Hello,
When I create a task on a new server I use, the task fails to auto detect the working directory and the repository, as in the attached image.
Consequently, I cannot run the task in clearml agent, getting " ERROR: Can not run task without repository or literal script in script.diff
" :
repository = branch = version_num = tag = docker_cmd = /home/rdekel/anaconda3/envs/P1 entry_point = working_dir = Executing Conda: /home/rdekel/anaconda3/condabin/conda env remove -p /home/rdekel/temp_cache/.trains/venvs-builds.6/3.7 --quiet --json Using pre-existing Conda environment from /home/rdekel/anaconda3/envs/P1 clearml_agent: ERROR: Can not run task without repository or literal script in
script.diff
I can reproduce this problem with a very simple code, running:
import multiprocessing as mp from clearml import Task mp.set_start_method('spawn') task = Task.init("OMD", task_name="bla") print(f"task.get_script={task.get_script()}") task.close()
Yields:
task.get_script={'working_dir': None, 'entry_point': '', 'branch': None, 'repository': ''}
As you can see, nothing in the working directory or the other fields.
This problem happens on a specific server, the other servers I use detect the repository as needed.
I tried updating clearml to the latest version and rebooting the machine, and no help.
It would be really helpful if you can help in any way, as I currently have 4 X A100 that sit idle and wait... 🥲
Thanks a lot,
Ron