500 Internal Server Error, 502 error

[pandaupc]

Certainly, here is the complete transcription of the error:

I'm currently able to import files and provide regular answers. However, an error occurs when providing document-based answers. Upon investigation, the error seems to occur when fetching the context for computing similarity, specifically when calling the similarity interface of Supabase. The error message is as follows:

arduino
Copy code
INFO: 127.0.0.1:59312 - "POST /chat/ef77183d-85bb-4389-afe9-6d13dcc0237b/question?brain_id=242a3a9c-f0a5-4993-b9f1-44efdf22c2b4 HTTP/1.1" 500 Internal Server Error
ERROR: Exception in ASGI application
Traceback (most recent call last):
...
File "/Users/huida/Applications/anaconda3/lib/python3.10/site-packages/fastapi/routing.py", line 163, in run_endpoint_function
return await dependant.call(**values)
File "/Users/huida/code/quivr/backend/routes/chat_routes.py", line 197, in create_question_handler
chat_answer = gpt_answer_generator.generate_answer(chat_question.question)
File "/Users/huida/code/quivr/backend/llm/openai_functions.py", line 204, in generate_answer
messages=self._construct_prompt(
File "/Users/huida/code/quivr/backend/llm/openai_functions.py", line 158, in _construct_prompt
chat_context = self._get_context(question)
File "/Users/huida/code/quivr/backend/llm/openai_functions.py", line 128, in _get_context
return self.vector_store.similarity_search(query=question)
File "/Users/huida/code/quivr/backend/vectorstore/supabase.py", line 42, in similarity_search
).execute()
File "/Users/huida/Applications/anaconda3/lib/python3.10/site-packages/postgrest/_sync/request_builder.py", line 55, in execute
r = self.session.request(
File "/Users/huida/Applications/anaconda3/lib/python3.10/site-packages/httpx/_client.py", line 821, in request
return self.send(request, auth=auth, follow_redirects=follow_redirects)
...
httpx.ReadTimeout: The read operation timed out
In the Supabase interface manager, a 502 error is displayed, but other interface calls are functioning normally.

[bolshoytoster]

@pandaupc There is a related issue of the supabase-py repo. Their solution was to increase the server's timeout from the default of 5 seconds. I suggest doing this as a temporary solution, but someone should probably look into why it's taking so long.

---

To increase the timeout, you'll have to make the same change in 3 files:
In backend/models/settings.py, add the line from supabase.lib.client_options import ClientOptions anywhere at the top of the file, then on line 30:
https://github.com/StanGirard/quivr/blob/72924b52bca989f68cf8bdb1be95451dc0fae602/backend/models/settings.py#L29-L31

Add , options=ClientOptions(timeout=60), where 60 is the timeout in seconds.

In backend/llm/openai.py, add the same line to the top of the file, then add the same bit to the end of line 70:
https://github.com/StanGirard/quivr/blob/72924b52bca989f68cf8bdb1be95451dc0fae602/backend/llm/openai.py#L69-L71

In backend/llm/openai_functions.py, add the same line to the top of the file, then add the same bit to the end of line 81:
https://github.com/StanGirard/quivr/blob/72924b52bca989f68cf8bdb1be95451dc0fae602/backend/llm/openai_functions.py#L80-L82

[github-actions]

Thanks for your contributions, we'll be closing this issue as it has gone stale. Feel free to reopen if you'd like to continue the discussion.