when I specify --packages I shoudl manually list them all not?

I assume it has nothing to do with my client version

Cool - so that means the fileserver which comes with the host will stay emtpy? Or is there anything else being stored there?

I guess what I want is a way to define environment variables in agents

For example I have a DATA_DIR environment variable which points to the directory where disk-data is stored

CostlyOstrich36 so why 1000:1000? My user and group are not that and so do all the otehr files I have under /opt/clearml

so in my code, I'll use this environment variable to read from disk

okay, that's acceptable

you can use pgrep -af "trains-agent"

I don't think the problem is setting that variable, I think it has something to do with it but not that obvious... Because it did work for me in the past, since then we docker-compose up/downed a few times, changed some other things etc... Can't figure out what made it get to this point

I'd be lying if I said I had time for that 🙂

why not use my user and group?

Sorry but its not clear

sorry I think it trimmed it

let me repay you with a nice trick

its like ps + grep together 😄

Thanks Alon

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

yep

The latest, I curl ed the docker-compose like 10 minutes ago

is this already available or only on github?

👍 thanks (again)

CostlyOstrich36 ?

Okay, so if my python script imports some other scripts I've written - I must use git?

I assume that at some points in the execution, the client (where the task is running) is sending JSONs to the mongo service, and that is what we see in the web UI.
Since we are talking about a case where there is no internet available, maybe these could be dumped into files/stdout and let the user manually insert them.
The manual insertion UX could be something like a CLI copy-paste or and endpoint for files - but since your UX is so good ( 🙂 ) I'm sure you'll figure this part out better

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

Oh I get it, that also makes sense with the docs directing this at inference jobs and avoiding GPU - because of the 1-N thing

does the services mode have a separate configuration for base image?