Yeah but I don't get what it is for - for now I have 2 agents, each listening to some queues. I actually ignore the "services" queue until now

I don't get the difference between how I'm using my agents now, just starting them on machines, and making them listen to queues, to using the "services" mode

yep

I mean, I barely have 20 experiments

why does it deplete so fast?

Increased to 20, lets see how long will it last 🙂

Where is the docker-compose file? It's not at /opt (again, I didn't place it anywhere, I'm just using the ami)

AgitatedDove14

so putting the docs aside, what permissions should I give to the IAM associated with trains' autoscale ?

so when I'm on the machine through SSH

this means that you supply the machine to execute the actions specified

it works without any problem

SuccessfulKoala55 ⬆ sorry for the delay, was at a meeting

I'll anonimyze it and paste it in a sec

id did not help

it is the location of the repo

I can't specify the absolute path, it is dynamically created

I can try something with $(pwd)

I also ran it without $(pwd) on the Create Clearml task templates section, I added it because of CostlyOstrich36 's comments but it didn't help

Why is that an issue?

Okay SuccessfulKoala55 , problem solved! Indeed the problem was that there is not .git folder. I updated necessary things to make the checkout action get the actual repo and now it works

It wasn't really clear to me what "standalone" means, maybe it will be better to add to the error

Error: Standalone

(no .git folder found)

script detected 'tasks/hp_optimization.py', but no requirements provided

to see if it contains the .git folder

what should I paste here to diagnose it?

Okay, so the agent automatically inherits the launching environment's variables?

Cool thanks 🤩

SuccessfulKoala55

I have a single IAM, my question is what kind of permissions I should associate with the IAM so that the autoscaler task will work

This is what I meant should be documented - the permissions...

😅

?

SuccessfulKoala55 ?