store/retrievers/retriever.py", line 112, in _aretrieve
return await self._aget_nodes_with_embeddings(
File "/usr/local/lib/python3.10/site-packages/llama_index/core/indices/vector_store/retrievers/retriever.py", line 184, in _aget_nodes_with_embeddings
query_result = await self._vector_store.aquery(query, **self._kwargs)
File "/usr/local/lib/python3.10/site-packages/llama_index/core/vector_stores/types.py", line 420, in aquery
return self.query(query, **kwargs)
File "/usr/local/lib/python3.10/site-packages/llama_index/vector_stores/lancedb/base.py", line 484, in query
self._table.search(
AttributeError: 'NoneType' object has no attribute 'search'
2025-07-21 11:58:52.871 ERROR {task_manager} [_run_llm_task] Something went wrong in llm: 'NoneType' object has no attribute 'search'
2025-07-21 11:58:53.309 INFO {deepgram_transcriber} [receiver] Value of is_transcript_sent_for_processing in utterance end - True
2025-07-21 11:58:53.498 INFO {deepgram_transcriber} [receiver] Received SpeechStarted event from deepgram

[0x-Jayveersinh-Raj]

In have tested with lancedb and it works, you have to do the following, as it assumes an existing lancedb table. Here is the overview:

• The LanceDB table wasn’t opened/created, so the underlying table handle is None. Can be one or more of the below causes:
  • Wrong or missing LANCEDB_DIR
  • Wrong vector_id (table name) in provider config
  • Table never created/ingested (no prior run to build the index)
  • Files missing/corrupted at the LanceDB path
    OpenAI embeddings succeeded, then the vector query immediately failed because there was no valid LanceDB table object to query.

RAG flow assumes a valid LanceDB table exists and works when LANCEDB_DIR and vector_id are correct and the table has been created.

You can do the following:

• Create a script at the root level create_lancedb_table.py to build the table (You can find examples online, or if you can't figure it out write again on this, I'll give you the script)
• run the script (python create_lancedb_table.py --docs_dir ./docs --vector_id --lancedb_dir )
• add LANCEDB_DIR=/data/lancedb (This will be living inside your container)
• Change the compose file to volume mount the local directory

NOTE: Pass "embedding_model": "text-embedding-3-large" in your json under llm_agent.llm_config.vector_store.provider_config, if you use that otherwise it defaults to text-embedding-3-small, keep in mind to change accordingly. Also, it works better with txt and md files.