All-Hands-AI
UnsupportedProtocol Error when prompting with openai llm model in Windows wsl environment
Describe the bug
When testing OpenDevin with OpenAI , every prompt request produces an OpenAIException which is raised by an UnsupportedProtocol Error telling me that a Request URL is missing an 'http://' or 'https://' protocol. Did I miss something when setting up the config? Mine looks pretty similar to others that I have seen.
Setup and configuration
Running OpenDevin in a Windows 11 wsl Ubuntu environment.
Current version:
707ab7b3f84fb5664ff63da0b52e7b0d2e4df545
My config.toml and environment vars (be sure to redact API keys):
LLM_MODEL="gpt-3.5-turbo-0125"
LLM_API_KEY="s...."
LLM_EMBEDDING_MODEL="openai"
WORKSPACE_DIR="./workspace"
My model and agent (you can see these settings in the UI):
- Model: gpt-3.5-turbo-0125
- Agent: MonologueAgent
Commands I ran to install and run OpenDevin:
git clone https://github.com/OpenDevin/OpenDevin.git
cd OpenDevin
sudo make build
sudo make setup-config
sudo make run
Steps to Reproduce:
- setting gpt-3.5-turbo-0125 as model and openai for embeddings
- make run
- Prompting the model
Logs, error messages, and screenshots:
ERROR:
OpenAIException - Traceback (most recent call last):
File "/root/.cache/pypoetry/virtualenvs/opendevin-w_6MAHcD-py3.11/lib/python3.11/site-packages/httpx/_transports/default.py", line 69, in map_httpcore_exceptions
yield
File "/root/.cache/pypoetry/virtualenvs/opendevin-w_6MAHcD-py3.11/lib/python3.11/site-packages/httpx/_transports/default.py", line 233, in handle_request
resp = self._pool.handle_request(req)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/root/.cache/pypoetry/virtualenvs/opendevin-w_6MAHcD-py3.11/lib/python3.11/site-packages/httpcore/_sync/connection_pool.py", line 167, in handle_request
raise UnsupportedProtocol(
httpcore.UnsupportedProtocol: Request URL is missing an 'http://' or 'https://' protocol.
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "/root/.cache/pypoetry/virtualenvs/opendevin-w_6MAHcD-py3.11/lib/python3.11/site-packages/openai/_base_client.py", line 931, in _request
response = self._client.send(
^^^^^^^^^^^^^^^^^^
File "/root/.cache/pypoetry/virtualenvs/opendevin-w_6MAHcD-py3.11/lib/python3.11/site-packages/httpx/_client.py", line 914, in send
response = self._send_handling_auth(
^^^^^^^^^^^^^^^^^^^^^^^^^
File "/root/.cache/pypoetry/virtualenvs/opendevin-w_6MAHcD-py3.11/lib/python3.11/site-packages/httpx/_client.py", line 942, in _send_handling_auth
response = self._send_handling_redirects(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/root/.cache/pypoetry/virtualenvs/opendevin-w_6MAHcD-py3.11/lib/python3.11/site-packages/httpx/_client.py", line 979, in _send_handling_redirects
response = self._send_single_request(request)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/root/.cache/pypoetry/virtualenvs/opendevin-w_6MAHcD-py3.11/lib/python3.11/site-packages/httpx/_client.py", line 1015, in _send_single_request
response = transport.handle_request(request)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/root/.cache/pypoetry/virtualenvs/opendevin-w_6MAHcD-py3.11/lib/python3.11/site-packages/httpx/_transports/default.py", line 232, in handle_request
with map_httpcore_exceptions():
File "/usr/lib/python3.11/contextlib.py", line 158, in __exit__
self.gen.throw(typ, value, traceback)
File "/root/.cache/pypoetry/virtualenvs/opendevin-w_6MAHcD-py3.11/lib/python3.11/site-packages/httpx/_transports/default.py", line 86, in map_httpcore_exceptions
raise mapped_exc(message) from exc
httpx.UnsupportedProtocol: Request URL is missing an 'http://' or 'https://' protocol.
Additional Context
The opendevin server is running fine, so I guess my setup seems to be correct. Maybe I am missing something obvious. I suppose I dont need the LLM_BASE_URL config parameter when using openai?
You're right, I don't have it set either and it works.
Can you please try the suggestion in this comment? https://github.com/OpenDevin/OpenDevin/issues/908#issuecomment-2046517745
You're right, I don't have it set either and it works.
Can you please try the suggestion in this comment? #908 (comment)
I tried that, didnt help unfortunately. I tried this:
https://github.com/OpenDevin/OpenDevin/issues/908#issuecomment-2046184211
Which then leads to this when doing "make run":
File "/root/.cache/pypoetry/virtualenvs/opendevin-4oYb6y1w-py3.11/lib/python3.11/site-packages/litellm/router.py", line 199, in __init__
self.set_model_list(model_list)
File "/root/.cache/pypoetry/virtualenvs/opendevin-4oYb6y1w-py3.11/lib/python3.11/site-packages/litellm/router.py", line 2091, in set_model_list
deployment = self._add_deployment(deployment=deployment)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/root/.cache/pypoetry/virtualenvs/opendevin-4oYb6y1w-py3.11/lib/python3.11/site-packages/litellm/router.py", line 2127, in _add_deployment
) = litellm.get_llm_provider(
^^^^^^^^^^^^^^^^^^^^^^^^^
File "/root/.cache/pypoetry/virtualenvs/opendevin-4oYb6y1w-py3.11/lib/python3.11/site-packages/litellm/utils.py", line 5850, in get_llm_provider
raise e
File "/root/.cache/pypoetry/virtualenvs/opendevin-4oYb6y1w-py3.11/lib/python3.11/site-packages/litellm/utils.py", line 5837, in get_llm_provider
raise litellm.exceptions.BadRequestError( # type: ignore
litellm.exceptions.BadRequestError: LLM Provider NOT provided. Pass in the LLM provider you are trying to call. You passed model=llama-7b-chat
Pass model as E.g. For 'Huggingface' inference endpoints pass in `completion(model='huggingface/starcoder',..)` Learn more: https://docs.litellm.ai/docs/providers
Which is kinda weird because llama-7b-chat was something I used in the beginning but my current config still looks like this:
LLM_MODEL="gpt-3.5-turbo-1106" LLM_API_KEY="s...." LLM_EMBEDDING_MODEL="openai" WORKSPACE_DIR="./workspace"
So maybe I have some problems with the poetry cache? I tried to clear the cache with poetry cache clear --all . but it says nothing is there to clear. How can I make sure that every cache or anything related to old stuff is cleaned before I start the make run again?
Aha, that's exactly the problem: your installation is sending stale information out there, but it's not poetry. We can fix it.
I'm going to reiterate some, but also complete here what I said in other threads on this. So the idea is simple: latest main branch now should work, but we need to get rid of leftover stale information on your environment.
- stop opendevin backend, frontend, close localhost tabs
- clear local storage in your browser, or use another browser you haven't used with opendevin in the last few days
- git pull
- delete ./cache folder in ./opendevin (important)
- make build
- make start-backend, allow it to finish until it says uvicorn ready or so
- make start-frontend
- open localhost:3001
