but I still had time to go inside the container, export the PATH variables for my poetry and python versions, and run the poetry install command there

Yes indeed, but what about the possibility to do the clone/poetry installation ourself in the init bash script of the task?

Yes I take the export statements from my bash script of the task

This is really extremely hard to debug. I am thinking to create another repo and iterate on the packages to hopefully find the problem, but it will take ages.

@<1523701070390366208:profile|CostlyOstrich36> @<1523701087100473344:profile|SuccessfulKoala55> I tried with dummy repo. Using Python and stripe packages ONLY in the pyproject.toml

Here is my result (still failing) :

Poetry Enabled: Ignoring requested python packages, using repository poetry lock file!
Creating virtualenv debug in /root/.clearml/venvs-builds/3.9/task_repository/clearmldebug.git/.venv
Using virtualenv: /root/.clearml/venvs-builds/3.9/task_repository/clearmldebug.git/.venv
Installing dependencies from lock file
Finding the necessary packages for the current system
Package operations: 6 installs, 0 updates, 0 removals
failed installing poetry requirements: Command '['poetry', 'install', '-n', '-v']' returned non-zero exit status 1.
Ignoring pip: markers 'python_version >= "3.10"' don't match your environment

Yes should be correct. Inside the bash script of the task.

the autoscaler always uses docker mode

I am literrally trying with 1 package and python and it fails. I tried with python 3.8 3.9 and 3.9.16. and it always fail --> not linked to python version. What is the problem then? I am wondering if there is not an intrinsic bug

I tried too. I do not have more logs inside the ClearML agent 😞

So the same?

I literrally connected to it at runtime, and ran poetry install -n and it worked

Is it a bug inside the AWS autoscaler??

I guess it makes no sense because of the steps a clearml-agent works...
I also thought about going to pip mode but not all packages are detected from our poetry.lock file unfortunately so cannot do that.

where did you specify it?

I am currently trying with a new dummy repo and I iterate over the dependencies of the pyproject.toml.

@<1556812486840160256:profile|SuccessfulRaven86> , did you install poetry inside the EC2 instance or inside the docker? Basically, where do you put the poetry installation bash script - in the 'init script' section of the autoscaler or on the task's 'setup shell script' in execution tab (This is basically the script that runs inside the docker)

It sounds like you're installing poetry on the ec2 instance itself but the experiment runs inside a docker container

Because I was ssh-ing to it before the fail. When poetry fails, it installs everything using PIP

When the task finally failed, I was kicked of from the container

How to make sure that the python version is correct?

Yes same error

I think you should try to manually start such a docker container and try to see what fails in the process. Attaching to an existing one has too many differences already

How do you explain that it works when I ssh-ed into the same AWS container instance from the autoscaler?

Yes, the problem is it's still really hidden (the error, I mean)

How is it still up is the task failed?

The autoscaler just runs it on an AWS instance, inside a docker container - there's no difference from running it yourself inside a docker container - did you try running it inside a docker container as well?

The same container

or just to the AWS instance?

into the same docker container running the task?

@<1556812486840160256:profile|SuccessfulRaven86> , to make things easier to debug, can you try running the agent locally?

@<1523701070390366208:profile|CostlyOstrich36> poetry is installed as part of the bash script of the task.

The init script of the AWS autoscaler only contains three export variables I set.