Thanks a lot!! @<1523701070390366208:profile|CostlyOstrich36> @<1523701087100473344:profile|SuccessfulKoala55>
If you want to sometimes run with docker and sometimes without, yes.
i have a systemd service that runs the following
ExecStart=/bin/bash -c 'source /home/clearml/.bashrc && clearml-agent daemon --queue gpu_3080 --foreground'
do i need to create another systemd service for the agent using docker?
you set up 2 agent runs - one with docker and the other without. Each agent should be listening to a different queue. Makes sense?
thanks @<1523701070390366208:profile|CostlyOstrich36> for the fast response!
do I need to update the field python_binary?
the path is correct.
I updated clearml.conf on the remote host
Set the python version to use when creating the virtual environment and launching the experiment
# Example values: "/usr/bin/python3" or "/usr/local/bin/python3.6"
# The default is the python executing the clearml_agent
python_binary: "/opt/conda/bin/python3"
How did you set it up? Are you sure that's the correct path?
the same service will run 2 agents?
I get the following error:
clearml_agent: ERROR: Python executable with version '/opt/conda/bin/python3' requested by the Task, key 'agent.default_python', not found in path, tried: ('/opt/conda/bin/python3', '/opt/conda/bin/python3')
@<1564060263047499776:profile|ThoughtfulCentipede62> how did you run the agent?
Hi @<1564060263047499776:profile|ThoughtfulCentipede62> , you can specify the Python interpreter in clearml.conf
on the remote machine (Search for 'binary' or 'python')
Also, yes ClearML can pull it from an artifactory as long as the machine has access 🙂
I'm running a systemd service
ExecStart=/bin/bash -c 'source /home/clearml/.bashrc && clearml-agent daemon --queue gpu_3080 --foreground'
Just making sure - This path, is it inside the docker or outside the container?
You can run agent as many times as you want on your machine
is the docker image is the docker that will be used for the task?
so the agent isn't running in docker mode and thus won't run the task inside a docker container
I don't think you have to, but if your environment is ready you can try 🙂
you need to add --docker <image-name>
actually, sometimes I will not use a docker for my tasks. is there a way to support both ways?
@<1564060263047499776:profile|ThoughtfulCentipede62> , you can run the agent twice on different queues - one with docker one without