Upgraded to 2.1.0, ran locally works fine but deployments don't work anymore as libraries could not be find. Do you have to somehow install libraries now to run the deployments???

Libraries need to be installed within both your deployment build and your execution environment What Prefect packages for you, though are your custom modules (helper functions, custom code, SQL scripts, configuration files etc ) which are present in your root directory

Thanks. Just to clarify I am running prefect + agent + db in a docker-compose service. I created a custom image for the agent including mlflow and all the other libraries necessary as installing mlflow on the HPC cluster where the docker service is running didn't work for me. Just for my information, is there a guide how to handle all those things on a remote server when not using prefect cloud?

no, I sort of count on community contributions here You actually don't need such a complicated setup, Prefect 2.0 is super lightweight and works just in pure Python without Docker

it's nice for testing, but I'd be nervous to run a tool in production when it depends on that the CLI is not closing and thus stopping server and agent. because that's what it would do if I dont misunderstand

for Production, you can sign up for Prefect Cloud - there is an always free tier, no trials, no talking to Sales unless you want to https://app.prefect.cloud/