[BUG] Error in handler LangChainTracer, handleChainStart: TimeoutError: The operation was aborted due to timeout #3292

[kientv]

It's always slow or timeout at first message of every conversation. Does the chain stop after a period of time?

chatbot | 2023-12-27 03:10:40 [INFO]: ⬆️ POST /api/v1/prediction/037cd272-8dbd-425a-a87a-4bd96b30116d chatbot | [chain/start] [1:chain:ConversationalRetrievalQAChain] Entering Chain run with input: { chatbot | "question": "Hello", chatbot | "chat_history": [] chatbot | } chatbot | Error in handler LangChainTracer, handleChainStart: TimeoutError: The operation was aborted due to timeout chatbot | [retriever/start] [1:chain:ConversationalRetrievalQAChain > 2:retriever:VectorStoreRetriever] Entering Retriever run with input: { chatbot | "query": "Hello" chatbot | } qdrant | 2023-12-27T03:12:36.603842Z INFO actix_web::middleware::logger: 192.168.16.4 "GET /collections HTTP/1.1" 200 184 "-" "qdrant-js" 0.088474 qdrant | 2023-12-27T03:12:39.099975Z INFO actix_web::middleware::logger: 192.168.16.4 "POST /collections/gemini768/points/search HTTP/1.1" 200 1624 "-" "qdrant-js" 1.449681 chatbot | [retriever/end] [1:chain:ConversationalRetrievalQAChain > 2:retriever:VectorStoreRetriever] [6.94s] Exiting Retriever run with output: { chatbot | "documents": [

...

chatbot | Error in handler LangChainTracer, handleLLMEnd: TimeoutError: The operation was aborted due to timeout chatbot | [chain/end] [1:chain:ConversationalRetrievalQAChain > 5:chain:StuffDocumentsChain > 6:chain:LLMChain] [149.38s] Exiting Chain run with output: {

[vinodkiran]

Do you have the LANGCHAIN_TRACING_V2 variable enabled in the server/.env file ? This error is seen when you have this flag enabled without a corresponding LANGCHAIN_ENDPOINT in the .env

Comment out these variables and use the Chatflow / Analytics to configure LANGCHAIN tracing.

[kientv]

Do you have the LANGCHAIN_TRACING_V2 variable enabled in the server/.env file ? This error is seen when you have this flag enabled without a corresponding LANGCHAIN_ENDPOINT in the .env

Comment out these variables and use the Chatflow / Analytics to configure LANGCHAIN tracing.

Thank you.

But I think that it's not problem of langsmith config. first message of the conversation is very slow. Nevertheless the performance after that (from second message) is good. First message maybe take 121.94s to finish as this trace: https://smith.langchain.com/public/430f55dc-691d-4512-8fea-499469619a09/r