.wslconfig as bottleneck for Docker users in Windows

[pedgomgal1]

I am just sharing info regarding the problem I have been facing when trying to infer a semantic segmentation model in large 2D images (>600 MB). BiaPy released an error as presented bellow:

/tmp/tmp8z1ohrf7: line 3: 16 Killed python3 -u /installations/BiaPy/main.py --config /BiaPy_files/input.yaml --result_dir /F/Lab/MouseMusclePOGLUT1/data/predictions/all_predictions_30012026_modelv2 --name input_modified --run_id 1 --dist_backend gloo
ERROR conda.cli.main_run:execute(127): conda run python3 -u /installations/BiaPy/main.py --config /BiaPy_files/input.yaml --result_dir /F/Lab/MouseMusclePOGLUT1/data/predictions/all_predictions_30012026_modelv2 --name input_modified --run_id 1 --dist_backend gloo failed. (See above for error)

I double-checked all the BiaPy settings and the resources of my computer. Everything was fine. After a very long chat with chatGPT we realised Docker was causing a bottleneck through its internal configuration file: ".wslconfig", accessed through: C:\Users<YOUR_USERNAME>.wslconfig

There, the memory parameter was only allocating 6GB, and there was no other way to visualise it in Docker but by accessing this file. There, I just modified the file by increasing the memo and the processors:

memory=64GB
processors=16
swap=32GB

right after I had to type in the Windows powershell->wsl --shutdown
and restart Docker. Once restarted, you can verify everything is fine by executing within the Docker image:

free -h
nproc

I hope this issue could help others! Best wishes.

[danifranco]

Thank you so much for reporting this Pedro! Maybe our documentation master (@iarganda) can add all this info in our documentation so other users can follow these steps in case they face the same error. Thanks again!