Hi John, thanks for your reply.
I do get this for the first task which is train_model :
task b22ef7f69f214beb8c35243160df650e pulled from 53f50b52fa944b08a325f34edc6da03d by worker imp:0, and I can also see that in the screenshot below, but it still stays in Queue. For this worker, I can clone and run individual tasks successfully.

Hi John,
This was solved by assigning a different to the pipeline and to the different tasks of the pipeline. Thanks for your help

Hi @<1635813046947418112:profile|FriendlyHedgehong10> , the pipeline basically creates tasks and pushes them into execution. You can click on each step and view the full details. In the info section you can see into which queue each step was pushed. I'm assuming there are no agents listening to the queue

Workers listen to queues, and tasks are assigned to queues. You can't directly assign a job to a worker per se..

I have another question. I can now run successfully a pipeline remotely via code, but I also want to run one using the UI and NEW RUN. However the Run button is not activated? Any idea of what I need to change?

I checked the info of the task and I am getting:

A worker can execute only a single job at a time

The pipeline has been assigned to both a queue and a worker but the first step/task only to a queue but not a worker