I have added a custom docker file along with the dependencies required for my repository. I was able to run the agent in docker mode using this custom docker file

from: None

ColossalHawk50 , you basically run the agent outside of a docker container and then use the --docker flag to run it in docker mode. Makes sense?

It is working now. Thanks

i had created a new task from the clearml web ui by pulling the code from git and ran it in a clearml-agent in normal mode (clearml-agent daemon --queue default) and it worked fine. but the same code when i run using the --docker flag is not working. (clearml-agent daemon --queue default --docker)

you can either:

• Build an image from your docker file and when running the task/experiment, tell it to use that docker image
• If the steps to install dependencies required for your repository is not too complicate, then you can use agent.extra_docker_shell_script in the clearml.conf in order to install all the dependencies inside the docker container launched by clearml in docker mode.
  

Is this the same issue as per here ?
In which case, can you make your script run using that docker container nvidia/cuda:11.0.3-cudnn8-runtime-ubuntu20.04 , manually, without ClearML ?
What change did you do to make it worked ? Does updating python enough ?

clearml_agent: ERROR: Command '['/root/.clearml/venvs-builds/3.10/bin/python', '-m', 'pip', '--disable-pip-version-check', 'install', '-r', '/tmp/requirements_alb8_42l.txt', '--use-deprecated=legacy-resolver']' returned non-zero exit status 1.

This is the error log I am getting. Should i upgrade the python version? if yes, how do i make changes in the clearml.conf file