thanks

and another question.. how can i get support over the phone? i feel like i have a lot of questions about setting this up and that would be easier

got it

ok ya i thought maybe to create a user for the agent

not the best options as they also need to be maintained and updated, the other repos i have are constantly worked on and changing too

users and passwords for the web login

the problem i have is, my code is using scripts that are located in multiple repos, when i execute remotely it only clones the repo the main script is in, so i end up having to clone the other ones manually in my code

sounds good then i think i'll try that, thanks!

hi i have another question- when setting up the agent on the worker machine, what credentials do i give it? a user's?

also, if i create a submodule, will it clone it automatically as well?

ok great thanks

the package is keras-contrib, it can't be installed regularly with pip install, you need to use pip install git+ https://www.github.com/keras-team/keras-contrib.git

running from pycharm, i want it to use the env i have on the agent, like when i do venv/bin/activate.

clearml-agent daemon --queue default --detached

ill have to wait until it drops again and see

ok so what seems to happen is that when the machine reboots, after it comes back up the agent is seemingly still working. clearml-agent list shoes the agent and i see it in the web ui. since the /tmp folder is deleted after reboot the log file is gone. when i try to run something on the agent only then the list is empty

what do you mean? doing a git clone within the script?

i'll test it, thanks

i meant the users from the machine i installed the server on, like in /etc/passwd

weird thing is it didn't realize it was dead until a task was trying to run, it looked like it was still up even after the reboot

AgitatedDove14 hi martin, i have another question- when in the remote debugging mode, does it take the agent off the queue? does it know it's not available to take tasks?

from what i see in the docs it's only for Jupyter / VS Code, i didn't see anything about pycharm

i need to run the debug on the remote agent because i don't have enough GPU on my local one

? AgitatedDove14

for example i have a package that can't be installed with pip regularly, but with pip from a git repo, and when the agent installs the requirements from the pycharm env it gets an error as it can't find it with pip. so i thought of just having the environment locally and use it