"TypeError: HTTPConnection.request() got an unexpected keyword argument 'chunked'" error while running AlphaFold

[RuiJoaoLoureiro2019]

Dear AlphaFold users and/or developers,

I'm struggling to find the root cause of two error messages "File "/home/rloureiro/alphafold/alphafold-env/lib/python3.10/site-packages/urllib3/connectionpool.py", line 497, in _make_request conn.request( TypeError: HTTPConnection.request() got an unexpected keyword argument 'chunked'" and "During handling of the above exception, another exception occurred: File "/home/rloureiro/alphafold/alphafold-env/lib/python3.10/site-packages/docker/api/client.py", line 221, in _retrieve_server_version raise DockerException( docker.errors.DockerException: Error while fetching server API version: HTTPConnection.request() got an unexpected keyword argument 'chunked'" while trying to run AlphaFold on a local machine with Ubuntu 22.04 LTS, CUDA 12.2 and the compatible latest version of Docker. I installed AlphaFold dependencies and tried to run AlphaFold on a Python environment having Python 3.10.12. I uninstalled and reinstalled Docker and AlphaFold but I got the same mistakes when trying to run it. Any support in solving this issue would be highly appreciated.

With best regards, Rui João Loureiro

[laolanllx]

same error here, running successfully before, with ubuntu 20.04, CUDA 11.5 and DRIVER 510.85.02

[laolanllx]

same error here, running successfully before, with ubuntu 20.04, CUDA 11.5 and DRIVER 510.85.02

solved, I used the wrong python. Changed the python path solved this.

[RuiJoaoLoureiro2019]

Hi @laolanllx !

Thanks for your quick feeedback. Did you mean you were running AlphaFold in a python environment and used Python from your system environment?

Thanks

[laolanllx]

Hi @laolanllx !

Thanks for your quick feeedback. Did you mean you were running AlphaFold in a python environment and used Python from your system environment?

Thanks

I installled alphafold based on conda environment. When I used which python to check , it showed it's not the alphafold one. Thus, I exit and login and reactivate my alphafold environment. Then it works. Hope this can help you.

[LaurinLang]

Could you share what the full package list for your environment you run alphafold in looks like now? I just got a fresh install today and face the same error.

[RuiJoaoLoureiro2019]

Hi @laolanllx! Did you change any of the Dockerfile settings to match your CUDA version and OS? I just did it and instead of solving the problem, I got an error in the build phase of the installation "ERROR: failed to solve: nvidia/cuda:12.2-cudnn8-runtime-ubuntu22.04: docker.io/nvidia/cuda:12.2-cudnn8-runtime-ubuntu22.04: not found". Previously to that, I have run AlphaFold outside a virtual environment just to test and it began to run but then it crashed in the HHSearch phase.

[laolanllx]

Hi @laolanllx! Did you change any of the Dockerfile settings to match your CUDA version and OS? I just did it and instead of solving the problem, I got an error in the build phase of the installation "ERROR: failed to solve: nvidia/cuda:12.2-cudnn8-runtime-ubuntu22.04: docker.io/nvidia/cuda:12.2-cudnn8-runtime-ubuntu22.04: not found". Previously to that, I have run AlphaFold outside a virtual environment just to test and it began to run but then it crashed in the HHSearch phase.

no, I didn't because I use an old version of alphafold. I think it's a 2.2.* one. This one works perfectly on my old machine. However, when I set up AF2.3.2 on a new one with 4090 GPU, Ubuntu22.04 CUDA 12.2 Driver 535.104.05. Some error appear and haven't solved till now #764 . Even though I change the Dockerfile like this, ARG CUDA=11.1.1------->ARG CUDA=11.8.0 FROM nvidia/cuda:${CUDA}-cudnn8-runtime-ubuntu18.04------->FROM nvidia/cuda:${CUDA}-cudnn8-devel-ubuntu20.04 https://github.com/google-deepmind/alphafold/issues/764#issuecomment-1679537433

[laolanllx]

Hi @laolanllx! Did you change any of the Dockerfile settings to match your CUDA version and OS? I just did it and instead of solving the problem, I got an error in the build phase of the installation "ERROR: failed to solve: nvidia/cuda:12.2-cudnn8-runtime-ubuntu22.04: docker.io/nvidia/cuda:12.2-cudnn8-runtime-ubuntu22.04: not found". Previously to that, I have run AlphaFold outside a virtual environment just to test and it began to run but then it crashed in the HHSearch phase.

321 mentioned a new way to change Dockerfile, you can try it. https://github.com/google-deepmind/alphafold/issues/321#issuecomment-1475245689 I tried yesterday, but didn't work for me, maybe the python version issue (it used python3.8). I'll try this again later.

[titusfx]

I dunno if is the same problem that I have but I fix that changing docker-compose to docker compose

[GuShaocheng]

Hi everyone, what is a real issue? I installed the latest version, had a similar issue: lib/python3.10/site-packages/docker/api/client.py", line 221, in _retrieve_server_version raise DockerException( docker.errors.DockerException: Error while fetching server API version: HTTPConnection.request() got an unexpected keyword argument 'chunked'

[damiano-sg]

Hi everyone, I had the same issue and I solved it by installing a different version of requests, I used version 2.28.1

[ocstx]

Just installed latest version of AF and found the same issue, but... First I got the following error: requests.exceptions.InvalidURL: Not supported URL scheme http+docker According to this: https://github.com/docker/docker-py/issues/3256 can be fixed by using the previous version of requests (2.31.0) So I did it It is with requests==2.31.0 that I found the "got an unexpected keyword argument 'chunked'" error. Your solution is still valid. requests==2.28.1 solved the issue. Perhaps requests==2.28.1 could be added to the "alphafold/docker/requirements.txt" file thank you very much