When performing"podman exec" into a container to launch a new process, it interferes with the performance of the existing container/database. So the customer is looking to launch a new, separate process inside the container using podman exec and apply specific CPU limits only to that new process.

The idea is to create a resource boundary for the script, ensuring it cannot interfere with the performance of the main database process running in the same container. They cannot create a separate database container for these tasks because of the stateful nature of the database; the scripts must run against the live data.

Essentially, they are looking for a feature for "podman exec" that is similar to the "--cpu" flag in "podman run" - a way to constrain the new process, not the entire container.