Error: sh 1: python not found

[rm-data-viz]

https://github.com/heartexlabs/label-studio-ml-backend/blob/4989ef5adf3673e3d41fac67964b90298ac07573/label_studio_ml/server.py#L112

I believe it should be:

os.system('python3 ' + wsgi + ' ' +' '.join(subprocess_params))

What are your thoughts?

[jonatassalgado]

I get the same error in version 1.0.1. Before, in old version, I had a lot ml services running, but until now I spend a few hours and I not figure out how to trespassing this issue :(

Nothing different in logs at uwsgi.log and rq.log

I tried following the simple tutorial in the README at this repo after run label-studio-ml start my_ml_backend. In a linux machine with python3.

[makseq]

@jonatassalgado @rm-data-viz Are you using venv or virtualenv? I strongly recommend to do this. Looks like you have some mix of python environments. Read more here: https://docs.python.org/3/tutorial/venv.html#creating-virtual-environments

[rm-data-viz]

@makseq - I'm using the docker version of Label-Studio. I don't think it's associated to python virtual environment.

[makseq]

In this case it's weird: LS 1.0 docker doesn't have ML backend at all.

[rm-data-viz]

It does have a django ml app. But, to make ml backend work, I cloned the label-studio-ml repository.

I then initiated the ml_backend, but during label-studio-ml start ml_backend. It throws this error.

[alex-eurec]

The same problem is happening to me, I chose not to use venv, because I created a VM with a 'labelstudio' user that will be only running LabelStudio and the ML module.

This is happening in an ubuntu 20.20 server, that only has python3 installed (by default installation).

[makseq]

Closed as outdated and probably solved.