Julia on VSCode Notebook does not clean RAM and hangs if notebook restarted

I have a notebook I use in VSCode, with a kernel Julia 1.9.3, that uses the package Agents. I launch a distributed simulation on 10 cores on my Ubuntu 18.04. The model loads a CSV file (50MB), then I launch, say, 30 simulations on my 10 workers.
Every time I launch the simulations, at the end my RAM increases of 6GB and stays that way indefinitely, no matter if I close the notebook: every time I run, I have some more RAM occupied and never cleared.
Also, I noticed that when I close or stop the notebook, I have a Julia process occupying a core on my machine indefinitely until I manually kill it…
Do you have any hints on how to spot memory leaks? or is this a known problem using Julia kernels in VSCode?

ok I have seen that to free memory you need to explicilty call rmprocs(workers())

I guess only point that remains is the problem with the notebook…