Hi WackyRabbit7 ,

Did you try using sdk.development.detect_with_pip_freeze as true in your ~/clearml.conf file? It should take the same environment as the one you are running from.

basically it run pip freeze in your execution env and create the requirements according to it, without any analysis

TimelyPenguin76 this fixed it, using the detect_with_pip_freeze as true solves the issue

I can easily see how this problem can reoccur

The non-pip freeze will have the package your are using in your script, and not the whole env, according to imports and usage

can't really see how this configuration effects the agents

Gotcha

Can you lend a few a words about how the not-pip freeze mechanism of detecting packages work?

I think a good idea is to add to the error message when the clearml agent fails due to import error, a suggestion ot try out with pip freeze

But does it disable the agent? or will the tasks still wait for the agent to dequeue?

It doesn’t, but if your issue is

since the task requirements is not logged correctly and then when it is cloned it fails

this should log the same requirements as you have in your machine, without any analysis. with the same environment the agent shouldn’t have this issue

I'll try it