The backend starts with a precomputed number of processes based on the available CPUs.
This value can be overridden by setting the number of workers in the configuration file.
Processes are automatically restarted after reaching the maximum number of handled requests
(default: 10000), which can also be adjusted in the configuration file.